diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,58516 @@
+{
+  "best_metric": 1.89534922643667e-13,
+  "best_model_checkpoint": "./checkpoint-1500",
+  "epoch": 2.999922845459455,
+  "global_step": 9720,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 9.9994,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 9.9823,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 9.5606,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 9.9595,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4670184702981194e-07,
+      "loss": 10.0083,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.934036940596239e-07,
+      "loss": 9.349,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.934036940596239e-07,
+      "loss": 9.4961,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.401055410894359e-07,
+      "loss": 9.9053,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.868073881192478e-07,
+      "loss": 9.9411,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2335092351490599e-06,
+      "loss": 10.2539,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4802110821788718e-06,
+      "loss": 9.2498,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4802110821788718e-06,
+      "loss": 9.3151,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7269129292086836e-06,
+      "loss": 9.4923,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9736147762384955e-06,
+      "loss": 9.7777,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2203166232683074e-06,
+      "loss": 8.9812,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4670184702981197e-06,
+      "loss": 9.112,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.713720317327931e-06,
+      "loss": 8.1211,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9604221643577435e-06,
+      "loss": 8.232,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2071240113875554e-06,
+      "loss": 7.7281,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4538258584173673e-06,
+      "loss": 7.882,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.700527705447179e-06,
+      "loss": 7.2153,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.947229552476991e-06,
+      "loss": 7.6867,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.193931399506803e-06,
+      "loss": 6.8404,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.440633246536615e-06,
+      "loss": 7.038,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.687335093566427e-06,
+      "loss": 6.6647,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9340369405962395e-06,
+      "loss": 6.3349,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.180738787626051e-06,
+      "loss": 6.1034,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.427440634655862e-06,
+      "loss": 6.4131,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.674142481685675e-06,
+      "loss": 5.7632,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.920844328715487e-06,
+      "loss": 6.1685,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.920844328715487e-06,
+      "loss": 5.935,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.167546175745299e-06,
+      "loss": 5.5854,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.414248022775111e-06,
+      "loss": 5.4124,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.660949869804922e-06,
+      "loss": 5.6406,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.907651716834735e-06,
+      "loss": 5.8216,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.154353563864547e-06,
+      "loss": 5.22,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.401055410894358e-06,
+      "loss": 5.0853,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.64775725792417e-06,
+      "loss": 5.2381,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.894459104953982e-06,
+      "loss": 5.1294,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.141160951983795e-06,
+      "loss": 5.0527,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.387862799013607e-06,
+      "loss": 4.7977,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.63456464604342e-06,
+      "loss": 4.8616,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.88126649307323e-06,
+      "loss": 4.7318,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.127968340103041e-06,
+      "loss": 4.9216,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.374670187132854e-06,
+      "loss": 4.4783,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.621372034162666e-06,
+      "loss": 4.4616,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.868073881192479e-06,
+      "loss": 4.3317,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.011477572822229e-05,
+      "loss": 4.0049,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0361477575252102e-05,
+      "loss": 3.6552,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0608179422281913e-05,
+      "loss": 3.5597,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0854881269311725e-05,
+      "loss": 6.5201,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1101583116341538e-05,
+      "loss": 6.2329,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.134828496337135e-05,
+      "loss": 6.0426,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1594986810401163e-05,
+      "loss": 6.0495,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1841688657430974e-05,
+      "loss": 5.8186,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2088390504460786e-05,
+      "loss": 5.8554,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2335092351490599e-05,
+      "loss": 5.555,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2581794198520408e-05,
+      "loss": 5.5688,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2828496045550222e-05,
+      "loss": 5.5471,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3075197892580033e-05,
+      "loss": 5.2925,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3321899739609845e-05,
+      "loss": 5.3143,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3568601586639658e-05,
+      "loss": 5.1692,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.381530343366947e-05,
+      "loss": 5.3045,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4062005280699282e-05,
+      "loss": 5.2949,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4308707127729094e-05,
+      "loss": 5.0509,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4555408974758905e-05,
+      "loss": 5.0167,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4802110821788717e-05,
+      "loss": 4.818,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5048812668818528e-05,
+      "loss": 5.0345,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.529551451584834e-05,
+      "loss": 5.051,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5542216362878153e-05,
+      "loss": 4.9864,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5788918209907964e-05,
+      "loss": 4.9994,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.603562005693778e-05,
+      "loss": 4.8565,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.628232190396759e-05,
+      "loss": 4.8892,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6529023750997402e-05,
+      "loss": 4.9114,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6775725598027214e-05,
+      "loss": 4.7279,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7022427445057025e-05,
+      "loss": 4.8821,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.726912929208684e-05,
+      "loss": 4.6722,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7515831139116648e-05,
+      "loss": 4.7111,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.776253298614646e-05,
+      "loss": 4.6077,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.800923483317627e-05,
+      "loss": 4.7383,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8255936680206082e-05,
+      "loss": 4.4168,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8502638527235897e-05,
+      "loss": 4.658,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.874934037426571e-05,
+      "loss": 4.5859,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.899604222129552e-05,
+      "loss": 4.4373,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.924274406832533e-05,
+      "loss": 4.3366,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9489445915355143e-05,
+      "loss": 4.5744,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9736147762384958e-05,
+      "loss": 4.5122,
+      "step": 87
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998284960941477e-05,
+      "loss": 4.3594,
+      "step": 88
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.022955145644458e-05,
+      "loss": 4.3232,
+      "step": 89
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0476253303474392e-05,
+      "loss": 4.2864,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0722955150504204e-05,
+      "loss": 4.2353,
+      "step": 91
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.096965699753402e-05,
+      "loss": 4.4,
+      "step": 92
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1216358844563827e-05,
+      "loss": 4.1035,
+      "step": 93
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1463060691593638e-05,
+      "loss": 3.693,
+      "step": 94
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.170976253862345e-05,
+      "loss": 4.0602,
+      "step": 95
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1956464385653264e-05,
+      "loss": 3.7158,
+      "step": 96
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.2203166232683076e-05,
+      "loss": 3.4228,
+      "step": 97
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.2449868079712887e-05,
+      "loss": 3.3592,
+      "step": 98
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.26965699267427e-05,
+      "loss": 3.4861,
+      "step": 99
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.294327177377251e-05,
+      "loss": 3.2347,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.3189973620802325e-05,
+      "loss": 6.2737,
+      "step": 101
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.3436675467832137e-05,
+      "loss": 6.0196,
+      "step": 102
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.3683377314861948e-05,
+      "loss": 5.7992,
+      "step": 103
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.393007916189176e-05,
+      "loss": 5.231,
+      "step": 104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.417678100892157e-05,
+      "loss": 5.2319,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.4423482855951386e-05,
+      "loss": 5.2672,
+      "step": 106
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.4670184702981197e-05,
+      "loss": 5.0631,
+      "step": 107
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.491688655001101e-05,
+      "loss": 4.9932,
+      "step": 108
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.5163588397040817e-05,
+      "loss": 5.0783,
+      "step": 109
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.541029024407063e-05,
+      "loss": 5.1133,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.5656992091100443e-05,
+      "loss": 4.7806,
+      "step": 111
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.5903693938130255e-05,
+      "loss": 4.9077,
+      "step": 112
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.6150395785160066e-05,
+      "loss": 4.6014,
+      "step": 113
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.6397097632189878e-05,
+      "loss": 4.6012,
+      "step": 114
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.664379947921969e-05,
+      "loss": 4.8114,
+      "step": 115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.6890501326249504e-05,
+      "loss": 4.665,
+      "step": 116
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.7137203173279315e-05,
+      "loss": 4.847,
+      "step": 117
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.7383905020309127e-05,
+      "loss": 4.6202,
+      "step": 118
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.763060686733894e-05,
+      "loss": 4.5651,
+      "step": 119
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.7877308714368753e-05,
+      "loss": 4.6493,
+      "step": 120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.8124010561398565e-05,
+      "loss": 4.6014,
+      "step": 121
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.8370712408428376e-05,
+      "loss": 4.536,
+      "step": 122
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.8617414255458188e-05,
+      "loss": 4.4809,
+      "step": 123
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.8864116102488e-05,
+      "loss": 4.3241,
+      "step": 124
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.911081794951781e-05,
+      "loss": 4.4554,
+      "step": 125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9357519796547622e-05,
+      "loss": 4.274,
+      "step": 126
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9604221643577433e-05,
+      "loss": 4.4213,
+      "step": 127
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9850923490607245e-05,
+      "loss": 4.4207,
+      "step": 128
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.0097625337637056e-05,
+      "loss": 4.6034,
+      "step": 129
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.034432718466687e-05,
+      "loss": 4.3281,
+      "step": 130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.059102903169668e-05,
+      "loss": 4.205,
+      "step": 131
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.0837730878726494e-05,
+      "loss": 4.1821,
+      "step": 132
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.1084432725756306e-05,
+      "loss": 4.0485,
+      "step": 133
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.133113457278612e-05,
+      "loss": 4.0697,
+      "step": 134
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.157783641981593e-05,
+      "loss": 3.9912,
+      "step": 135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.182453826684574e-05,
+      "loss": 4.2788,
+      "step": 136
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.207124011387556e-05,
+      "loss": 3.9497,
+      "step": 137
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.231794196090537e-05,
+      "loss": 4.2084,
+      "step": 138
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.256464380793518e-05,
+      "loss": 4.1596,
+      "step": 139
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.281134565496499e-05,
+      "loss": 3.8781,
+      "step": 140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.3058047501994804e-05,
+      "loss": 3.8881,
+      "step": 141
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.3304749349024616e-05,
+      "loss": 3.8212,
+      "step": 142
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.355145119605443e-05,
+      "loss": 3.9197,
+      "step": 143
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.379815304308424e-05,
+      "loss": 3.864,
+      "step": 144
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.404485489011405e-05,
+      "loss": 3.8519,
+      "step": 145
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.429155673714386e-05,
+      "loss": 3.7788,
+      "step": 146
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.453825858417368e-05,
+      "loss": 3.5859,
+      "step": 147
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.4784960431203484e-05,
+      "loss": 3.4278,
+      "step": 148
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.5031662278233296e-05,
+      "loss": 3.1378,
+      "step": 149
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.527836412526311e-05,
+      "loss": 3.0606,
+      "step": 150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.552506597229292e-05,
+      "loss": 5.9656,
+      "step": 151
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.577176781932273e-05,
+      "loss": 5.9505,
+      "step": 152
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.601846966635254e-05,
+      "loss": 5.591,
+      "step": 153
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.626517151338235e-05,
+      "loss": 5.2512,
+      "step": 154
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.6511873360412165e-05,
+      "loss": 4.9191,
+      "step": 155
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.6758575207441976e-05,
+      "loss": 4.6688,
+      "step": 156
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.7005277054471794e-05,
+      "loss": 4.915,
+      "step": 157
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.7251978901501606e-05,
+      "loss": 4.8045,
+      "step": 158
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.749868074853142e-05,
+      "loss": 4.7794,
+      "step": 159
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.774538259556123e-05,
+      "loss": 4.5804,
+      "step": 160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.799208444259104e-05,
+      "loss": 4.5253,
+      "step": 161
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.823878628962085e-05,
+      "loss": 4.5609,
+      "step": 162
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.848548813665066e-05,
+      "loss": 4.5709,
+      "step": 163
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.8732189983680475e-05,
+      "loss": 4.4665,
+      "step": 164
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.8978891830710286e-05,
+      "loss": 4.6487,
+      "step": 165
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9225593677740104e-05,
+      "loss": 4.3506,
+      "step": 166
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9472295524769916e-05,
+      "loss": 4.8238,
+      "step": 167
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.971899737179973e-05,
+      "loss": 4.4556,
+      "step": 168
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996569921882954e-05,
+      "loss": 4.3084,
+      "step": 169
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.021240106585935e-05,
+      "loss": 4.4329,
+      "step": 170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.045910291288916e-05,
+      "loss": 4.1638,
+      "step": 171
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.070580475991897e-05,
+      "loss": 4.2945,
+      "step": 172
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0952506606948785e-05,
+      "loss": 4.6448,
+      "step": 173
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.1199208453978596e-05,
+      "loss": 4.1398,
+      "step": 174
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.144591030100841e-05,
+      "loss": 4.3881,
+      "step": 175
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.1692612148038226e-05,
+      "loss": 4.1718,
+      "step": 176
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.193931399506804e-05,
+      "loss": 4.4054,
+      "step": 177
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.218601584209785e-05,
+      "loss": 4.3241,
+      "step": 178
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.243271768912765e-05,
+      "loss": 4.4298,
+      "step": 179
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.2679419536157465e-05,
+      "loss": 4.1339,
+      "step": 180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.2926121383187276e-05,
+      "loss": 4.3067,
+      "step": 181
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.317282323021709e-05,
+      "loss": 3.9943,
+      "step": 182
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.34195250772469e-05,
+      "loss": 3.9987,
+      "step": 183
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.366622692427671e-05,
+      "loss": 4.1286,
+      "step": 184
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.391292877130653e-05,
+      "loss": 4.1267,
+      "step": 185
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.415963061833634e-05,
+      "loss": 4.1323,
+      "step": 186
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.440633246536615e-05,
+      "loss": 3.9972,
+      "step": 187
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.465303431239596e-05,
+      "loss": 3.9931,
+      "step": 188
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.4899736159425775e-05,
+      "loss": 3.9881,
+      "step": 189
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.5146438006455586e-05,
+      "loss": 3.8725,
+      "step": 190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.53931398534854e-05,
+      "loss": 3.9095,
+      "step": 191
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.563984170051521e-05,
+      "loss": 3.7215,
+      "step": 192
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.588654354754502e-05,
+      "loss": 3.8142,
+      "step": 193
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.613324539457483e-05,
+      "loss": 3.3684,
+      "step": 194
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.637994724160465e-05,
+      "loss": 3.6073,
+      "step": 195
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.662664908863446e-05,
+      "loss": 3.307,
+      "step": 196
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.687335093566427e-05,
+      "loss": 3.2923,
+      "step": 197
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7120052782694085e-05,
+      "loss": 3.3907,
+      "step": 198
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7366754629723896e-05,
+      "loss": 3.1184,
+      "step": 199
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.761345647675371e-05,
+      "loss": 3.0719,
+      "step": 200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.786015832378352e-05,
+      "loss": 6.1078,
+      "step": 201
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.810686017081333e-05,
+      "loss": 5.6465,
+      "step": 202
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.835356201784314e-05,
+      "loss": 5.3461,
+      "step": 203
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8600263864872954e-05,
+      "loss": 4.8343,
+      "step": 204
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.884696571190277e-05,
+      "loss": 4.6286,
+      "step": 205
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.909366755893258e-05,
+      "loss": 4.6812,
+      "step": 206
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9340369405962395e-05,
+      "loss": 4.5481,
+      "step": 207
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9587071252992206e-05,
+      "loss": 4.536,
+      "step": 208
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.983377310002202e-05,
+      "loss": 4.5787,
+      "step": 209
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.008047494705183e-05,
+      "loss": 4.4325,
+      "step": 210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.0327176794081634e-05,
+      "loss": 4.5575,
+      "step": 211
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.0573878641111445e-05,
+      "loss": 4.4885,
+      "step": 212
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.082058048814126e-05,
+      "loss": 4.411,
+      "step": 213
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.1067282335171075e-05,
+      "loss": 4.5176,
+      "step": 214
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.1313984182200886e-05,
+      "loss": 4.3857,
+      "step": 215
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.15606860292307e-05,
+      "loss": 4.4306,
+      "step": 216
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.180738787626051e-05,
+      "loss": 4.2458,
+      "step": 217
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.205408972329032e-05,
+      "loss": 4.2212,
+      "step": 218
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.230079157032013e-05,
+      "loss": 4.2546,
+      "step": 219
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.2547493417349944e-05,
+      "loss": 4.3256,
+      "step": 220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.2794195264379755e-05,
+      "loss": 4.4029,
+      "step": 221
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.304089711140957e-05,
+      "loss": 4.2838,
+      "step": 222
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.328759895843938e-05,
+      "loss": 4.494,
+      "step": 223
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.3534300805469196e-05,
+      "loss": 4.3454,
+      "step": 224
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.378100265249901e-05,
+      "loss": 4.3382,
+      "step": 225
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.402770449952882e-05,
+      "loss": 4.1983,
+      "step": 226
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.427440634655863e-05,
+      "loss": 4.289,
+      "step": 227
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.452110819358844e-05,
+      "loss": 4.1536,
+      "step": 228
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.4767810040618254e-05,
+      "loss": 4.2602,
+      "step": 229
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.5014511887648065e-05,
+      "loss": 4.1487,
+      "step": 230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.526121373467788e-05,
+      "loss": 4.1293,
+      "step": 231
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.550791558170769e-05,
+      "loss": 4.0864,
+      "step": 232
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.5754617428737506e-05,
+      "loss": 4.2018,
+      "step": 233
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.600131927576732e-05,
+      "loss": 4.0122,
+      "step": 234
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.624802112279713e-05,
+      "loss": 4.197,
+      "step": 235
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.649472296982694e-05,
+      "loss": 4.0332,
+      "step": 236
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.674142481685675e-05,
+      "loss": 3.969,
+      "step": 237
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.6988126663886564e-05,
+      "loss": 3.9932,
+      "step": 238
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.7234828510916375e-05,
+      "loss": 4.097,
+      "step": 239
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.748153035794619e-05,
+      "loss": 3.9676,
+      "step": 240
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.7728232204976e-05,
+      "loss": 3.7376,
+      "step": 241
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.79749340520058e-05,
+      "loss": 3.9942,
+      "step": 242
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.822163589903562e-05,
+      "loss": 3.6519,
+      "step": 243
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.846833774606543e-05,
+      "loss": 3.6726,
+      "step": 244
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.8715039593095244e-05,
+      "loss": 3.324,
+      "step": 245
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.8961741440125055e-05,
+      "loss": 3.4168,
+      "step": 246
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.920844328715487e-05,
+      "loss": 3.4067,
+      "step": 247
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.945514513418468e-05,
+      "loss": 3.153,
+      "step": 248
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.970184698121449e-05,
+      "loss": 2.9696,
+      "step": 249
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.99485488282443e-05,
+      "loss": 2.8854,
+      "step": 250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.019525067527411e-05,
+      "loss": 5.5135,
+      "step": 251
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.0441952522303924e-05,
+      "loss": 5.2969,
+      "step": 252
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.068865436933374e-05,
+      "loss": 4.9749,
+      "step": 253
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.0935356216363554e-05,
+      "loss": 4.7775,
+      "step": 254
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.118205806339337e-05,
+      "loss": 4.613,
+      "step": 255
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.142875991042317e-05,
+      "loss": 4.5745,
+      "step": 256
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.167546175745299e-05,
+      "loss": 4.7139,
+      "step": 257
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.19221636044828e-05,
+      "loss": 4.6129,
+      "step": 258
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.216886545151261e-05,
+      "loss": 4.4759,
+      "step": 259
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.241556729854243e-05,
+      "loss": 4.424,
+      "step": 260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.266226914557223e-05,
+      "loss": 4.3594,
+      "step": 261
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.290897099260205e-05,
+      "loss": 4.3573,
+      "step": 262
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.315567283963186e-05,
+      "loss": 4.3436,
+      "step": 263
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.340237468666168e-05,
+      "loss": 4.4567,
+      "step": 264
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.364907653369148e-05,
+      "loss": 4.2672,
+      "step": 265
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.38957783807213e-05,
+      "loss": 4.4398,
+      "step": 266
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.414248022775112e-05,
+      "loss": 4.2977,
+      "step": 267
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.438918207478092e-05,
+      "loss": 4.2575,
+      "step": 268
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.463588392181074e-05,
+      "loss": 4.2958,
+      "step": 269
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.488258576884054e-05,
+      "loss": 4.1785,
+      "step": 270
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.512928761587036e-05,
+      "loss": 4.2825,
+      "step": 271
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.537598946290017e-05,
+      "loss": 4.4133,
+      "step": 272
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.562269130992999e-05,
+      "loss": 4.4458,
+      "step": 273
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.586939315695979e-05,
+      "loss": 4.2545,
+      "step": 274
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.611609500398961e-05,
+      "loss": 4.1371,
+      "step": 275
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.636279685101941e-05,
+      "loss": 4.2689,
+      "step": 276
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.660949869804923e-05,
+      "loss": 4.2919,
+      "step": 277
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.685620054507905e-05,
+      "loss": 4.3332,
+      "step": 278
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.710290239210885e-05,
+      "loss": 4.2571,
+      "step": 279
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.734960423913867e-05,
+      "loss": 4.0688,
+      "step": 280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.759630608616848e-05,
+      "loss": 4.0989,
+      "step": 281
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.78430079331983e-05,
+      "loss": 3.9163,
+      "step": 282
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.80897097802281e-05,
+      "loss": 4.0267,
+      "step": 283
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.833641162725792e-05,
+      "loss": 4.2511,
+      "step": 284
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.858311347428772e-05,
+      "loss": 3.9217,
+      "step": 285
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.882981532131754e-05,
+      "loss": 3.9477,
+      "step": 286
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.907651716834736e-05,
+      "loss": 3.8026,
+      "step": 287
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.932321901537716e-05,
+      "loss": 3.8774,
+      "step": 288
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.956992086240697e-05,
+      "loss": 4.038,
+      "step": 289
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.981662270943677e-05,
+      "loss": 3.606,
+      "step": 290
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.006332455646659e-05,
+      "loss": 3.6562,
+      "step": 291
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.03100264034964e-05,
+      "loss": 3.8134,
+      "step": 292
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.055672825052621e-05,
+      "loss": 3.4138,
+      "step": 293
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.080343009755602e-05,
+      "loss": 3.2232,
+      "step": 294
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.105013194458584e-05,
+      "loss": 3.4359,
+      "step": 295
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.129683379161566e-05,
+      "loss": 3.1689,
+      "step": 296
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.154353563864546e-05,
+      "loss": 3.3418,
+      "step": 297
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.179023748567528e-05,
+      "loss": 3.3289,
+      "step": 298
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.203693933270508e-05,
+      "loss": 3.0747,
+      "step": 299
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.22836411797349e-05,
+      "loss": 2.7899,
+      "step": 300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.25303430267647e-05,
+      "loss": 5.8501,
+      "step": 301
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.277704487379452e-05,
+      "loss": 5.3189,
+      "step": 302
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.302374672082433e-05,
+      "loss": 5.2962,
+      "step": 303
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.327044856785415e-05,
+      "loss": 4.6646,
+      "step": 304
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.351715041488395e-05,
+      "loss": 4.5088,
+      "step": 305
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.376385226191377e-05,
+      "loss": 4.5518,
+      "step": 306
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.401055410894359e-05,
+      "loss": 4.6864,
+      "step": 307
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.42572559559734e-05,
+      "loss": 4.5587,
+      "step": 308
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.450395780300321e-05,
+      "loss": 4.4907,
+      "step": 309
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.475065965003302e-05,
+      "loss": 4.1272,
+      "step": 310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.499736149706283e-05,
+      "loss": 4.3515,
+      "step": 311
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.524406334409264e-05,
+      "loss": 4.2094,
+      "step": 312
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.549076519112246e-05,
+      "loss": 4.6832,
+      "step": 313
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.573746703815226e-05,
+      "loss": 4.3876,
+      "step": 314
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.598416888518208e-05,
+      "loss": 4.2954,
+      "step": 315
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.62308707322119e-05,
+      "loss": 4.4803,
+      "step": 316
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.64775725792417e-05,
+      "loss": 4.2748,
+      "step": 317
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.672427442627152e-05,
+      "loss": 4.5386,
+      "step": 318
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.697097627330133e-05,
+      "loss": 4.1999,
+      "step": 319
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.721767812033114e-05,
+      "loss": 4.0726,
+      "step": 320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.746437996736095e-05,
+      "loss": 4.3165,
+      "step": 321
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.771108181439077e-05,
+      "loss": 4.1606,
+      "step": 322
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.795778366142057e-05,
+      "loss": 4.1394,
+      "step": 323
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.820448550845039e-05,
+      "loss": 4.0756,
+      "step": 324
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.845118735548021e-05,
+      "loss": 4.2201,
+      "step": 325
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.869788920251001e-05,
+      "loss": 4.0369,
+      "step": 326
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.894459104953983e-05,
+      "loss": 4.3879,
+      "step": 327
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.919129289656964e-05,
+      "loss": 3.9861,
+      "step": 328
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.943799474359945e-05,
+      "loss": 4.0601,
+      "step": 329
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.968469659062926e-05,
+      "loss": 4.0512,
+      "step": 330
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.993139843765908e-05,
+      "loss": 4.2947,
+      "step": 331
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.017810028468888e-05,
+      "loss": 4.0716,
+      "step": 332
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.04248021317187e-05,
+      "loss": 3.9328,
+      "step": 333
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.067150397874852e-05,
+      "loss": 3.781,
+      "step": 334
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.091820582577832e-05,
+      "loss": 4.1123,
+      "step": 335
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.116490767280814e-05,
+      "loss": 3.8107,
+      "step": 336
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.141160951983795e-05,
+      "loss": 3.9465,
+      "step": 337
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.165831136686776e-05,
+      "loss": 3.9798,
+      "step": 338
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.190501321389757e-05,
+      "loss": 3.8881,
+      "step": 339
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.215171506092739e-05,
+      "loss": 3.7747,
+      "step": 340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.239841690795719e-05,
+      "loss": 3.6815,
+      "step": 341
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.264511875498701e-05,
+      "loss": 3.6616,
+      "step": 342
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.289182060201682e-05,
+      "loss": 3.4601,
+      "step": 343
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.313852244904663e-05,
+      "loss": 3.2104,
+      "step": 344
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.338522429607645e-05,
+      "loss": 3.2139,
+      "step": 345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.363192614310626e-05,
+      "loss": 3.3863,
+      "step": 346
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.387862799013607e-05,
+      "loss": 3.2103,
+      "step": 347
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.412532983716588e-05,
+      "loss": 3.2033,
+      "step": 348
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.43720316841957e-05,
+      "loss": 2.8542,
+      "step": 349
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.46187335312255e-05,
+      "loss": 3.0374,
+      "step": 350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.48654353782553e-05,
+      "loss": 5.5316,
+      "step": 351
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.511213722528511e-05,
+      "loss": 5.3367,
+      "step": 352
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.535883907231493e-05,
+      "loss": 4.9158,
+      "step": 353
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.560554091934475e-05,
+      "loss": 4.729,
+      "step": 354
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.585224276637455e-05,
+      "loss": 4.6709,
+      "step": 355
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.609894461340437e-05,
+      "loss": 4.6134,
+      "step": 356
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.634564646043418e-05,
+      "loss": 4.4264,
+      "step": 357
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.6592348307464e-05,
+      "loss": 4.5311,
+      "step": 358
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.68390501544938e-05,
+      "loss": 4.2861,
+      "step": 359
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.708575200152362e-05,
+      "loss": 4.5318,
+      "step": 360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.733245384855342e-05,
+      "loss": 4.4549,
+      "step": 361
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.757915569558324e-05,
+      "loss": 4.3015,
+      "step": 362
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.782585754261306e-05,
+      "loss": 4.4215,
+      "step": 363
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.807255938964286e-05,
+      "loss": 4.3765,
+      "step": 364
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.831926123667268e-05,
+      "loss": 4.4003,
+      "step": 365
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.856596308370249e-05,
+      "loss": 4.3718,
+      "step": 366
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.88126649307323e-05,
+      "loss": 4.1877,
+      "step": 367
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.905936677776211e-05,
+      "loss": 4.2822,
+      "step": 368
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.930606862479193e-05,
+      "loss": 4.3035,
+      "step": 369
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.955277047182173e-05,
+      "loss": 4.3714,
+      "step": 370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.979947231885155e-05,
+      "loss": 4.0867,
+      "step": 371
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.004617416588135e-05,
+      "loss": 3.962,
+      "step": 372
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.029287601291117e-05,
+      "loss": 4.4638,
+      "step": 373
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.053957785994099e-05,
+      "loss": 4.2593,
+      "step": 374
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.07862797069708e-05,
+      "loss": 4.307,
+      "step": 375
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.103298155400061e-05,
+      "loss": 4.299,
+      "step": 376
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.127968340103042e-05,
+      "loss": 4.0364,
+      "step": 377
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.152638524806024e-05,
+      "loss": 4.1699,
+      "step": 378
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.177308709509004e-05,
+      "loss": 4.2143,
+      "step": 379
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.201978894211986e-05,
+      "loss": 4.0192,
+      "step": 380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.226649078914966e-05,
+      "loss": 4.0221,
+      "step": 381
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.251319263617948e-05,
+      "loss": 4.0027,
+      "step": 382
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.27598944832093e-05,
+      "loss": 4.1677,
+      "step": 383
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.30065963302391e-05,
+      "loss": 4.1318,
+      "step": 384
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.325329817726892e-05,
+      "loss": 3.9487,
+      "step": 385
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.350000002429873e-05,
+      "loss": 3.879,
+      "step": 386
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.374670187132855e-05,
+      "loss": 4.0456,
+      "step": 387
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.399340371835835e-05,
+      "loss": 3.9708,
+      "step": 388
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.424010556538817e-05,
+      "loss": 3.7171,
+      "step": 389
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.448680741241797e-05,
+      "loss": 3.7966,
+      "step": 390
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.473350925944779e-05,
+      "loss": 3.6115,
+      "step": 391
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.498021110647761e-05,
+      "loss": 3.7155,
+      "step": 392
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.522691295350742e-05,
+      "loss": 3.6877,
+      "step": 393
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.547361480053723e-05,
+      "loss": 3.5269,
+      "step": 394
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.572031664756704e-05,
+      "loss": 3.4064,
+      "step": 395
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.596701849459686e-05,
+      "loss": 3.3396,
+      "step": 396
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.621372034162666e-05,
+      "loss": 3.3875,
+      "step": 397
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.646042218865648e-05,
+      "loss": 3.0848,
+      "step": 398
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.670712403568628e-05,
+      "loss": 3.003,
+      "step": 399
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.69538258827161e-05,
+      "loss": 2.9789,
+      "step": 400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.720052772974591e-05,
+      "loss": 5.5924,
+      "step": 401
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.744722957677573e-05,
+      "loss": 5.2388,
+      "step": 402
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.769393142380554e-05,
+      "loss": 4.5494,
+      "step": 403
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.794063327083535e-05,
+      "loss": 4.6223,
+      "step": 404
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.818733511786517e-05,
+      "loss": 4.8478,
+      "step": 405
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.843403696489497e-05,
+      "loss": 4.534,
+      "step": 406
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.868073881192479e-05,
+      "loss": 4.625,
+      "step": 407
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.89274406589546e-05,
+      "loss": 4.4475,
+      "step": 408
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.917414250598441e-05,
+      "loss": 4.3715,
+      "step": 409
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.942084435301422e-05,
+      "loss": 4.3184,
+      "step": 410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.966754620004404e-05,
+      "loss": 4.2362,
+      "step": 411
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.991424804707385e-05,
+      "loss": 4.2693,
+      "step": 412
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00010016094989410366,
+      "loss": 4.4151,
+      "step": 413
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00010040765174113346,
+      "loss": 4.3752,
+      "step": 414
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00010065435358816327,
+      "loss": 4.4849,
+      "step": 415
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00010090105543519309,
+      "loss": 4.3668,
+      "step": 416
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00010114775728222289,
+      "loss": 4.6339,
+      "step": 417
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00010139445912925271,
+      "loss": 4.2481,
+      "step": 418
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00010164116097628251,
+      "loss": 4.2447,
+      "step": 419
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00010188786282331233,
+      "loss": 4.2562,
+      "step": 420
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00010213456467034215,
+      "loss": 4.2921,
+      "step": 421
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00010238126651737195,
+      "loss": 4.0857,
+      "step": 422
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00010262796836440177,
+      "loss": 4.1084,
+      "step": 423
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00010287467021143158,
+      "loss": 4.0326,
+      "step": 424
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001031213720584614,
+      "loss": 4.1429,
+      "step": 425
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001033680739054912,
+      "loss": 3.8682,
+      "step": 426
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00010361477575252102,
+      "loss": 4.3055,
+      "step": 427
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00010386147759955082,
+      "loss": 4.0385,
+      "step": 428
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00010410817944658064,
+      "loss": 4.189,
+      "step": 429
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00010435488129361045,
+      "loss": 4.1096,
+      "step": 430
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00010460158314064026,
+      "loss": 4.0669,
+      "step": 431
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00010484828498767008,
+      "loss": 4.045,
+      "step": 432
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00010509498683469989,
+      "loss": 4.0539,
+      "step": 433
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001053416886817297,
+      "loss": 4.0231,
+      "step": 434
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00010558839052875951,
+      "loss": 4.1554,
+      "step": 435
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00010558839052875951,
+      "loss": 3.8854,
+      "step": 436
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00010583509237578933,
+      "loss": 3.9812,
+      "step": 437
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00010608179422281913,
+      "loss": 3.8511,
+      "step": 438
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00010632849606984895,
+      "loss": 3.7371,
+      "step": 439
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00010657519791687876,
+      "loss": 3.8623,
+      "step": 440
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00010682189976390857,
+      "loss": 3.919,
+      "step": 441
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00010706860161093839,
+      "loss": 3.7473,
+      "step": 442
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001073153034579682,
+      "loss": 3.6142,
+      "step": 443
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00010756200530499802,
+      "loss": 3.7105,
+      "step": 444
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00010780870715202782,
+      "loss": 3.4896,
+      "step": 445
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00010805540899905764,
+      "loss": 3.5483,
+      "step": 446
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00010830211084608744,
+      "loss": 3.7472,
+      "step": 447
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00010854881269311726,
+      "loss": 3.2245,
+      "step": 448
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00010879551454014707,
+      "loss": 3.2113,
+      "step": 449
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00010904221638717688,
+      "loss": 2.7866,
+      "step": 450
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001092889182342067,
+      "loss": 5.727,
+      "step": 451
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00010953562008123651,
+      "loss": 5.1451,
+      "step": 452
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00010978232192826633,
+      "loss": 4.7538,
+      "step": 453
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00011002902377529613,
+      "loss": 4.5769,
+      "step": 454
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00011027572562232595,
+      "loss": 4.5682,
+      "step": 455
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00011052242746935575,
+      "loss": 4.5233,
+      "step": 456
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00011076912931638557,
+      "loss": 4.4451,
+      "step": 457
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00011101583116341538,
+      "loss": 4.2981,
+      "step": 458
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001112625330104452,
+      "loss": 4.2268,
+      "step": 459
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00011150923485747501,
+      "loss": 4.363,
+      "step": 460
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00011175593670450482,
+      "loss": 4.4011,
+      "step": 461
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00011200263855153464,
+      "loss": 4.1197,
+      "step": 462
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00011224934039856444,
+      "loss": 4.4562,
+      "step": 463
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00011249604224559426,
+      "loss": 4.5259,
+      "step": 464
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00011274274409262406,
+      "loss": 4.3707,
+      "step": 465
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00011298944593965388,
+      "loss": 3.8887,
+      "step": 466
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00011323614778668369,
+      "loss": 4.2003,
+      "step": 467
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001134828496337135,
+      "loss": 4.1289,
+      "step": 468
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00011372955148074331,
+      "loss": 4.1508,
+      "step": 469
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00011397625332777313,
+      "loss": 4.2498,
+      "step": 470
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00011422295517480295,
+      "loss": 4.3538,
+      "step": 471
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00011446965702183275,
+      "loss": 4.591,
+      "step": 472
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00011471635886886257,
+      "loss": 4.2215,
+      "step": 473
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00011496306071589237,
+      "loss": 4.3044,
+      "step": 474
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00011520976256292219,
+      "loss": 4.2299,
+      "step": 475
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000115456464409952,
+      "loss": 4.2404,
+      "step": 476
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001157031662569818,
+      "loss": 4.0391,
+      "step": 477
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001159498681040116,
+      "loss": 4.2259,
+      "step": 478
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00011619656995104142,
+      "loss": 3.9942,
+      "step": 479
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00011644327179807124,
+      "loss": 4.0105,
+      "step": 480
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00011668997364510105,
+      "loss": 3.9026,
+      "step": 481
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00011693667549213087,
+      "loss": 4.0074,
+      "step": 482
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00011718337733916067,
+      "loss": 4.0125,
+      "step": 483
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00011743007918619049,
+      "loss": 3.8485,
+      "step": 484
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00011767678103322029,
+      "loss": 4.0881,
+      "step": 485
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00011792348288025011,
+      "loss": 3.8867,
+      "step": 486
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00011817018472727992,
+      "loss": 3.8611,
+      "step": 487
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00011841688657430973,
+      "loss": 3.7431,
+      "step": 488
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00011866358842133955,
+      "loss": 4.0315,
+      "step": 489
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00011891029026836936,
+      "loss": 3.809,
+      "step": 490
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00011915699211539918,
+      "loss": 3.88,
+      "step": 491
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00011940369396242898,
+      "loss": 3.3458,
+      "step": 492
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001196503958094588,
+      "loss": 3.7114,
+      "step": 493
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001198970976564886,
+      "loss": 3.6372,
+      "step": 494
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012014379950351842,
+      "loss": 3.739,
+      "step": 495
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012039050135054823,
+      "loss": 3.1826,
+      "step": 496
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012063720319757804,
+      "loss": 3.2308,
+      "step": 497
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012088390504460785,
+      "loss": 3.1079,
+      "step": 498
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012113060689163767,
+      "loss": 2.8073,
+      "step": 499
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012137730873866748,
+      "loss": 2.8303,
+      "step": 500
+    },
+    {
+      "epoch": 0.15,
+      "eval_bleu": 0.0,
+      "eval_loss": 4.923842430114746,
+      "eval_runtime": 2550.431,
+      "eval_samples_per_second": 5.787,
+      "eval_steps_per_second": 0.723,
+      "step": 500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012162401058569729,
+      "loss": 5.9959,
+      "step": 501
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00012187071243272711,
+      "loss": 5.6859,
+      "step": 502
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001221174142797569,
+      "loss": 4.9284,
+      "step": 503
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012236411612678673,
+      "loss": 4.7624,
+      "step": 504
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012261081797381655,
+      "loss": 4.5076,
+      "step": 505
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012285751982084634,
+      "loss": 4.5359,
+      "step": 506
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012310422166787616,
+      "loss": 4.7318,
+      "step": 507
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012335092351490598,
+      "loss": 4.4868,
+      "step": 508
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012333754488979588,
+      "loss": 4.5527,
+      "step": 509
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012332416626468582,
+      "loss": 4.3696,
+      "step": 510
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012331078763957575,
+      "loss": 4.4506,
+      "step": 511
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012329740901446568,
+      "loss": 4.3571,
+      "step": 512
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001232840303893556,
+      "loss": 4.3717,
+      "step": 513
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012327065176424552,
+      "loss": 4.346,
+      "step": 514
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012325727313913546,
+      "loss": 3.9544,
+      "step": 515
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012324389451402536,
+      "loss": 4.3818,
+      "step": 516
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012323051588891527,
+      "loss": 4.4631,
+      "step": 517
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001232171372638052,
+      "loss": 4.0567,
+      "step": 518
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012320375863869514,
+      "loss": 4.3836,
+      "step": 519
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012319038001358507,
+      "loss": 4.3008,
+      "step": 520
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012317700138847498,
+      "loss": 4.6076,
+      "step": 521
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001231636227633649,
+      "loss": 4.994,
+      "step": 522
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012315024413825484,
+      "loss": 4.9677,
+      "step": 523
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012313686551314475,
+      "loss": 4.9555,
+      "step": 524
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012312348688803468,
+      "loss": 4.8203,
+      "step": 525
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001231101082629246,
+      "loss": 4.8152,
+      "step": 526
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012309672963781452,
+      "loss": 4.8872,
+      "step": 527
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012308335101270446,
+      "loss": 5.1067,
+      "step": 528
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012306997238759436,
+      "loss": 5.6763,
+      "step": 529
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001230565937624843,
+      "loss": 5.5465,
+      "step": 530
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012304321513737423,
+      "loss": 5.1378,
+      "step": 531
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012302983651226414,
+      "loss": 4.6955,
+      "step": 532
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012301645788715407,
+      "loss": 4.8377,
+      "step": 533
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00012300307926204398,
+      "loss": 4.5607,
+      "step": 534
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001229897006369339,
+      "loss": 4.2408,
+      "step": 535
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012297632201182385,
+      "loss": 4.5209,
+      "step": 536
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012296294338671375,
+      "loss": 4.7752,
+      "step": 537
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012294956476160369,
+      "loss": 4.4023,
+      "step": 538
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001229361861364936,
+      "loss": 4.1366,
+      "step": 539
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012292280751138353,
+      "loss": 4.3019,
+      "step": 540
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012290942888627346,
+      "loss": 4.0776,
+      "step": 541
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012289605026116337,
+      "loss": 3.9817,
+      "step": 542
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001228826716360533,
+      "loss": 3.9172,
+      "step": 543
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012286929301094323,
+      "loss": 3.9923,
+      "step": 544
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012285591438583314,
+      "loss": 3.8953,
+      "step": 545
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012284253576072307,
+      "loss": 3.9234,
+      "step": 546
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012282915713561298,
+      "loss": 3.4393,
+      "step": 547
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001228157785105029,
+      "loss": 3.2897,
+      "step": 548
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012280239988539285,
+      "loss": 3.4355,
+      "step": 549
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012278902126028275,
+      "loss": 3.3794,
+      "step": 550
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012277564263517269,
+      "loss": 5.8921,
+      "step": 551
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012276226401006262,
+      "loss": 5.4134,
+      "step": 552
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012274888538495253,
+      "loss": 5.02,
+      "step": 553
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012273550675984246,
+      "loss": 4.8916,
+      "step": 554
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012272212813473237,
+      "loss": 4.6914,
+      "step": 555
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001227087495096223,
+      "loss": 4.5255,
+      "step": 556
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012269537088451223,
+      "loss": 4.7455,
+      "step": 557
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012268199225940214,
+      "loss": 4.5667,
+      "step": 558
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012266861363429207,
+      "loss": 4.6288,
+      "step": 559
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000122655235009182,
+      "loss": 4.5168,
+      "step": 560
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001226418563840719,
+      "loss": 4.7083,
+      "step": 561
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012262847775896185,
+      "loss": 4.5492,
+      "step": 562
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012261509913385175,
+      "loss": 4.6801,
+      "step": 563
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001226017205087417,
+      "loss": 4.5671,
+      "step": 564
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012258834188363162,
+      "loss": 4.6073,
+      "step": 565
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012257496325852153,
+      "loss": 4.3533,
+      "step": 566
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00012256158463341146,
+      "loss": 4.6854,
+      "step": 567
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001225482060083014,
+      "loss": 4.5101,
+      "step": 568
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001225348273831913,
+      "loss": 4.6245,
+      "step": 569
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00012252144875808123,
+      "loss": 4.6061,
+      "step": 570
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00012250807013297114,
+      "loss": 4.3197,
+      "step": 571
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00012249469150786107,
+      "loss": 4.3263,
+      "step": 572
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000122481312882751,
+      "loss": 4.3852,
+      "step": 573
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00012246793425764091,
+      "loss": 4.5023,
+      "step": 574
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00012245455563253085,
+      "loss": 4.293,
+      "step": 575
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00012244117700742078,
+      "loss": 4.2612,
+      "step": 576
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001224277983823107,
+      "loss": 4.3834,
+      "step": 577
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00012241441975720062,
+      "loss": 4.3958,
+      "step": 578
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00012240104113209053,
+      "loss": 4.3465,
+      "step": 579
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00012238766250698046,
+      "loss": 4.3482,
+      "step": 580
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001223742838818704,
+      "loss": 4.3616,
+      "step": 581
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001223609052567603,
+      "loss": 4.2006,
+      "step": 582
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00012234752663165023,
+      "loss": 4.3303,
+      "step": 583
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00012233414800654017,
+      "loss": 4.2894,
+      "step": 584
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00012232076938143007,
+      "loss": 4.382,
+      "step": 585
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00012230739075632,
+      "loss": 4.2585,
+      "step": 586
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00012229401213120991,
+      "loss": 4.1285,
+      "step": 587
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00012228063350609985,
+      "loss": 4.3608,
+      "step": 588
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00012226725488098978,
+      "loss": 4.2241,
+      "step": 589
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001222538762558797,
+      "loss": 3.9742,
+      "step": 590
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00012224049763076962,
+      "loss": 4.3244,
+      "step": 591
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00012222711900565956,
+      "loss": 3.8326,
+      "step": 592
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00012221374038054946,
+      "loss": 4.1109,
+      "step": 593
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001222003617554394,
+      "loss": 3.5887,
+      "step": 594
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001221869831303293,
+      "loss": 4.0595,
+      "step": 595
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00012217360450521924,
+      "loss": 3.779,
+      "step": 596
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00012216022588010917,
+      "loss": 3.6884,
+      "step": 597
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00012214684725499908,
+      "loss": 3.6359,
+      "step": 598
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000122133468629889,
+      "loss": 3.5836,
+      "step": 599
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00012212009000477894,
+      "loss": 3.255,
+      "step": 600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00012210671137966885,
+      "loss": 5.6803,
+      "step": 601
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00012209333275455878,
+      "loss": 5.6142,
+      "step": 602
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001220799541294487,
+      "loss": 5.0492,
+      "step": 603
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00012206657550433864,
+      "loss": 4.7204,
+      "step": 604
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00012205319687922854,
+      "loss": 4.4783,
+      "step": 605
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00012203981825411846,
+      "loss": 4.8103,
+      "step": 606
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001220264396290084,
+      "loss": 4.6155,
+      "step": 607
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00012201306100389832,
+      "loss": 4.8471,
+      "step": 608
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00012199968237878825,
+      "loss": 4.4469,
+      "step": 609
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00012198630375367817,
+      "loss": 4.4243,
+      "step": 610
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00012197292512856809,
+      "loss": 4.413,
+      "step": 611
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00012195954650345802,
+      "loss": 4.5389,
+      "step": 612
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00012194616787834793,
+      "loss": 4.424,
+      "step": 613
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00012193278925323786,
+      "loss": 4.5006,
+      "step": 614
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00012191941062812778,
+      "loss": 4.4158,
+      "step": 615
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001219060320030177,
+      "loss": 4.7135,
+      "step": 616
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00012189265337790764,
+      "loss": 4.3617,
+      "step": 617
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00012187927475279756,
+      "loss": 4.1773,
+      "step": 618
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00012186589612768748,
+      "loss": 4.1398,
+      "step": 619
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001218525175025774,
+      "loss": 4.3508,
+      "step": 620
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00012183913887746732,
+      "loss": 4.3629,
+      "step": 621
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00012182576025235725,
+      "loss": 4.1533,
+      "step": 622
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00012181238162724717,
+      "loss": 4.1764,
+      "step": 623
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00012179900300213709,
+      "loss": 4.2879,
+      "step": 624
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00012178562437702702,
+      "loss": 4.3449,
+      "step": 625
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00012177224575191694,
+      "loss": 4.0918,
+      "step": 626
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00012175886712680688,
+      "loss": 4.4934,
+      "step": 627
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00012174548850169678,
+      "loss": 4.1859,
+      "step": 628
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001217321098765867,
+      "loss": 4.3091,
+      "step": 629
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00012171873125147664,
+      "loss": 4.735,
+      "step": 630
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00012170535262636656,
+      "loss": 4.1607,
+      "step": 631
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00012169197400125648,
+      "loss": 4.2881,
+      "step": 632
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00012167859537614641,
+      "loss": 4.1848,
+      "step": 633
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00012166521675103633,
+      "loss": 4.0926,
+      "step": 634
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00012165183812592626,
+      "loss": 3.9356,
+      "step": 635
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00012163845950081617,
+      "loss": 4.358,
+      "step": 636
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00012162508087570609,
+      "loss": 4.0504,
+      "step": 637
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00012161170225059602,
+      "loss": 4.1675,
+      "step": 638
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00012159832362548594,
+      "loss": 3.886,
+      "step": 639
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00012158494500037588,
+      "loss": 3.9011,
+      "step": 640
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001215715663752658,
+      "loss": 3.7507,
+      "step": 641
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00012155818775015572,
+      "loss": 4.036,
+      "step": 642
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00012154480912504564,
+      "loss": 3.8348,
+      "step": 643
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00012153143049993556,
+      "loss": 3.7055,
+      "step": 644
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00012151805187482548,
+      "loss": 3.5662,
+      "step": 645
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00012150467324971541,
+      "loss": 3.4458,
+      "step": 646
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00012149129462460533,
+      "loss": 3.408,
+      "step": 647
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00012147791599949526,
+      "loss": 3.4834,
+      "step": 648
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00012146453737438518,
+      "loss": 3.2232,
+      "step": 649
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001214511587492751,
+      "loss": 3.0037,
+      "step": 650
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00012143778012416502,
+      "loss": 5.6602,
+      "step": 651
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00012142440149905494,
+      "loss": 5.4904,
+      "step": 652
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00012141102287394486,
+      "loss": 5.0785,
+      "step": 653
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001213976442488348,
+      "loss": 4.7764,
+      "step": 654
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00012138426562372472,
+      "loss": 4.9849,
+      "step": 655
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00012137088699861465,
+      "loss": 4.7167,
+      "step": 656
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00012135750837350457,
+      "loss": 4.6299,
+      "step": 657
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00012134412974839449,
+      "loss": 4.6506,
+      "step": 658
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00012133075112328441,
+      "loss": 4.3563,
+      "step": 659
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00012131737249817433,
+      "loss": 4.5771,
+      "step": 660
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00012130399387306427,
+      "loss": 4.4903,
+      "step": 661
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00012129061524795419,
+      "loss": 4.4992,
+      "step": 662
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001212772366228441,
+      "loss": 4.6997,
+      "step": 663
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00012126385799773404,
+      "loss": 4.4887,
+      "step": 664
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00012125047937262396,
+      "loss": 4.4604,
+      "step": 665
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00012123710074751387,
+      "loss": 4.3907,
+      "step": 666
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001212237221224038,
+      "loss": 4.748,
+      "step": 667
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00012121034349729372,
+      "loss": 4.4132,
+      "step": 668
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00012119696487218365,
+      "loss": 4.4036,
+      "step": 669
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00012118358624707357,
+      "loss": 4.614,
+      "step": 670
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00012117020762196349,
+      "loss": 4.5547,
+      "step": 671
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00012115682899685343,
+      "loss": 4.634,
+      "step": 672
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00012114345037174335,
+      "loss": 4.6578,
+      "step": 673
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00012113007174663327,
+      "loss": 4.8709,
+      "step": 674
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00012111669312152319,
+      "loss": 4.3829,
+      "step": 675
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001211033144964131,
+      "loss": 4.6295,
+      "step": 676
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00012108993587130304,
+      "loss": 4.2637,
+      "step": 677
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00012107655724619296,
+      "loss": 4.1953,
+      "step": 678
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00012106317862108288,
+      "loss": 4.2317,
+      "step": 679
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00012104979999597281,
+      "loss": 4.1843,
+      "step": 680
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00012103642137086273,
+      "loss": 4.1675,
+      "step": 681
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00012102304274575265,
+      "loss": 4.014,
+      "step": 682
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00012100966412064257,
+      "loss": 4.223,
+      "step": 683
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001209962854955325,
+      "loss": 3.7616,
+      "step": 684
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00012098290687042243,
+      "loss": 3.9112,
+      "step": 685
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00012096952824531235,
+      "loss": 3.8686,
+      "step": 686
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00012095614962020227,
+      "loss": 3.8139,
+      "step": 687
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001209427709950922,
+      "loss": 3.6491,
+      "step": 688
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00012092939236998212,
+      "loss": 3.6294,
+      "step": 689
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00012091601374487204,
+      "loss": 3.696,
+      "step": 690
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00012090263511976196,
+      "loss": 3.8591,
+      "step": 691
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00012088925649465188,
+      "loss": 3.4703,
+      "step": 692
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00012087587786954181,
+      "loss": 3.7024,
+      "step": 693
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00012086249924443173,
+      "loss": 3.2265,
+      "step": 694
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00012084912061932167,
+      "loss": 3.3069,
+      "step": 695
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00012083574199421159,
+      "loss": 3.2252,
+      "step": 696
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001208223633691015,
+      "loss": 3.1884,
+      "step": 697
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00012080898474399143,
+      "loss": 3.1327,
+      "step": 698
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00012079560611888135,
+      "loss": 2.7246,
+      "step": 699
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00012078222749377127,
+      "loss": 2.8562,
+      "step": 700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001207688488686612,
+      "loss": 5.4132,
+      "step": 701
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00012075547024355112,
+      "loss": 5.2615,
+      "step": 702
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00012074209161844105,
+      "loss": 5.1017,
+      "step": 703
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00012072871299333097,
+      "loss": 4.8549,
+      "step": 704
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00012071533436822088,
+      "loss": 4.6373,
+      "step": 705
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00012070195574311081,
+      "loss": 4.5066,
+      "step": 706
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00012068857711800073,
+      "loss": 4.6969,
+      "step": 707
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00012067519849289067,
+      "loss": 4.2511,
+      "step": 708
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00012066181986778059,
+      "loss": 4.2244,
+      "step": 709
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00012064844124267051,
+      "loss": 4.2681,
+      "step": 710
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00012063506261756044,
+      "loss": 4.2439,
+      "step": 711
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00012062168399245036,
+      "loss": 4.2464,
+      "step": 712
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00012060830536734027,
+      "loss": 4.2153,
+      "step": 713
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001205949267422302,
+      "loss": 4.3904,
+      "step": 714
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00012058154811712012,
+      "loss": 4.3978,
+      "step": 715
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00012056816949201006,
+      "loss": 4.1834,
+      "step": 716
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00012055479086689998,
+      "loss": 4.1508,
+      "step": 717
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001205414122417899,
+      "loss": 4.1399,
+      "step": 718
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00012052803361667983,
+      "loss": 4.1044,
+      "step": 719
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00012051465499156974,
+      "loss": 4.1943,
+      "step": 720
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00012050127636645967,
+      "loss": 4.0906,
+      "step": 721
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00012048789774134959,
+      "loss": 4.1518,
+      "step": 722
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00012047451911623951,
+      "loss": 4.04,
+      "step": 723
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00012046114049112944,
+      "loss": 4.1327,
+      "step": 724
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00012044776186601936,
+      "loss": 4.0004,
+      "step": 725
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00012043438324090928,
+      "loss": 4.0169,
+      "step": 726
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00012042100461579922,
+      "loss": 4.0855,
+      "step": 727
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00012040762599068912,
+      "loss": 3.8438,
+      "step": 728
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00012039424736557906,
+      "loss": 3.9036,
+      "step": 729
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00012038086874046898,
+      "loss": 3.7812,
+      "step": 730
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001203674901153589,
+      "loss": 3.9648,
+      "step": 731
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00012035411149024883,
+      "loss": 3.9352,
+      "step": 732
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00012034073286513875,
+      "loss": 3.7022,
+      "step": 733
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00012032735424002867,
+      "loss": 3.927,
+      "step": 734
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001203139756149186,
+      "loss": 3.8299,
+      "step": 735
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00012030059698980851,
+      "loss": 3.7367,
+      "step": 736
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00012028721836469844,
+      "loss": 3.6928,
+      "step": 737
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00012027383973958836,
+      "loss": 3.7227,
+      "step": 738
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00012026046111447828,
+      "loss": 3.3979,
+      "step": 739
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00012024708248936822,
+      "loss": 3.762,
+      "step": 740
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00012023370386425814,
+      "loss": 3.7172,
+      "step": 741
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00012022032523914807,
+      "loss": 3.4368,
+      "step": 742
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00012020694661403799,
+      "loss": 3.4843,
+      "step": 743
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001201935679889279,
+      "loss": 3.267,
+      "step": 744
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00012018018936381783,
+      "loss": 3.2355,
+      "step": 745
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00012016681073870775,
+      "loss": 3.02,
+      "step": 746
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00012015343211359767,
+      "loss": 3.0818,
+      "step": 747
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001201400534884876,
+      "loss": 3.0638,
+      "step": 748
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00012012667486337752,
+      "loss": 2.7221,
+      "step": 749
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00012011329623826746,
+      "loss": 2.636,
+      "step": 750
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00012009991761315736,
+      "loss": 5.413,
+      "step": 751
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00012008653898804728,
+      "loss": 4.8036,
+      "step": 752
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00012007316036293722,
+      "loss": 4.8767,
+      "step": 753
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00012005978173782714,
+      "loss": 4.4234,
+      "step": 754
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00012004640311271707,
+      "loss": 4.4816,
+      "step": 755
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00012003302448760699,
+      "loss": 4.3345,
+      "step": 756
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00012001964586249691,
+      "loss": 4.1955,
+      "step": 757
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00012000626723738684,
+      "loss": 4.0613,
+      "step": 758
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00011999288861227675,
+      "loss": 4.2979,
+      "step": 759
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00011997950998716667,
+      "loss": 4.2565,
+      "step": 760
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001199661313620566,
+      "loss": 4.1554,
+      "step": 761
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00011995275273694652,
+      "loss": 4.1364,
+      "step": 762
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00011993937411183646,
+      "loss": 4.2789,
+      "step": 763
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00011992599548672638,
+      "loss": 4.3077,
+      "step": 764
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001199126168616163,
+      "loss": 4.0292,
+      "step": 765
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00011989923823650623,
+      "loss": 4.132,
+      "step": 766
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00011988585961139614,
+      "loss": 4.2988,
+      "step": 767
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00011987248098628606,
+      "loss": 4.1221,
+      "step": 768
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00011985910236117599,
+      "loss": 4.1591,
+      "step": 769
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00011984572373606591,
+      "loss": 4.0694,
+      "step": 770
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00011983234511095585,
+      "loss": 4.1995,
+      "step": 771
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00011981896648584577,
+      "loss": 3.979,
+      "step": 772
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00011980558786073569,
+      "loss": 3.9501,
+      "step": 773
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001197922092356256,
+      "loss": 3.8659,
+      "step": 774
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00011977883061051553,
+      "loss": 4.0409,
+      "step": 775
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00011976545198540546,
+      "loss": 4.0888,
+      "step": 776
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00011975207336029538,
+      "loss": 4.1929,
+      "step": 777
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001197386947351853,
+      "loss": 3.8422,
+      "step": 778
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00011972531611007523,
+      "loss": 3.8215,
+      "step": 779
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00011971193748496515,
+      "loss": 3.9469,
+      "step": 780
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00011969855885985507,
+      "loss": 4.094,
+      "step": 781
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00011968518023474499,
+      "loss": 3.8995,
+      "step": 782
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00011967180160963491,
+      "loss": 3.8155,
+      "step": 783
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00011965842298452485,
+      "loss": 3.7928,
+      "step": 784
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00011964504435941477,
+      "loss": 3.9603,
+      "step": 785
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00011963166573430469,
+      "loss": 3.7096,
+      "step": 786
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00011961828710919462,
+      "loss": 3.6168,
+      "step": 787
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00011960490848408454,
+      "loss": 3.9447,
+      "step": 788
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00011959152985897447,
+      "loss": 3.7617,
+      "step": 789
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00011957815123386438,
+      "loss": 3.5554,
+      "step": 790
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001195647726087543,
+      "loss": 3.4808,
+      "step": 791
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00011955139398364423,
+      "loss": 3.4568,
+      "step": 792
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00011953801535853415,
+      "loss": 3.4892,
+      "step": 793
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00011952463673342407,
+      "loss": 3.5827,
+      "step": 794
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000119511258108314,
+      "loss": 3.5314,
+      "step": 795
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00011949787948320393,
+      "loss": 3.2315,
+      "step": 796
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00011948450085809386,
+      "loss": 3.3781,
+      "step": 797
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00011947112223298377,
+      "loss": 2.8728,
+      "step": 798
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00011945774360787369,
+      "loss": 2.6696,
+      "step": 799
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00011944436498276362,
+      "loss": 2.7633,
+      "step": 800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00011943098635765354,
+      "loss": 5.2428,
+      "step": 801
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00011941760773254346,
+      "loss": 5.0211,
+      "step": 802
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001194042291074334,
+      "loss": 4.6765,
+      "step": 803
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00011939085048232331,
+      "loss": 4.3398,
+      "step": 804
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00011937747185721323,
+      "loss": 4.5166,
+      "step": 805
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00011936409323210315,
+      "loss": 4.3344,
+      "step": 806
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00011935071460699307,
+      "loss": 4.5373,
+      "step": 807
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00011933733598188301,
+      "loss": 4.2417,
+      "step": 808
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00011932395735677293,
+      "loss": 4.5244,
+      "step": 809
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00011931057873166286,
+      "loss": 4.3566,
+      "step": 810
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00011929720010655278,
+      "loss": 3.966,
+      "step": 811
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001192838214814427,
+      "loss": 4.1979,
+      "step": 812
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00011927044285633262,
+      "loss": 4.317,
+      "step": 813
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00011925706423122254,
+      "loss": 4.0557,
+      "step": 814
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00011924368560611246,
+      "loss": 4.3108,
+      "step": 815
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001192303069810024,
+      "loss": 4.1464,
+      "step": 816
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00011921692835589231,
+      "loss": 4.2143,
+      "step": 817
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00011920354973078225,
+      "loss": 4.2171,
+      "step": 818
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00011919017110567217,
+      "loss": 3.8697,
+      "step": 819
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00011917679248056209,
+      "loss": 4.0609,
+      "step": 820
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00011916341385545201,
+      "loss": 4.3286,
+      "step": 821
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00011915003523034193,
+      "loss": 4.0048,
+      "step": 822
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00011913665660523186,
+      "loss": 3.965,
+      "step": 823
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00011912327798012178,
+      "loss": 4.0199,
+      "step": 824
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001191098993550117,
+      "loss": 3.8507,
+      "step": 825
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00011909652072990164,
+      "loss": 3.7397,
+      "step": 826
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00011908314210479156,
+      "loss": 4.0744,
+      "step": 827
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00011906976347968146,
+      "loss": 4.0631,
+      "step": 828
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001190563848545714,
+      "loss": 3.8633,
+      "step": 829
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00011904300622946132,
+      "loss": 3.7681,
+      "step": 830
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00011902962760435125,
+      "loss": 3.9462,
+      "step": 831
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00011901624897924117,
+      "loss": 3.8938,
+      "step": 832
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00011900287035413109,
+      "loss": 3.8217,
+      "step": 833
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00011898949172902102,
+      "loss": 3.6824,
+      "step": 834
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00011897611310391094,
+      "loss": 3.7959,
+      "step": 835
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00011896273447880086,
+      "loss": 3.5203,
+      "step": 836
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00011894935585369078,
+      "loss": 3.8038,
+      "step": 837
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001189359772285807,
+      "loss": 3.7128,
+      "step": 838
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00011892259860347064,
+      "loss": 3.5859,
+      "step": 839
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00011890921997836056,
+      "loss": 3.5265,
+      "step": 840
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00011889584135325048,
+      "loss": 3.5559,
+      "step": 841
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00011888246272814041,
+      "loss": 3.4266,
+      "step": 842
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00011886908410303033,
+      "loss": 3.1939,
+      "step": 843
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00011885570547792025,
+      "loss": 3.1178,
+      "step": 844
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00011884232685281017,
+      "loss": 3.252,
+      "step": 845
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00011882894822770009,
+      "loss": 3.1803,
+      "step": 846
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00011881556960259002,
+      "loss": 2.8996,
+      "step": 847
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00011880219097747994,
+      "loss": 2.8522,
+      "step": 848
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00011878881235236986,
+      "loss": 2.7377,
+      "step": 849
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001187754337272598,
+      "loss": 2.6207,
+      "step": 850
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00011876205510214972,
+      "loss": 5.3191,
+      "step": 851
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00011874867647703964,
+      "loss": 4.7815,
+      "step": 852
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00011873529785192956,
+      "loss": 4.7361,
+      "step": 853
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00011872191922681948,
+      "loss": 4.3847,
+      "step": 854
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00011870854060170941,
+      "loss": 4.2154,
+      "step": 855
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00011869516197659933,
+      "loss": 4.3191,
+      "step": 856
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00011868178335148926,
+      "loss": 4.2787,
+      "step": 857
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00011866840472637918,
+      "loss": 4.1901,
+      "step": 858
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00011865502610126909,
+      "loss": 4.1654,
+      "step": 859
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00011864164747615902,
+      "loss": 4.1299,
+      "step": 860
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00011862826885104894,
+      "loss": 3.9869,
+      "step": 861
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00011861489022593886,
+      "loss": 4.2874,
+      "step": 862
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001186015116008288,
+      "loss": 4.0805,
+      "step": 863
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00011858813297571872,
+      "loss": 4.2155,
+      "step": 864
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00011857475435060865,
+      "loss": 4.2275,
+      "step": 865
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00011856137572549857,
+      "loss": 4.1073,
+      "step": 866
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00011854799710038848,
+      "loss": 4.1098,
+      "step": 867
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00011853461847527841,
+      "loss": 4.0736,
+      "step": 868
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00011852123985016833,
+      "loss": 4.4054,
+      "step": 869
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00011850786122505826,
+      "loss": 4.1285,
+      "step": 870
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00011849448259994818,
+      "loss": 3.9414,
+      "step": 871
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001184811039748381,
+      "loss": 3.9301,
+      "step": 872
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00011846772534972804,
+      "loss": 3.9247,
+      "step": 873
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00011845434672461796,
+      "loss": 4.2731,
+      "step": 874
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00011844096809950786,
+      "loss": 4.029,
+      "step": 875
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001184275894743978,
+      "loss": 3.9142,
+      "step": 876
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00011841421084928772,
+      "loss": 3.9867,
+      "step": 877
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00011840083222417765,
+      "loss": 3.9116,
+      "step": 878
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00011838745359906757,
+      "loss": 3.8666,
+      "step": 879
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00011837407497395749,
+      "loss": 3.715,
+      "step": 880
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00011836069634884742,
+      "loss": 3.6912,
+      "step": 881
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00011834731772373733,
+      "loss": 3.5959,
+      "step": 882
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00011833393909862725,
+      "loss": 3.7422,
+      "step": 883
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00011832056047351718,
+      "loss": 3.6521,
+      "step": 884
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001183071818484071,
+      "loss": 3.6406,
+      "step": 885
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00011829380322329704,
+      "loss": 3.43,
+      "step": 886
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00011828042459818696,
+      "loss": 3.66,
+      "step": 887
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00011826704597307688,
+      "loss": 3.4506,
+      "step": 888
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00011825366734796681,
+      "loss": 3.5952,
+      "step": 889
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00011824028872285672,
+      "loss": 3.2202,
+      "step": 890
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00011822691009774665,
+      "loss": 3.3703,
+      "step": 891
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00011821353147263657,
+      "loss": 3.204,
+      "step": 892
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00011820015284752649,
+      "loss": 3.207,
+      "step": 893
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00011818677422241643,
+      "loss": 3.151,
+      "step": 894
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00011817339559730635,
+      "loss": 3.2191,
+      "step": 895
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00011816001697219627,
+      "loss": 2.8205,
+      "step": 896
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001181466383470862,
+      "loss": 2.6755,
+      "step": 897
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001181332597219761,
+      "loss": 2.7717,
+      "step": 898
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00011811988109686604,
+      "loss": 2.9115,
+      "step": 899
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00011810650247175596,
+      "loss": 2.402,
+      "step": 900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00011809312384664588,
+      "loss": 5.3379,
+      "step": 901
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00011807974522153581,
+      "loss": 4.9145,
+      "step": 902
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00011806636659642573,
+      "loss": 4.6728,
+      "step": 903
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00011805298797131567,
+      "loss": 4.6578,
+      "step": 904
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00011803960934620559,
+      "loss": 4.1938,
+      "step": 905
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00011802623072109549,
+      "loss": 4.1966,
+      "step": 906
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00011801285209598543,
+      "loss": 4.3542,
+      "step": 907
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00011799947347087535,
+      "loss": 4.357,
+      "step": 908
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00011798609484576527,
+      "loss": 4.1798,
+      "step": 909
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001179727162206552,
+      "loss": 4.0289,
+      "step": 910
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00011795933759554512,
+      "loss": 4.1311,
+      "step": 911
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00011794595897043505,
+      "loss": 4.026,
+      "step": 912
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00011793258034532496,
+      "loss": 4.0865,
+      "step": 913
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00011791920172021488,
+      "loss": 4.0928,
+      "step": 914
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00011790582309510481,
+      "loss": 4.1276,
+      "step": 915
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00011789244446999473,
+      "loss": 4.2001,
+      "step": 916
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00011787906584488467,
+      "loss": 3.8628,
+      "step": 917
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00011786568721977459,
+      "loss": 4.189,
+      "step": 918
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00011785230859466451,
+      "loss": 4.0161,
+      "step": 919
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00011783892996955444,
+      "loss": 4.0884,
+      "step": 920
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00011782555134444435,
+      "loss": 3.933,
+      "step": 921
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00011781217271933427,
+      "loss": 3.9382,
+      "step": 922
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001177987940942242,
+      "loss": 3.9188,
+      "step": 923
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011778541546911412,
+      "loss": 3.9214,
+      "step": 924
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011777203684400405,
+      "loss": 3.73,
+      "step": 925
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011775865821889397,
+      "loss": 3.9926,
+      "step": 926
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001177452795937839,
+      "loss": 3.985,
+      "step": 927
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011773190096867383,
+      "loss": 3.9329,
+      "step": 928
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011771852234356373,
+      "loss": 3.6416,
+      "step": 929
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011770514371845365,
+      "loss": 3.7744,
+      "step": 930
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011769176509334359,
+      "loss": 3.6589,
+      "step": 931
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011767838646823351,
+      "loss": 3.7609,
+      "step": 932
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011766500784312344,
+      "loss": 3.9485,
+      "step": 933
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011765162921801336,
+      "loss": 3.8828,
+      "step": 934
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011763825059290328,
+      "loss": 3.373,
+      "step": 935
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001176248719677932,
+      "loss": 3.6739,
+      "step": 936
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011761149334268312,
+      "loss": 3.5979,
+      "step": 937
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011759811471757305,
+      "loss": 3.5828,
+      "step": 938
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011758473609246297,
+      "loss": 3.4539,
+      "step": 939
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001175713574673529,
+      "loss": 3.5869,
+      "step": 940
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011755797884224283,
+      "loss": 3.4094,
+      "step": 941
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011754460021713275,
+      "loss": 3.2794,
+      "step": 942
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011753122159202267,
+      "loss": 3.257,
+      "step": 943
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011751784296691259,
+      "loss": 3.2481,
+      "step": 944
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011750446434180251,
+      "loss": 3.171,
+      "step": 945
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011749108571669244,
+      "loss": 3.0366,
+      "step": 946
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011747770709158236,
+      "loss": 2.8523,
+      "step": 947
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011746432846647228,
+      "loss": 2.8948,
+      "step": 948
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011745094984136222,
+      "loss": 2.6685,
+      "step": 949
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011743757121625214,
+      "loss": 2.4777,
+      "step": 950
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011742419259114207,
+      "loss": 5.4026,
+      "step": 951
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011741081396603198,
+      "loss": 5.0879,
+      "step": 952
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001173974353409219,
+      "loss": 4.8459,
+      "step": 953
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011738405671581183,
+      "loss": 4.5631,
+      "step": 954
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011737067809070175,
+      "loss": 4.411,
+      "step": 955
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011735729946559167,
+      "loss": 4.2704,
+      "step": 956
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001173439208404816,
+      "loss": 4.2826,
+      "step": 957
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011733054221537152,
+      "loss": 4.4019,
+      "step": 958
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011731716359026146,
+      "loss": 4.1995,
+      "step": 959
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011730378496515136,
+      "loss": 4.0342,
+      "step": 960
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011729040634004128,
+      "loss": 4.0123,
+      "step": 961
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011727702771493122,
+      "loss": 4.2402,
+      "step": 962
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011726364908982114,
+      "loss": 4.0871,
+      "step": 963
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011725027046471106,
+      "loss": 3.9297,
+      "step": 964
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011723689183960099,
+      "loss": 4.1386,
+      "step": 965
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011722351321449091,
+      "loss": 4.054,
+      "step": 966
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011721013458938083,
+      "loss": 4.2414,
+      "step": 967
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011719675596427075,
+      "loss": 3.9786,
+      "step": 968
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011718337733916067,
+      "loss": 3.8723,
+      "step": 969
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001171699987140506,
+      "loss": 3.8843,
+      "step": 970
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011715662008894052,
+      "loss": 3.8656,
+      "step": 971
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011714324146383046,
+      "loss": 3.938,
+      "step": 972
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011712986283872038,
+      "loss": 3.8207,
+      "step": 973
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001171164842136103,
+      "loss": 3.9087,
+      "step": 974
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011710310558850022,
+      "loss": 3.9235,
+      "step": 975
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011708972696339014,
+      "loss": 3.6883,
+      "step": 976
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011707634833828006,
+      "loss": 3.6174,
+      "step": 977
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011706296971316999,
+      "loss": 3.8834,
+      "step": 978
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011704959108805991,
+      "loss": 3.7033,
+      "step": 979
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011703621246294984,
+      "loss": 3.8132,
+      "step": 980
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011702283383783976,
+      "loss": 3.7872,
+      "step": 981
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011700945521272968,
+      "loss": 3.8707,
+      "step": 982
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001169960765876196,
+      "loss": 3.8047,
+      "step": 983
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011698269796250952,
+      "loss": 3.4402,
+      "step": 984
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011696931933739946,
+      "loss": 3.6349,
+      "step": 985
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011695594071228938,
+      "loss": 3.4723,
+      "step": 986
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001169425620871793,
+      "loss": 3.5127,
+      "step": 987
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011692918346206923,
+      "loss": 3.2948,
+      "step": 988
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00011691580483695915,
+      "loss": 3.6125,
+      "step": 989
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00011690242621184906,
+      "loss": 3.1067,
+      "step": 990
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00011688904758673899,
+      "loss": 3.3753,
+      "step": 991
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00011687566896162891,
+      "loss": 3.2689,
+      "step": 992
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00011686229033651884,
+      "loss": 3.1529,
+      "step": 993
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00011684891171140876,
+      "loss": 3.0119,
+      "step": 994
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00011683553308629868,
+      "loss": 2.8843,
+      "step": 995
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00011682215446118862,
+      "loss": 2.7233,
+      "step": 996
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00011680877583607854,
+      "loss": 2.6406,
+      "step": 997
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00011679539721096844,
+      "loss": 2.8193,
+      "step": 998
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00011678201858585838,
+      "loss": 2.5364,
+      "step": 999
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001167686399607483,
+      "loss": 2.4062,
+      "step": 1000
+    },
+    {
+      "epoch": 0.31,
+      "eval_bleu": 0.0,
+      "eval_loss": 4.401661396026611,
+      "eval_runtime": 2573.0246,
+      "eval_samples_per_second": 5.736,
+      "eval_steps_per_second": 0.717,
+      "step": 1000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00011675526133563823,
+      "loss": 5.357,
+      "step": 1001
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00011674188271052815,
+      "loss": 5.033,
+      "step": 1002
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00011672850408541807,
+      "loss": 4.4124,
+      "step": 1003
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000116715125460308,
+      "loss": 4.709,
+      "step": 1004
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00011670174683519793,
+      "loss": 4.2526,
+      "step": 1005
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00011668836821008785,
+      "loss": 4.4101,
+      "step": 1006
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00011667498958497777,
+      "loss": 4.1548,
+      "step": 1007
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00011666161095986769,
+      "loss": 4.1352,
+      "step": 1008
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00011664823233475762,
+      "loss": 4.217,
+      "step": 1009
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00011663485370964754,
+      "loss": 4.3548,
+      "step": 1010
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00011662147508453746,
+      "loss": 4.2089,
+      "step": 1011
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00011660809645942739,
+      "loss": 3.7767,
+      "step": 1012
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001165947178343173,
+      "loss": 3.839,
+      "step": 1013
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00011658133920920723,
+      "loss": 4.0316,
+      "step": 1014
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00011656796058409715,
+      "loss": 4.0314,
+      "step": 1015
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00011655458195898707,
+      "loss": 4.1192,
+      "step": 1016
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000116541203333877,
+      "loss": 4.218,
+      "step": 1017
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00011652782470876693,
+      "loss": 3.769,
+      "step": 1018
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00011651444608365686,
+      "loss": 3.9649,
+      "step": 1019
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00011650106745854678,
+      "loss": 4.02,
+      "step": 1020
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00011648768883343669,
+      "loss": 4.1289,
+      "step": 1021
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00011647431020832662,
+      "loss": 3.8979,
+      "step": 1022
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00011646093158321654,
+      "loss": 3.8312,
+      "step": 1023
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00011644755295810646,
+      "loss": 3.9109,
+      "step": 1024
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00011643417433299639,
+      "loss": 3.6291,
+      "step": 1025
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00011642079570788631,
+      "loss": 3.4722,
+      "step": 1026
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00011640741708277625,
+      "loss": 3.7165,
+      "step": 1027
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00011639403845766617,
+      "loss": 3.8078,
+      "step": 1028
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00011638065983255607,
+      "loss": 3.8205,
+      "step": 1029
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000116367281207446,
+      "loss": 3.8619,
+      "step": 1030
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00011635390258233593,
+      "loss": 3.6046,
+      "step": 1031
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00011634052395722586,
+      "loss": 3.7037,
+      "step": 1032
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00011632714533211578,
+      "loss": 3.5694,
+      "step": 1033
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001163137667070057,
+      "loss": 3.5795,
+      "step": 1034
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00011630038808189563,
+      "loss": 3.3325,
+      "step": 1035
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00011628700945678555,
+      "loss": 3.4844,
+      "step": 1036
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00011627363083167546,
+      "loss": 3.3225,
+      "step": 1037
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001162602522065654,
+      "loss": 3.3438,
+      "step": 1038
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00011624687358145531,
+      "loss": 3.3813,
+      "step": 1039
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00011623349495634525,
+      "loss": 3.3748,
+      "step": 1040
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00011622011633123517,
+      "loss": 3.3635,
+      "step": 1041
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00011620673770612509,
+      "loss": 3.2493,
+      "step": 1042
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00011619335908101502,
+      "loss": 3.2222,
+      "step": 1043
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00011617998045590493,
+      "loss": 3.0538,
+      "step": 1044
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00011616660183079485,
+      "loss": 3.0005,
+      "step": 1045
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00011615322320568478,
+      "loss": 3.0796,
+      "step": 1046
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001161398445805747,
+      "loss": 2.8762,
+      "step": 1047
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00011612646595546463,
+      "loss": 2.5621,
+      "step": 1048
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00011611308733035455,
+      "loss": 2.6853,
+      "step": 1049
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00011609970870524447,
+      "loss": 2.4124,
+      "step": 1050
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00011608633008013441,
+      "loss": 5.5708,
+      "step": 1051
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00011607295145502431,
+      "loss": 4.6196,
+      "step": 1052
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00011605957282991425,
+      "loss": 4.5833,
+      "step": 1053
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00011604619420480417,
+      "loss": 4.6327,
+      "step": 1054
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00011603281557969409,
+      "loss": 4.4059,
+      "step": 1055
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00011601943695458402,
+      "loss": 4.3137,
+      "step": 1056
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00011600605832947394,
+      "loss": 4.0986,
+      "step": 1057
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00011599267970436386,
+      "loss": 4.2964,
+      "step": 1058
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001159793010792538,
+      "loss": 4.1307,
+      "step": 1059
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001159659224541437,
+      "loss": 3.9759,
+      "step": 1060
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00011595254382903364,
+      "loss": 4.2584,
+      "step": 1061
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00011593916520392356,
+      "loss": 3.9553,
+      "step": 1062
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00011592578657881348,
+      "loss": 3.8624,
+      "step": 1063
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00011591240795370341,
+      "loss": 3.9956,
+      "step": 1064
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00011589902932859333,
+      "loss": 3.9698,
+      "step": 1065
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00011588565070348326,
+      "loss": 3.9469,
+      "step": 1066
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00011587227207837317,
+      "loss": 3.77,
+      "step": 1067
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00011585889345326309,
+      "loss": 3.9409,
+      "step": 1068
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00011584551482815302,
+      "loss": 3.9935,
+      "step": 1069
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00011583213620304294,
+      "loss": 3.882,
+      "step": 1070
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00011581875757793286,
+      "loss": 3.9012,
+      "step": 1071
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001158053789528228,
+      "loss": 3.8196,
+      "step": 1072
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00011579200032771272,
+      "loss": 3.7804,
+      "step": 1073
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00011577862170260265,
+      "loss": 3.8186,
+      "step": 1074
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00011576524307749256,
+      "loss": 3.6668,
+      "step": 1075
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00011575186445238248,
+      "loss": 3.7345,
+      "step": 1076
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00011573848582727241,
+      "loss": 3.6726,
+      "step": 1077
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00011572510720216233,
+      "loss": 3.7943,
+      "step": 1078
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00011571172857705225,
+      "loss": 3.5872,
+      "step": 1079
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00011569834995194218,
+      "loss": 3.7842,
+      "step": 1080
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001156849713268321,
+      "loss": 3.645,
+      "step": 1081
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00011567159270172204,
+      "loss": 3.6889,
+      "step": 1082
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00011565821407661194,
+      "loss": 3.5333,
+      "step": 1083
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00011564483545150186,
+      "loss": 3.2074,
+      "step": 1084
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001156314568263918,
+      "loss": 3.4914,
+      "step": 1085
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00011561807820128172,
+      "loss": 3.6842,
+      "step": 1086
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00011560469957617165,
+      "loss": 3.3834,
+      "step": 1087
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00011559132095106157,
+      "loss": 3.2788,
+      "step": 1088
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00011557794232595149,
+      "loss": 3.1994,
+      "step": 1089
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00011556456370084142,
+      "loss": 3.4681,
+      "step": 1090
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00011555118507573133,
+      "loss": 3.2365,
+      "step": 1091
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00011553780645062125,
+      "loss": 3.1054,
+      "step": 1092
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00011552442782551118,
+      "loss": 3.0266,
+      "step": 1093
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001155110492004011,
+      "loss": 2.8698,
+      "step": 1094
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00011549767057529104,
+      "loss": 2.9816,
+      "step": 1095
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00011548429195018096,
+      "loss": 2.5871,
+      "step": 1096
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00011547091332507088,
+      "loss": 2.6161,
+      "step": 1097
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001154575346999608,
+      "loss": 2.5678,
+      "step": 1098
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00011544415607485072,
+      "loss": 2.4859,
+      "step": 1099
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00011543077744974065,
+      "loss": 2.6307,
+      "step": 1100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00011541739882463057,
+      "loss": 5.0397,
+      "step": 1101
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00011540402019952049,
+      "loss": 4.7119,
+      "step": 1102
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00011539064157441042,
+      "loss": 4.6501,
+      "step": 1103
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00011537726294930034,
+      "loss": 4.2657,
+      "step": 1104
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00011536388432419026,
+      "loss": 3.9446,
+      "step": 1105
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00011535050569908018,
+      "loss": 4.0682,
+      "step": 1106
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001153371270739701,
+      "loss": 4.4486,
+      "step": 1107
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00011532374844886004,
+      "loss": 4.1464,
+      "step": 1108
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00011531036982374996,
+      "loss": 3.9531,
+      "step": 1109
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00011529699119863988,
+      "loss": 4.0982,
+      "step": 1110
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00011528361257352981,
+      "loss": 4.0009,
+      "step": 1111
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00011527023394841973,
+      "loss": 3.812,
+      "step": 1112
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00011525685532330967,
+      "loss": 3.9902,
+      "step": 1113
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00011524347669819957,
+      "loss": 4.1126,
+      "step": 1114
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00011523009807308949,
+      "loss": 4.0814,
+      "step": 1115
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00011521671944797943,
+      "loss": 3.8563,
+      "step": 1116
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00011520334082286935,
+      "loss": 3.7882,
+      "step": 1117
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00011518996219775927,
+      "loss": 3.8479,
+      "step": 1118
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001151765835726492,
+      "loss": 3.7565,
+      "step": 1119
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00011516320494753912,
+      "loss": 3.7954,
+      "step": 1120
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00011514982632242904,
+      "loss": 3.7955,
+      "step": 1121
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00011513644769731896,
+      "loss": 3.7873,
+      "step": 1122
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00011512306907220888,
+      "loss": 3.7054,
+      "step": 1123
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00011510969044709881,
+      "loss": 3.7683,
+      "step": 1124
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00011509631182198873,
+      "loss": 3.6615,
+      "step": 1125
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00011508293319687865,
+      "loss": 3.4977,
+      "step": 1126
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00011506955457176859,
+      "loss": 3.6514,
+      "step": 1127
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001150561759466585,
+      "loss": 3.5261,
+      "step": 1128
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00011504279732154843,
+      "loss": 3.7045,
+      "step": 1129
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00011502941869643835,
+      "loss": 3.5247,
+      "step": 1130
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00011501604007132827,
+      "loss": 3.7823,
+      "step": 1131
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001150026614462182,
+      "loss": 3.4978,
+      "step": 1132
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00011498928282110812,
+      "loss": 3.6035,
+      "step": 1133
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00011497590419599805,
+      "loss": 3.5412,
+      "step": 1134
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00011496252557088797,
+      "loss": 3.4305,
+      "step": 1135
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00011494914694577789,
+      "loss": 3.4225,
+      "step": 1136
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00011493576832066781,
+      "loss": 2.9879,
+      "step": 1137
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00011492238969555773,
+      "loss": 3.2837,
+      "step": 1138
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00011490901107044765,
+      "loss": 3.08,
+      "step": 1139
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00011489563244533759,
+      "loss": 3.0544,
+      "step": 1140
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001148822538202275,
+      "loss": 3.1062,
+      "step": 1141
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00011486887519511744,
+      "loss": 2.9521,
+      "step": 1142
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00011485549657000736,
+      "loss": 2.7877,
+      "step": 1143
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00011484211794489727,
+      "loss": 3.1213,
+      "step": 1144
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001148287393197872,
+      "loss": 3.1196,
+      "step": 1145
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00011481536069467712,
+      "loss": 2.8165,
+      "step": 1146
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00011480198206956705,
+      "loss": 2.4731,
+      "step": 1147
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00011478860344445697,
+      "loss": 2.4453,
+      "step": 1148
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001147752248193469,
+      "loss": 2.3895,
+      "step": 1149
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00011476184619423683,
+      "loss": 2.2447,
+      "step": 1150
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00011474846756912675,
+      "loss": 5.0798,
+      "step": 1151
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00011473508894401665,
+      "loss": 4.7653,
+      "step": 1152
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00011472171031890659,
+      "loss": 4.7029,
+      "step": 1153
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00011470833169379651,
+      "loss": 4.546,
+      "step": 1154
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00011469495306868644,
+      "loss": 4.0571,
+      "step": 1155
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00011468157444357636,
+      "loss": 4.1215,
+      "step": 1156
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00011466819581846628,
+      "loss": 4.0926,
+      "step": 1157
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00011465481719335621,
+      "loss": 4.2055,
+      "step": 1158
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00011464143856824613,
+      "loss": 4.0501,
+      "step": 1159
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00011462805994313604,
+      "loss": 3.961,
+      "step": 1160
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00011461468131802597,
+      "loss": 3.8048,
+      "step": 1161
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001146013026929159,
+      "loss": 3.8922,
+      "step": 1162
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00011458792406780583,
+      "loss": 3.8667,
+      "step": 1163
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00011457454544269575,
+      "loss": 3.8412,
+      "step": 1164
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00011456116681758567,
+      "loss": 3.733,
+      "step": 1165
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001145477881924756,
+      "loss": 4.0065,
+      "step": 1166
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00011453440956736552,
+      "loss": 3.6497,
+      "step": 1167
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00011452103094225544,
+      "loss": 3.9308,
+      "step": 1168
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00011450765231714536,
+      "loss": 3.7676,
+      "step": 1169
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00011449427369203528,
+      "loss": 3.8859,
+      "step": 1170
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00011448089506692521,
+      "loss": 3.5971,
+      "step": 1171
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00011446751644181513,
+      "loss": 3.8938,
+      "step": 1172
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00011445413781670505,
+      "loss": 4.1355,
+      "step": 1173
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00011444075919159499,
+      "loss": 4.1524,
+      "step": 1174
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001144273805664849,
+      "loss": 3.8213,
+      "step": 1175
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00011441400194137483,
+      "loss": 4.1402,
+      "step": 1176
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00011440062331626475,
+      "loss": 3.975,
+      "step": 1177
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00011438724469115467,
+      "loss": 3.8081,
+      "step": 1178
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001143738660660446,
+      "loss": 4.0083,
+      "step": 1179
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00011436048744093452,
+      "loss": 3.6284,
+      "step": 1180
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00011434710881582446,
+      "loss": 4.1683,
+      "step": 1181
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00011433373019071438,
+      "loss": 3.8377,
+      "step": 1182
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00011432035156560428,
+      "loss": 3.5418,
+      "step": 1183
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00011430697294049422,
+      "loss": 3.8074,
+      "step": 1184
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00011429359431538414,
+      "loss": 3.4783,
+      "step": 1185
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00011428021569027406,
+      "loss": 3.5146,
+      "step": 1186
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00011426683706516399,
+      "loss": 3.3452,
+      "step": 1187
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00011425345844005391,
+      "loss": 3.1917,
+      "step": 1188
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00011424007981494384,
+      "loss": 3.2938,
+      "step": 1189
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00011422670118983376,
+      "loss": 3.209,
+      "step": 1190
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00011421332256472367,
+      "loss": 3.368,
+      "step": 1191
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001141999439396136,
+      "loss": 2.8557,
+      "step": 1192
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00011418656531450352,
+      "loss": 2.8708,
+      "step": 1193
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00011417318668939344,
+      "loss": 2.8461,
+      "step": 1194
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00011415980806428338,
+      "loss": 2.8313,
+      "step": 1195
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001141464294391733,
+      "loss": 2.5797,
+      "step": 1196
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00011413305081406323,
+      "loss": 2.5016,
+      "step": 1197
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00011411967218895314,
+      "loss": 2.4904,
+      "step": 1198
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00011410629356384306,
+      "loss": 2.2821,
+      "step": 1199
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00011409291493873299,
+      "loss": 2.3145,
+      "step": 1200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00011407953631362291,
+      "loss": 5.0331,
+      "step": 1201
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00011406615768851284,
+      "loss": 4.8974,
+      "step": 1202
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00011405277906340276,
+      "loss": 4.8102,
+      "step": 1203
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00011403940043829268,
+      "loss": 4.7,
+      "step": 1204
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00011402602181318262,
+      "loss": 4.3582,
+      "step": 1205
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00011401264318807252,
+      "loss": 4.413,
+      "step": 1206
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00011399926456296244,
+      "loss": 4.1213,
+      "step": 1207
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00011398588593785238,
+      "loss": 4.018,
+      "step": 1208
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001139725073127423,
+      "loss": 4.0686,
+      "step": 1209
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00011395912868763223,
+      "loss": 4.2432,
+      "step": 1210
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00011394575006252215,
+      "loss": 4.0145,
+      "step": 1211
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00011393237143741207,
+      "loss": 3.9932,
+      "step": 1212
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000113918992812302,
+      "loss": 3.8357,
+      "step": 1213
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00011390561418719191,
+      "loss": 3.8939,
+      "step": 1214
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00011389223556208184,
+      "loss": 3.5397,
+      "step": 1215
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00011387885693697176,
+      "loss": 3.6141,
+      "step": 1216
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00011386547831186168,
+      "loss": 3.8005,
+      "step": 1217
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00011385209968675162,
+      "loss": 3.896,
+      "step": 1218
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00011383872106164154,
+      "loss": 3.5686,
+      "step": 1219
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00011382534243653146,
+      "loss": 3.6332,
+      "step": 1220
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00011381196381142139,
+      "loss": 3.6103,
+      "step": 1221
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001137985851863113,
+      "loss": 3.77,
+      "step": 1222
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00011378520656120123,
+      "loss": 3.6481,
+      "step": 1223
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00011377182793609115,
+      "loss": 3.5826,
+      "step": 1224
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00011375844931098107,
+      "loss": 3.6392,
+      "step": 1225
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000113745070685871,
+      "loss": 3.745,
+      "step": 1226
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00011373169206076092,
+      "loss": 3.7214,
+      "step": 1227
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00011371831343565086,
+      "loss": 3.54,
+      "step": 1228
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00011370493481054076,
+      "loss": 3.3323,
+      "step": 1229
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00011369155618543068,
+      "loss": 3.4613,
+      "step": 1230
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00011367817756032062,
+      "loss": 3.1968,
+      "step": 1231
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00011366479893521054,
+      "loss": 3.495,
+      "step": 1232
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00011365142031010046,
+      "loss": 3.181,
+      "step": 1233
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00011363804168499039,
+      "loss": 3.2589,
+      "step": 1234
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00011362466305988031,
+      "loss": 3.1378,
+      "step": 1235
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00011361128443477025,
+      "loss": 3.1856,
+      "step": 1236
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00011359790580966015,
+      "loss": 3.0043,
+      "step": 1237
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00011358452718455007,
+      "loss": 3.209,
+      "step": 1238
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00011357114855944,
+      "loss": 2.9475,
+      "step": 1239
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00011355776993432993,
+      "loss": 2.8116,
+      "step": 1240
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00011354439130921985,
+      "loss": 3.0458,
+      "step": 1241
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00011353101268410978,
+      "loss": 2.9641,
+      "step": 1242
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001135176340589997,
+      "loss": 2.9518,
+      "step": 1243
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00011350425543388963,
+      "loss": 2.7537,
+      "step": 1244
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00011349087680877954,
+      "loss": 2.5479,
+      "step": 1245
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00011347749818366946,
+      "loss": 2.5805,
+      "step": 1246
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00011346411955855939,
+      "loss": 2.5881,
+      "step": 1247
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00011345074093344931,
+      "loss": 2.3238,
+      "step": 1248
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00011343736230833925,
+      "loss": 2.1842,
+      "step": 1249
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00011342398368322917,
+      "loss": 2.0312,
+      "step": 1250
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00011341060505811909,
+      "loss": 5.0088,
+      "step": 1251
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000113397226433009,
+      "loss": 4.723,
+      "step": 1252
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00011338384780789893,
+      "loss": 4.414,
+      "step": 1253
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00011337046918278885,
+      "loss": 4.3106,
+      "step": 1254
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00011335709055767878,
+      "loss": 4.111,
+      "step": 1255
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001133437119325687,
+      "loss": 4.0575,
+      "step": 1256
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00011333033330745863,
+      "loss": 3.9518,
+      "step": 1257
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00011331695468234855,
+      "loss": 4.0905,
+      "step": 1258
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00011330357605723847,
+      "loss": 3.8396,
+      "step": 1259
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00011329019743212839,
+      "loss": 3.7184,
+      "step": 1260
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00011327681880701831,
+      "loss": 3.8814,
+      "step": 1261
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00011326344018190825,
+      "loss": 3.8185,
+      "step": 1262
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00011325006155679817,
+      "loss": 3.7233,
+      "step": 1263
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00011323668293168809,
+      "loss": 3.4773,
+      "step": 1264
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00011322330430657802,
+      "loss": 3.7191,
+      "step": 1265
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00011320992568146794,
+      "loss": 3.9206,
+      "step": 1266
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00011319654705635786,
+      "loss": 3.6206,
+      "step": 1267
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00011318316843124778,
+      "loss": 3.691,
+      "step": 1268
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001131697898061377,
+      "loss": 3.8661,
+      "step": 1269
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00011315641118102763,
+      "loss": 3.6171,
+      "step": 1270
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00011314303255591755,
+      "loss": 3.54,
+      "step": 1271
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00011312965393080747,
+      "loss": 3.4782,
+      "step": 1272
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00011311627530569741,
+      "loss": 3.4746,
+      "step": 1273
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00011310289668058733,
+      "loss": 3.5564,
+      "step": 1274
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00011308951805547725,
+      "loss": 3.3971,
+      "step": 1275
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00011307613943036717,
+      "loss": 3.519,
+      "step": 1276
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00011306276080525709,
+      "loss": 3.4912,
+      "step": 1277
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00011304938218014702,
+      "loss": 3.2689,
+      "step": 1278
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00011303600355503694,
+      "loss": 3.1875,
+      "step": 1279
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00011302262492992686,
+      "loss": 3.2598,
+      "step": 1280
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001130092463048168,
+      "loss": 3.4091,
+      "step": 1281
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00011299586767970671,
+      "loss": 3.3329,
+      "step": 1282
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00011298248905459663,
+      "loss": 3.239,
+      "step": 1283
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00011296911042948655,
+      "loss": 3.2354,
+      "step": 1284
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00011295573180437647,
+      "loss": 3.1335,
+      "step": 1285
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00011294235317926641,
+      "loss": 3.2876,
+      "step": 1286
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00011292897455415633,
+      "loss": 3.0815,
+      "step": 1287
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00011291559592904625,
+      "loss": 2.9928,
+      "step": 1288
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00011290221730393618,
+      "loss": 3.0472,
+      "step": 1289
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001128888386788261,
+      "loss": 2.9032,
+      "step": 1290
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00011287546005371602,
+      "loss": 2.8466,
+      "step": 1291
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00011286208142860594,
+      "loss": 2.9736,
+      "step": 1292
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00011284870280349586,
+      "loss": 2.7651,
+      "step": 1293
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001128353241783858,
+      "loss": 2.6522,
+      "step": 1294
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00011282194555327572,
+      "loss": 2.4532,
+      "step": 1295
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00011280856692816565,
+      "loss": 2.2228,
+      "step": 1296
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00011279518830305557,
+      "loss": 2.5466,
+      "step": 1297
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00011278180967794549,
+      "loss": 2.1872,
+      "step": 1298
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00011276843105283541,
+      "loss": 2.2953,
+      "step": 1299
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00011275505242772533,
+      "loss": 1.9457,
+      "step": 1300
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00011274167380261525,
+      "loss": 4.7859,
+      "step": 1301
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00011272829517750518,
+      "loss": 4.5864,
+      "step": 1302
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001127149165523951,
+      "loss": 4.048,
+      "step": 1303
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00011270153792728504,
+      "loss": 3.8965,
+      "step": 1304
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00011268815930217496,
+      "loss": 4.0971,
+      "step": 1305
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00011267478067706486,
+      "loss": 3.723,
+      "step": 1306
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001126614020519548,
+      "loss": 3.9887,
+      "step": 1307
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00011264802342684472,
+      "loss": 3.8866,
+      "step": 1308
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00011263464480173464,
+      "loss": 3.6946,
+      "step": 1309
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00011262126617662457,
+      "loss": 3.7646,
+      "step": 1310
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00011260788755151449,
+      "loss": 3.6768,
+      "step": 1311
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00011259450892640442,
+      "loss": 3.8485,
+      "step": 1312
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00011258113030129434,
+      "loss": 3.6996,
+      "step": 1313
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00011256775167618425,
+      "loss": 3.6045,
+      "step": 1314
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00011255437305107418,
+      "loss": 3.5345,
+      "step": 1315
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001125409944259641,
+      "loss": 3.6557,
+      "step": 1316
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00011252761580085404,
+      "loss": 3.6043,
+      "step": 1317
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00011251423717574396,
+      "loss": 3.4231,
+      "step": 1318
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00011250085855063388,
+      "loss": 3.4771,
+      "step": 1319
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00011248747992552381,
+      "loss": 3.5948,
+      "step": 1320
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00011247410130041373,
+      "loss": 3.5317,
+      "step": 1321
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00011246072267530364,
+      "loss": 3.4307,
+      "step": 1322
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00011244734405019357,
+      "loss": 3.3129,
+      "step": 1323
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00011243396542508349,
+      "loss": 3.4476,
+      "step": 1324
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00011242058679997342,
+      "loss": 3.1799,
+      "step": 1325
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00011240720817486334,
+      "loss": 3.4767,
+      "step": 1326
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00011239382954975326,
+      "loss": 3.3237,
+      "step": 1327
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001123804509246432,
+      "loss": 3.3302,
+      "step": 1328
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001123670722995331,
+      "loss": 3.3581,
+      "step": 1329
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00011235369367442304,
+      "loss": 3.3294,
+      "step": 1330
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00011234031504931296,
+      "loss": 3.1518,
+      "step": 1331
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00011232693642420288,
+      "loss": 2.9894,
+      "step": 1332
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00011231355779909281,
+      "loss": 3.0272,
+      "step": 1333
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00011230017917398273,
+      "loss": 3.2779,
+      "step": 1334
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00011228680054887265,
+      "loss": 3.0728,
+      "step": 1335
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00011227342192376258,
+      "loss": 2.9228,
+      "step": 1336
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00011226004329865249,
+      "loss": 3.1192,
+      "step": 1337
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00011224666467354242,
+      "loss": 2.8703,
+      "step": 1338
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00011223328604843234,
+      "loss": 2.9831,
+      "step": 1339
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00011221990742332226,
+      "loss": 2.744,
+      "step": 1340
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001122065287982122,
+      "loss": 2.723,
+      "step": 1341
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00011219315017310212,
+      "loss": 2.8749,
+      "step": 1342
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00011217977154799205,
+      "loss": 2.8916,
+      "step": 1343
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00011216639292288197,
+      "loss": 2.6373,
+      "step": 1344
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011215301429777188,
+      "loss": 2.3631,
+      "step": 1345
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011213963567266181,
+      "loss": 2.6787,
+      "step": 1346
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011212625704755173,
+      "loss": 2.1983,
+      "step": 1347
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011211287842244165,
+      "loss": 2.0912,
+      "step": 1348
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011209949979733159,
+      "loss": 2.2622,
+      "step": 1349
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001120861211722215,
+      "loss": 1.9463,
+      "step": 1350
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011207274254711144,
+      "loss": 4.8788,
+      "step": 1351
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011205936392200136,
+      "loss": 4.4668,
+      "step": 1352
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011204598529689127,
+      "loss": 3.986,
+      "step": 1353
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001120326066717812,
+      "loss": 4.0261,
+      "step": 1354
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011201922804667112,
+      "loss": 3.8636,
+      "step": 1355
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011200584942156104,
+      "loss": 3.8136,
+      "step": 1356
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011199247079645097,
+      "loss": 3.5296,
+      "step": 1357
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011197909217134089,
+      "loss": 3.7353,
+      "step": 1358
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011196571354623083,
+      "loss": 3.647,
+      "step": 1359
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011195233492112073,
+      "loss": 3.8598,
+      "step": 1360
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011193895629601065,
+      "loss": 3.7061,
+      "step": 1361
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011192557767090059,
+      "loss": 3.6227,
+      "step": 1362
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001119121990457905,
+      "loss": 3.4388,
+      "step": 1363
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011189882042068044,
+      "loss": 3.5226,
+      "step": 1364
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011188544179557036,
+      "loss": 3.5021,
+      "step": 1365
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011187206317046028,
+      "loss": 3.3148,
+      "step": 1366
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011185868454535021,
+      "loss": 3.4855,
+      "step": 1367
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011184530592024012,
+      "loss": 3.4555,
+      "step": 1368
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011183192729513004,
+      "loss": 3.642,
+      "step": 1369
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011181854867001997,
+      "loss": 3.6983,
+      "step": 1370
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011180517004490989,
+      "loss": 3.4566,
+      "step": 1371
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011179179141979983,
+      "loss": 3.3444,
+      "step": 1372
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011177841279468975,
+      "loss": 3.3791,
+      "step": 1373
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011176503416957967,
+      "loss": 3.2171,
+      "step": 1374
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001117516555444696,
+      "loss": 3.2764,
+      "step": 1375
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001117382769193595,
+      "loss": 3.1906,
+      "step": 1376
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011172489829424944,
+      "loss": 3.3562,
+      "step": 1377
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011171151966913936,
+      "loss": 3.1631,
+      "step": 1378
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011169814104402928,
+      "loss": 3.2195,
+      "step": 1379
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011168476241891921,
+      "loss": 3.2229,
+      "step": 1380
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011167138379380913,
+      "loss": 3.0755,
+      "step": 1381
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011165800516869905,
+      "loss": 3.3287,
+      "step": 1382
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011164462654358897,
+      "loss": 3.1735,
+      "step": 1383
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001116312479184789,
+      "loss": 2.9212,
+      "step": 1384
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011161786929336883,
+      "loss": 2.9097,
+      "step": 1385
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011160449066825875,
+      "loss": 2.9321,
+      "step": 1386
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011159111204314867,
+      "loss": 2.9358,
+      "step": 1387
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001115777334180386,
+      "loss": 2.9639,
+      "step": 1388
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011156435479292852,
+      "loss": 3.0145,
+      "step": 1389
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011155097616781844,
+      "loss": 2.9277,
+      "step": 1390
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011153759754270836,
+      "loss": 2.5737,
+      "step": 1391
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011152421891759828,
+      "loss": 2.5317,
+      "step": 1392
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011151084029248821,
+      "loss": 2.683,
+      "step": 1393
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011149746166737813,
+      "loss": 2.5851,
+      "step": 1394
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011148408304226805,
+      "loss": 2.5563,
+      "step": 1395
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011147070441715799,
+      "loss": 2.3509,
+      "step": 1396
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011145732579204791,
+      "loss": 2.2843,
+      "step": 1397
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011144394716693784,
+      "loss": 1.9618,
+      "step": 1398
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011143056854182775,
+      "loss": 2.103,
+      "step": 1399
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011141718991671767,
+      "loss": 2.1388,
+      "step": 1400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001114038112916076,
+      "loss": 4.7329,
+      "step": 1401
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011139043266649752,
+      "loss": 4.3818,
+      "step": 1402
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011137705404138744,
+      "loss": 4.3114,
+      "step": 1403
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011136367541627737,
+      "loss": 3.9008,
+      "step": 1404
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001113502967911673,
+      "loss": 3.7817,
+      "step": 1405
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011133691816605723,
+      "loss": 3.5171,
+      "step": 1406
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011132353954094713,
+      "loss": 3.8082,
+      "step": 1407
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011131016091583705,
+      "loss": 3.3768,
+      "step": 1408
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011129678229072699,
+      "loss": 3.5902,
+      "step": 1409
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011128340366561691,
+      "loss": 3.514,
+      "step": 1410
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011127002504050684,
+      "loss": 3.5394,
+      "step": 1411
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011125664641539676,
+      "loss": 3.5406,
+      "step": 1412
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011124326779028668,
+      "loss": 3.4886,
+      "step": 1413
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001112298891651766,
+      "loss": 3.4425,
+      "step": 1414
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011121651054006652,
+      "loss": 3.3981,
+      "step": 1415
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011120313191495644,
+      "loss": 3.304,
+      "step": 1416
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011118975328984638,
+      "loss": 3.2206,
+      "step": 1417
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001111763746647363,
+      "loss": 3.4546,
+      "step": 1418
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011116299603962623,
+      "loss": 3.3046,
+      "step": 1419
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011114961741451615,
+      "loss": 3.281,
+      "step": 1420
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011113623878940607,
+      "loss": 3.2634,
+      "step": 1421
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011112286016429599,
+      "loss": 3.4966,
+      "step": 1422
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011110948153918591,
+      "loss": 3.2493,
+      "step": 1423
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011109610291407584,
+      "loss": 3.4803,
+      "step": 1424
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011108272428896576,
+      "loss": 3.3123,
+      "step": 1425
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011106934566385568,
+      "loss": 3.4245,
+      "step": 1426
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011105596703874562,
+      "loss": 3.2904,
+      "step": 1427
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011104258841363554,
+      "loss": 3.1257,
+      "step": 1428
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011102920978852546,
+      "loss": 3.1289,
+      "step": 1429
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011101583116341538,
+      "loss": 2.9837,
+      "step": 1430
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001110024525383053,
+      "loss": 3.3553,
+      "step": 1431
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011098907391319523,
+      "loss": 2.9292,
+      "step": 1432
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011097569528808515,
+      "loss": 3.0231,
+      "step": 1433
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011096231666297507,
+      "loss": 3.1111,
+      "step": 1434
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000110948938037865,
+      "loss": 2.869,
+      "step": 1435
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011093555941275492,
+      "loss": 2.8193,
+      "step": 1436
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011092218078764483,
+      "loss": 2.9502,
+      "step": 1437
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011090880216253476,
+      "loss": 2.5276,
+      "step": 1438
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011089542353742468,
+      "loss": 2.566,
+      "step": 1439
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011088204491231462,
+      "loss": 2.869,
+      "step": 1440
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011086866628720454,
+      "loss": 2.6115,
+      "step": 1441
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011085528766209446,
+      "loss": 2.4596,
+      "step": 1442
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011084190903698439,
+      "loss": 2.494,
+      "step": 1443
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011082853041187431,
+      "loss": 2.5573,
+      "step": 1444
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011081515178676423,
+      "loss": 2.4597,
+      "step": 1445
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011080177316165415,
+      "loss": 2.254,
+      "step": 1446
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011078839453654407,
+      "loss": 2.1121,
+      "step": 1447
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000110775015911434,
+      "loss": 1.8925,
+      "step": 1448
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011076163728632392,
+      "loss": 2.0516,
+      "step": 1449
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011074825866121384,
+      "loss": 1.954,
+      "step": 1450
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011073488003610378,
+      "loss": 5.1723,
+      "step": 1451
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001107215014109937,
+      "loss": 4.701,
+      "step": 1452
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011070812278588362,
+      "loss": 4.3224,
+      "step": 1453
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011069474416077354,
+      "loss": 4.0479,
+      "step": 1454
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011068136553566346,
+      "loss": 3.8328,
+      "step": 1455
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011066798691055339,
+      "loss": 3.6558,
+      "step": 1456
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011065460828544331,
+      "loss": 3.6213,
+      "step": 1457
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011064122966033324,
+      "loss": 3.3516,
+      "step": 1458
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011062785103522316,
+      "loss": 3.4103,
+      "step": 1459
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011061447241011308,
+      "loss": 3.4662,
+      "step": 1460
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000110601093785003,
+      "loss": 3.4702,
+      "step": 1461
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011058771515989292,
+      "loss": 3.3184,
+      "step": 1462
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011057433653478284,
+      "loss": 3.3716,
+      "step": 1463
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011056095790967278,
+      "loss": 3.3767,
+      "step": 1464
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001105475792845627,
+      "loss": 3.3906,
+      "step": 1465
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011053420065945263,
+      "loss": 3.2746,
+      "step": 1466
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011052082203434255,
+      "loss": 3.4216,
+      "step": 1467
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011050744340923246,
+      "loss": 3.2336,
+      "step": 1468
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011049406478412239,
+      "loss": 3.3426,
+      "step": 1469
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011048068615901231,
+      "loss": 3.6974,
+      "step": 1470
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011046730753390223,
+      "loss": 3.1189,
+      "step": 1471
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011045392890879217,
+      "loss": 3.0309,
+      "step": 1472
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011044055028368209,
+      "loss": 3.0612,
+      "step": 1473
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011042717165857202,
+      "loss": 3.2096,
+      "step": 1474
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011041379303346194,
+      "loss": 3.1176,
+      "step": 1475
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011040041440835185,
+      "loss": 3.0923,
+      "step": 1476
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011038703578324178,
+      "loss": 2.9637,
+      "step": 1477
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001103736571581317,
+      "loss": 3.0492,
+      "step": 1478
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011036027853302163,
+      "loss": 2.8166,
+      "step": 1479
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011034689990791155,
+      "loss": 2.802,
+      "step": 1480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011033352128280147,
+      "loss": 2.9042,
+      "step": 1481
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001103201426576914,
+      "loss": 3.1374,
+      "step": 1482
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011030676403258133,
+      "loss": 2.7653,
+      "step": 1483
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011029338540747123,
+      "loss": 2.8816,
+      "step": 1484
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011028000678236117,
+      "loss": 2.7777,
+      "step": 1485
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011026662815725109,
+      "loss": 2.8862,
+      "step": 1486
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011025324953214102,
+      "loss": 2.7836,
+      "step": 1487
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011023987090703094,
+      "loss": 2.6697,
+      "step": 1488
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011022649228192086,
+      "loss": 2.6949,
+      "step": 1489
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001102131136568108,
+      "loss": 2.6052,
+      "step": 1490
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001101997350317007,
+      "loss": 2.4787,
+      "step": 1491
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011018635640659063,
+      "loss": 2.5955,
+      "step": 1492
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011017297778148055,
+      "loss": 2.4765,
+      "step": 1493
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011015959915637047,
+      "loss": 2.2732,
+      "step": 1494
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011014622053126041,
+      "loss": 2.2335,
+      "step": 1495
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011013284190615033,
+      "loss": 2.1731,
+      "step": 1496
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011011946328104025,
+      "loss": 2.2276,
+      "step": 1497
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011010608465593018,
+      "loss": 2.0301,
+      "step": 1498
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011009270603082009,
+      "loss": 2.0934,
+      "step": 1499
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011007932740571002,
+      "loss": 1.9171,
+      "step": 1500
+    },
+    {
+      "epoch": 0.46,
+      "eval_bleu": 1.89534922643667e-13,
+      "eval_loss": 3.6431376934051514,
+      "eval_runtime": 2558.2889,
+      "eval_samples_per_second": 5.769,
+      "eval_steps_per_second": 0.721,
+      "step": 1500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011006594878059994,
+      "loss": 4.5347,
+      "step": 1501
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011005257015548986,
+      "loss": 4.4146,
+      "step": 1502
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001100391915303798,
+      "loss": 3.9244,
+      "step": 1503
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011002581290526971,
+      "loss": 3.8558,
+      "step": 1504
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011001243428015963,
+      "loss": 3.7018,
+      "step": 1505
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00010999905565504957,
+      "loss": 3.5729,
+      "step": 1506
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010998567702993947,
+      "loss": 3.2647,
+      "step": 1507
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010997229840482941,
+      "loss": 3.6812,
+      "step": 1508
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010995891977971933,
+      "loss": 3.3717,
+      "step": 1509
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010994554115460925,
+      "loss": 3.0918,
+      "step": 1510
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010993216252949918,
+      "loss": 3.3421,
+      "step": 1511
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001099187839043891,
+      "loss": 3.5035,
+      "step": 1512
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010990540527927903,
+      "loss": 3.2051,
+      "step": 1513
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010989202665416895,
+      "loss": 3.241,
+      "step": 1514
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010987864802905886,
+      "loss": 3.2923,
+      "step": 1515
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001098652694039488,
+      "loss": 3.257,
+      "step": 1516
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010985189077883871,
+      "loss": 3.2456,
+      "step": 1517
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010983851215372863,
+      "loss": 3.1612,
+      "step": 1518
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010982513352861857,
+      "loss": 3.2378,
+      "step": 1519
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010981175490350849,
+      "loss": 3.169,
+      "step": 1520
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010979837627839842,
+      "loss": 2.7696,
+      "step": 1521
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010978499765328833,
+      "loss": 2.8914,
+      "step": 1522
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010977161902817825,
+      "loss": 2.9091,
+      "step": 1523
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010975824040306818,
+      "loss": 2.9101,
+      "step": 1524
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001097448617779581,
+      "loss": 2.9641,
+      "step": 1525
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010973148315284804,
+      "loss": 3.0903,
+      "step": 1526
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010971810452773796,
+      "loss": 2.9581,
+      "step": 1527
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010970472590262788,
+      "loss": 2.9802,
+      "step": 1528
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010969134727751781,
+      "loss": 2.9185,
+      "step": 1529
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010967796865240772,
+      "loss": 2.67,
+      "step": 1530
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010966459002729764,
+      "loss": 2.6196,
+      "step": 1531
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010965121140218757,
+      "loss": 2.8138,
+      "step": 1532
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010963783277707749,
+      "loss": 2.7384,
+      "step": 1533
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010962445415196742,
+      "loss": 2.8416,
+      "step": 1534
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010961107552685734,
+      "loss": 2.6563,
+      "step": 1535
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010959769690174726,
+      "loss": 2.6816,
+      "step": 1536
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001095843182766372,
+      "loss": 2.5792,
+      "step": 1537
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001095709396515271,
+      "loss": 2.595,
+      "step": 1538
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010955756102641704,
+      "loss": 2.6428,
+      "step": 1539
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010954418240130696,
+      "loss": 2.3565,
+      "step": 1540
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010953080377619688,
+      "loss": 2.8248,
+      "step": 1541
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010951742515108681,
+      "loss": 2.5987,
+      "step": 1542
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010950404652597673,
+      "loss": 2.3853,
+      "step": 1543
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010949066790086665,
+      "loss": 2.2349,
+      "step": 1544
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010947728927575657,
+      "loss": 2.1381,
+      "step": 1545
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010946391065064649,
+      "loss": 2.2791,
+      "step": 1546
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010945053202553642,
+      "loss": 1.8002,
+      "step": 1547
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010943715340042634,
+      "loss": 1.9439,
+      "step": 1548
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010942377477531626,
+      "loss": 1.8611,
+      "step": 1549
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001094103961502062,
+      "loss": 1.8827,
+      "step": 1550
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010939701752509612,
+      "loss": 4.5243,
+      "step": 1551
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010938363889998604,
+      "loss": 4.268,
+      "step": 1552
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010937026027487596,
+      "loss": 3.7906,
+      "step": 1553
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010935688164976588,
+      "loss": 3.6094,
+      "step": 1554
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010934350302465581,
+      "loss": 3.3772,
+      "step": 1555
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010933012439954573,
+      "loss": 3.536,
+      "step": 1556
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010931674577443565,
+      "loss": 3.6182,
+      "step": 1557
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010930336714932558,
+      "loss": 3.5043,
+      "step": 1558
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001092899885242155,
+      "loss": 3.0472,
+      "step": 1559
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010927660989910544,
+      "loss": 3.3081,
+      "step": 1560
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010926323127399534,
+      "loss": 3.098,
+      "step": 1561
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010924985264888526,
+      "loss": 3.2902,
+      "step": 1562
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001092364740237752,
+      "loss": 2.8644,
+      "step": 1563
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010922309539866512,
+      "loss": 3.1793,
+      "step": 1564
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010920971677355504,
+      "loss": 3.2934,
+      "step": 1565
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010919633814844497,
+      "loss": 3.1172,
+      "step": 1566
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010918295952333489,
+      "loss": 3.2046,
+      "step": 1567
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010916958089822482,
+      "loss": 2.9789,
+      "step": 1568
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010915620227311473,
+      "loss": 3.2985,
+      "step": 1569
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010914282364800465,
+      "loss": 3.1946,
+      "step": 1570
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010912944502289458,
+      "loss": 2.8788,
+      "step": 1571
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001091160663977845,
+      "loss": 2.9118,
+      "step": 1572
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010910268777267444,
+      "loss": 3.0702,
+      "step": 1573
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010908930914756436,
+      "loss": 3.0204,
+      "step": 1574
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010907593052245428,
+      "loss": 2.999,
+      "step": 1575
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001090625518973442,
+      "loss": 2.9038,
+      "step": 1576
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010904917327223412,
+      "loss": 2.8708,
+      "step": 1577
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010903579464712404,
+      "loss": 2.6582,
+      "step": 1578
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010902241602201397,
+      "loss": 2.892,
+      "step": 1579
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010900903739690389,
+      "loss": 3.1004,
+      "step": 1580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010899565877179383,
+      "loss": 2.9463,
+      "step": 1581
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010898228014668375,
+      "loss": 2.7336,
+      "step": 1582
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010896890152157367,
+      "loss": 2.5711,
+      "step": 1583
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010895552289646359,
+      "loss": 2.5238,
+      "step": 1584
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001089421442713535,
+      "loss": 2.779,
+      "step": 1585
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010892876564624343,
+      "loss": 2.6695,
+      "step": 1586
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010891538702113336,
+      "loss": 2.4674,
+      "step": 1587
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010890200839602328,
+      "loss": 2.6305,
+      "step": 1588
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010888862977091321,
+      "loss": 2.5774,
+      "step": 1589
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010887525114580313,
+      "loss": 2.5741,
+      "step": 1590
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010886187252069305,
+      "loss": 2.3251,
+      "step": 1591
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010884849389558297,
+      "loss": 2.328,
+      "step": 1592
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010883511527047289,
+      "loss": 2.3744,
+      "step": 1593
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010882173664536283,
+      "loss": 1.9846,
+      "step": 1594
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010880835802025275,
+      "loss": 2.5091,
+      "step": 1595
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010879497939514267,
+      "loss": 1.9455,
+      "step": 1596
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001087816007700326,
+      "loss": 1.9997,
+      "step": 1597
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010876822214492252,
+      "loss": 2.0347,
+      "step": 1598
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010875484351981243,
+      "loss": 1.9016,
+      "step": 1599
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010874146489470236,
+      "loss": 1.9777,
+      "step": 1600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010872808626959228,
+      "loss": 4.6691,
+      "step": 1601
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010871470764448221,
+      "loss": 4.0281,
+      "step": 1602
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010870132901937213,
+      "loss": 3.7956,
+      "step": 1603
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010868795039426205,
+      "loss": 3.5917,
+      "step": 1604
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010867457176915199,
+      "loss": 3.5091,
+      "step": 1605
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001086611931440419,
+      "loss": 3.2889,
+      "step": 1606
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010864781451893183,
+      "loss": 3.369,
+      "step": 1607
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010863443589382175,
+      "loss": 3.165,
+      "step": 1608
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010862105726871167,
+      "loss": 3.0757,
+      "step": 1609
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001086076786436016,
+      "loss": 3.2941,
+      "step": 1610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010859430001849152,
+      "loss": 2.967,
+      "step": 1611
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010858092139338144,
+      "loss": 3.037,
+      "step": 1612
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010856754276827137,
+      "loss": 3.0505,
+      "step": 1613
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001085541641431613,
+      "loss": 3.0074,
+      "step": 1614
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010854078551805121,
+      "loss": 2.9648,
+      "step": 1615
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010852740689294113,
+      "loss": 2.9853,
+      "step": 1616
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010851402826783105,
+      "loss": 3.1198,
+      "step": 1617
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010850064964272099,
+      "loss": 2.8321,
+      "step": 1618
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010848727101761091,
+      "loss": 2.9486,
+      "step": 1619
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010847389239250084,
+      "loss": 3.0079,
+      "step": 1620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010846051376739076,
+      "loss": 2.963,
+      "step": 1621
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010844713514228067,
+      "loss": 2.7996,
+      "step": 1622
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001084337565171706,
+      "loss": 2.8355,
+      "step": 1623
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010842037789206052,
+      "loss": 3.1094,
+      "step": 1624
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010840699926695044,
+      "loss": 2.9885,
+      "step": 1625
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010839362064184037,
+      "loss": 2.9498,
+      "step": 1626
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001083802420167303,
+      "loss": 2.8089,
+      "step": 1627
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010836686339162023,
+      "loss": 3.0346,
+      "step": 1628
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010835348476651015,
+      "loss": 2.959,
+      "step": 1629
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010834010614140005,
+      "loss": 2.9599,
+      "step": 1630
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010832672751628999,
+      "loss": 2.7998,
+      "step": 1631
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010831334889117991,
+      "loss": 2.927,
+      "step": 1632
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010829997026606983,
+      "loss": 2.7667,
+      "step": 1633
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010828659164095976,
+      "loss": 2.6815,
+      "step": 1634
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010827321301584968,
+      "loss": 2.7541,
+      "step": 1635
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010825983439073962,
+      "loss": 2.6258,
+      "step": 1636
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010824645576562954,
+      "loss": 2.381,
+      "step": 1637
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010823307714051944,
+      "loss": 2.36,
+      "step": 1638
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010821969851540938,
+      "loss": 2.5608,
+      "step": 1639
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001082063198902993,
+      "loss": 2.4585,
+      "step": 1640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010819294126518923,
+      "loss": 2.5415,
+      "step": 1641
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010817956264007915,
+      "loss": 2.5564,
+      "step": 1642
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010816618401496907,
+      "loss": 2.3356,
+      "step": 1643
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000108152805389859,
+      "loss": 2.3302,
+      "step": 1644
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010813942676474892,
+      "loss": 2.1353,
+      "step": 1645
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010812604813963883,
+      "loss": 2.082,
+      "step": 1646
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010811266951452876,
+      "loss": 2.0618,
+      "step": 1647
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010809929088941868,
+      "loss": 1.8425,
+      "step": 1648
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010808591226430862,
+      "loss": 1.7506,
+      "step": 1649
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010807253363919854,
+      "loss": 1.925,
+      "step": 1650
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010805915501408846,
+      "loss": 4.3193,
+      "step": 1651
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010804577638897839,
+      "loss": 4.1279,
+      "step": 1652
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001080323977638683,
+      "loss": 3.777,
+      "step": 1653
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010801901913875823,
+      "loss": 3.5972,
+      "step": 1654
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010800564051364815,
+      "loss": 3.6019,
+      "step": 1655
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010799226188853807,
+      "loss": 3.3004,
+      "step": 1656
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000107978883263428,
+      "loss": 3.3957,
+      "step": 1657
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010796550463831792,
+      "loss": 3.1782,
+      "step": 1658
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010795212601320784,
+      "loss": 3.1619,
+      "step": 1659
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010793874738809778,
+      "loss": 2.9938,
+      "step": 1660
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010792536876298768,
+      "loss": 3.2,
+      "step": 1661
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010791199013787762,
+      "loss": 3.3073,
+      "step": 1662
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010789861151276754,
+      "loss": 3.03,
+      "step": 1663
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010788523288765746,
+      "loss": 3.0453,
+      "step": 1664
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010787185426254739,
+      "loss": 3.1175,
+      "step": 1665
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010785847563743731,
+      "loss": 2.9733,
+      "step": 1666
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010784509701232723,
+      "loss": 2.9321,
+      "step": 1667
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010783171838721716,
+      "loss": 2.6592,
+      "step": 1668
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010781833976210707,
+      "loss": 3.0763,
+      "step": 1669
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000107804961136997,
+      "loss": 2.9854,
+      "step": 1670
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010779158251188692,
+      "loss": 3.0332,
+      "step": 1671
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010777820388677684,
+      "loss": 2.9357,
+      "step": 1672
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010776482526166678,
+      "loss": 3.1358,
+      "step": 1673
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001077514466365567,
+      "loss": 2.9969,
+      "step": 1674
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010773806801144663,
+      "loss": 2.6537,
+      "step": 1675
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010772468938633654,
+      "loss": 2.896,
+      "step": 1676
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010771131076122646,
+      "loss": 2.8438,
+      "step": 1677
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010769793213611639,
+      "loss": 3.0306,
+      "step": 1678
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010768455351100631,
+      "loss": 2.6998,
+      "step": 1679
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010767117488589623,
+      "loss": 2.8651,
+      "step": 1680
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010765779626078616,
+      "loss": 3.0646,
+      "step": 1681
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010764441763567608,
+      "loss": 2.6915,
+      "step": 1682
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010763103901056602,
+      "loss": 2.6811,
+      "step": 1683
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010761766038545592,
+      "loss": 2.6065,
+      "step": 1684
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010760428176034584,
+      "loss": 2.433,
+      "step": 1685
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010759090313523578,
+      "loss": 2.4267,
+      "step": 1686
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001075775245101257,
+      "loss": 2.571,
+      "step": 1687
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010756414588501563,
+      "loss": 2.4633,
+      "step": 1688
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010755076725990555,
+      "loss": 2.4808,
+      "step": 1689
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010753738863479547,
+      "loss": 2.3858,
+      "step": 1690
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001075240100096854,
+      "loss": 2.1206,
+      "step": 1691
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010751063138457531,
+      "loss": 2.4384,
+      "step": 1692
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010749725275946523,
+      "loss": 2.4576,
+      "step": 1693
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010748387413435516,
+      "loss": 2.3868,
+      "step": 1694
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010747049550924508,
+      "loss": 2.1689,
+      "step": 1695
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010745711688413502,
+      "loss": 2.0849,
+      "step": 1696
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010744373825902494,
+      "loss": 1.8281,
+      "step": 1697
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010743035963391486,
+      "loss": 1.7294,
+      "step": 1698
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010741698100880479,
+      "loss": 1.6693,
+      "step": 1699
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001074036023836947,
+      "loss": 1.6206,
+      "step": 1700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010739022375858462,
+      "loss": 4.2878,
+      "step": 1701
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010737684513347455,
+      "loss": 3.9623,
+      "step": 1702
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010736346650836447,
+      "loss": 3.6515,
+      "step": 1703
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001073500878832544,
+      "loss": 3.48,
+      "step": 1704
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010733670925814433,
+      "loss": 3.1691,
+      "step": 1705
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010732333063303425,
+      "loss": 3.2898,
+      "step": 1706
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010730995200792417,
+      "loss": 3.221,
+      "step": 1707
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010729657338281409,
+      "loss": 3.0183,
+      "step": 1708
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010728319475770402,
+      "loss": 2.9669,
+      "step": 1709
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010726981613259394,
+      "loss": 3.1849,
+      "step": 1710
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010725643750748386,
+      "loss": 3.1746,
+      "step": 1711
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010724305888237379,
+      "loss": 2.9404,
+      "step": 1712
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010722968025726371,
+      "loss": 2.9474,
+      "step": 1713
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010721630163215363,
+      "loss": 2.868,
+      "step": 1714
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010720292300704355,
+      "loss": 2.9367,
+      "step": 1715
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010718954438193347,
+      "loss": 2.9912,
+      "step": 1716
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001071761657568234,
+      "loss": 2.9286,
+      "step": 1717
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010716278713171333,
+      "loss": 3.05,
+      "step": 1718
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010714940850660325,
+      "loss": 2.8008,
+      "step": 1719
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010713602988149318,
+      "loss": 2.9048,
+      "step": 1720
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001071226512563831,
+      "loss": 2.6858,
+      "step": 1721
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010710927263127303,
+      "loss": 2.7255,
+      "step": 1722
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010709589400616294,
+      "loss": 2.7559,
+      "step": 1723
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010708251538105286,
+      "loss": 2.6286,
+      "step": 1724
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001070691367559428,
+      "loss": 2.7647,
+      "step": 1725
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010705575813083271,
+      "loss": 2.7929,
+      "step": 1726
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010704237950572263,
+      "loss": 2.9627,
+      "step": 1727
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010702900088061257,
+      "loss": 2.4862,
+      "step": 1728
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010701562225550249,
+      "loss": 2.7699,
+      "step": 1729
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010700224363039241,
+      "loss": 2.8554,
+      "step": 1730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010698886500528233,
+      "loss": 2.6372,
+      "step": 1731
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010697548638017225,
+      "loss": 2.6469,
+      "step": 1732
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010696210775506218,
+      "loss": 2.6462,
+      "step": 1733
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001069487291299521,
+      "loss": 2.6529,
+      "step": 1734
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010693535050484203,
+      "loss": 2.3828,
+      "step": 1735
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010692197187973195,
+      "loss": 2.6061,
+      "step": 1736
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010690859325462187,
+      "loss": 2.4256,
+      "step": 1737
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001068952146295118,
+      "loss": 2.6431,
+      "step": 1738
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010688183600440171,
+      "loss": 2.3305,
+      "step": 1739
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010686845737929163,
+      "loss": 2.497,
+      "step": 1740
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010685507875418157,
+      "loss": 2.2571,
+      "step": 1741
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010684170012907149,
+      "loss": 2.5116,
+      "step": 1742
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010682832150396142,
+      "loss": 2.0883,
+      "step": 1743
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010681494287885134,
+      "loss": 2.1896,
+      "step": 1744
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010680156425374126,
+      "loss": 1.9815,
+      "step": 1745
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010678818562863118,
+      "loss": 2.1223,
+      "step": 1746
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001067748070035211,
+      "loss": 2.0007,
+      "step": 1747
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010676142837841102,
+      "loss": 1.9921,
+      "step": 1748
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010674804975330095,
+      "loss": 1.7413,
+      "step": 1749
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010673467112819087,
+      "loss": 1.6894,
+      "step": 1750
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010672129250308081,
+      "loss": 4.2376,
+      "step": 1751
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010670791387797073,
+      "loss": 3.5338,
+      "step": 1752
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010669453525286063,
+      "loss": 3.2912,
+      "step": 1753
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010668115662775057,
+      "loss": 2.8981,
+      "step": 1754
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010666777800264049,
+      "loss": 3.0308,
+      "step": 1755
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010665439937753042,
+      "loss": 3.23,
+      "step": 1756
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010664102075242034,
+      "loss": 3.314,
+      "step": 1757
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010662764212731026,
+      "loss": 3.0063,
+      "step": 1758
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001066142635022002,
+      "loss": 2.9404,
+      "step": 1759
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010660088487709012,
+      "loss": 2.9421,
+      "step": 1760
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010658750625198002,
+      "loss": 2.826,
+      "step": 1761
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010657412762686996,
+      "loss": 3.0438,
+      "step": 1762
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010656074900175988,
+      "loss": 2.7802,
+      "step": 1763
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010654737037664981,
+      "loss": 2.9517,
+      "step": 1764
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010653399175153973,
+      "loss": 2.8389,
+      "step": 1765
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010652061312642965,
+      "loss": 2.9294,
+      "step": 1766
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010650723450131958,
+      "loss": 2.8726,
+      "step": 1767
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001064938558762095,
+      "loss": 2.9689,
+      "step": 1768
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010648047725109942,
+      "loss": 2.9454,
+      "step": 1769
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010646709862598934,
+      "loss": 2.7571,
+      "step": 1770
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010645372000087926,
+      "loss": 2.7546,
+      "step": 1771
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001064403413757692,
+      "loss": 2.7533,
+      "step": 1772
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010642696275065912,
+      "loss": 2.6923,
+      "step": 1773
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010641358412554904,
+      "loss": 2.7145,
+      "step": 1774
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010640020550043897,
+      "loss": 2.6988,
+      "step": 1775
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010638682687532889,
+      "loss": 2.5393,
+      "step": 1776
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010637344825021881,
+      "loss": 2.4391,
+      "step": 1777
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010636006962510873,
+      "loss": 2.6317,
+      "step": 1778
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010634669099999865,
+      "loss": 2.6828,
+      "step": 1779
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010633331237488858,
+      "loss": 2.5477,
+      "step": 1780
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001063199337497785,
+      "loss": 2.4256,
+      "step": 1781
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010630655512466842,
+      "loss": 2.5379,
+      "step": 1782
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010629317649955836,
+      "loss": 2.4993,
+      "step": 1783
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010627979787444826,
+      "loss": 2.5532,
+      "step": 1784
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001062664192493382,
+      "loss": 2.1367,
+      "step": 1785
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010625304062422812,
+      "loss": 2.5448,
+      "step": 1786
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010623966199911804,
+      "loss": 2.4004,
+      "step": 1787
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010622628337400797,
+      "loss": 2.2416,
+      "step": 1788
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010621290474889789,
+      "loss": 2.3302,
+      "step": 1789
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010619952612378782,
+      "loss": 2.4364,
+      "step": 1790
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010618614749867774,
+      "loss": 1.7887,
+      "step": 1791
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010617276887356765,
+      "loss": 2.0656,
+      "step": 1792
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010615939024845758,
+      "loss": 2.4396,
+      "step": 1793
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001061460116233475,
+      "loss": 2.1626,
+      "step": 1794
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010613263299823742,
+      "loss": 2.0999,
+      "step": 1795
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010611925437312736,
+      "loss": 1.7575,
+      "step": 1796
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010610587574801728,
+      "loss": 1.8741,
+      "step": 1797
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010609249712290721,
+      "loss": 1.6764,
+      "step": 1798
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00010607911849779713,
+      "loss": 1.7414,
+      "step": 1799
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00010606573987268704,
+      "loss": 1.5391,
+      "step": 1800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00010605236124757697,
+      "loss": 4.3118,
+      "step": 1801
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00010603898262246689,
+      "loss": 3.6997,
+      "step": 1802
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00010602560399735682,
+      "loss": 3.6229,
+      "step": 1803
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00010601222537224674,
+      "loss": 3.1708,
+      "step": 1804
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00010599884674713666,
+      "loss": 3.143,
+      "step": 1805
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001059854681220266,
+      "loss": 2.7982,
+      "step": 1806
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001059720894969165,
+      "loss": 2.7496,
+      "step": 1807
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00010595871087180642,
+      "loss": 2.9021,
+      "step": 1808
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00010594533224669636,
+      "loss": 2.8597,
+      "step": 1809
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00010593195362158628,
+      "loss": 2.6996,
+      "step": 1810
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00010591857499647621,
+      "loss": 2.9677,
+      "step": 1811
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00010590519637136613,
+      "loss": 2.9241,
+      "step": 1812
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00010589181774625605,
+      "loss": 2.7109,
+      "step": 1813
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00010587843912114599,
+      "loss": 2.7139,
+      "step": 1814
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00010586506049603589,
+      "loss": 2.6286,
+      "step": 1815
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00010585168187092583,
+      "loss": 2.7662,
+      "step": 1816
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00010583830324581575,
+      "loss": 2.6871,
+      "step": 1817
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00010582492462070567,
+      "loss": 2.5846,
+      "step": 1818
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001058115459955956,
+      "loss": 2.6742,
+      "step": 1819
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00010579816737048552,
+      "loss": 2.5828,
+      "step": 1820
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00010578478874537544,
+      "loss": 2.5607,
+      "step": 1821
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00010577141012026537,
+      "loss": 2.7955,
+      "step": 1822
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00010575803149515528,
+      "loss": 2.7642,
+      "step": 1823
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00010574465287004521,
+      "loss": 2.7158,
+      "step": 1824
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00010573127424493513,
+      "loss": 2.3409,
+      "step": 1825
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00010571789561982505,
+      "loss": 2.6397,
+      "step": 1826
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00010570451699471499,
+      "loss": 2.6591,
+      "step": 1827
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001056911383696049,
+      "loss": 2.6056,
+      "step": 1828
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00010567775974449483,
+      "loss": 2.6905,
+      "step": 1829
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00010566438111938476,
+      "loss": 2.3994,
+      "step": 1830
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00010565100249427467,
+      "loss": 2.5425,
+      "step": 1831
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001056376238691646,
+      "loss": 2.4134,
+      "step": 1832
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00010562424524405452,
+      "loss": 2.5227,
+      "step": 1833
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00010561086661894444,
+      "loss": 2.1716,
+      "step": 1834
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00010559748799383437,
+      "loss": 2.1985,
+      "step": 1835
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00010558410936872429,
+      "loss": 2.2949,
+      "step": 1836
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00010557073074361423,
+      "loss": 2.1555,
+      "step": 1837
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00010555735211850413,
+      "loss": 2.4348,
+      "step": 1838
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00010554397349339405,
+      "loss": 2.4158,
+      "step": 1839
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00010553059486828399,
+      "loss": 2.3857,
+      "step": 1840
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001055172162431739,
+      "loss": 2.1138,
+      "step": 1841
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00010550383761806383,
+      "loss": 2.1253,
+      "step": 1842
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00010549045899295376,
+      "loss": 2.0852,
+      "step": 1843
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00010547708036784368,
+      "loss": 1.8534,
+      "step": 1844
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00010546370174273361,
+      "loss": 2.0488,
+      "step": 1845
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00010545032311762352,
+      "loss": 1.6837,
+      "step": 1846
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00010543694449251344,
+      "loss": 1.7963,
+      "step": 1847
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00010542356586740337,
+      "loss": 1.6488,
+      "step": 1848
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001054101872422933,
+      "loss": 1.6738,
+      "step": 1849
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00010539680861718323,
+      "loss": 1.4779,
+      "step": 1850
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00010538342999207315,
+      "loss": 3.8426,
+      "step": 1851
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00010537005136696307,
+      "loss": 3.454,
+      "step": 1852
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000105356672741853,
+      "loss": 3.2554,
+      "step": 1853
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00010534329411674291,
+      "loss": 3.1169,
+      "step": 1854
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00010532991549163283,
+      "loss": 3.1904,
+      "step": 1855
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00010531653686652276,
+      "loss": 2.8055,
+      "step": 1856
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00010530315824141268,
+      "loss": 3.135,
+      "step": 1857
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00010528977961630261,
+      "loss": 2.7327,
+      "step": 1858
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00010527640099119253,
+      "loss": 2.6952,
+      "step": 1859
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00010526302236608245,
+      "loss": 2.8462,
+      "step": 1860
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00010524964374097237,
+      "loss": 2.9298,
+      "step": 1861
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001052362651158623,
+      "loss": 3.0013,
+      "step": 1862
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00010522288649075221,
+      "loss": 2.7311,
+      "step": 1863
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00010520950786564215,
+      "loss": 2.9035,
+      "step": 1864
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00010519612924053207,
+      "loss": 2.7059,
+      "step": 1865
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000105182750615422,
+      "loss": 2.5842,
+      "step": 1866
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00010516937199031192,
+      "loss": 2.6636,
+      "step": 1867
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00010515599336520184,
+      "loss": 2.7264,
+      "step": 1868
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00010514261474009176,
+      "loss": 2.93,
+      "step": 1869
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00010512923611498168,
+      "loss": 2.5953,
+      "step": 1870
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00010511585748987162,
+      "loss": 2.6203,
+      "step": 1871
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00010510247886476154,
+      "loss": 2.5838,
+      "step": 1872
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00010508910023965146,
+      "loss": 2.3694,
+      "step": 1873
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00010507572161454139,
+      "loss": 2.4634,
+      "step": 1874
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00010506234298943131,
+      "loss": 2.5652,
+      "step": 1875
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00010504896436432123,
+      "loss": 2.6277,
+      "step": 1876
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00010503558573921115,
+      "loss": 2.8507,
+      "step": 1877
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00010502220711410107,
+      "loss": 2.6752,
+      "step": 1878
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000105008828488991,
+      "loss": 2.2766,
+      "step": 1879
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00010499544986388092,
+      "loss": 2.3409,
+      "step": 1880
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00010498207123877084,
+      "loss": 2.476,
+      "step": 1881
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00010496869261366078,
+      "loss": 2.3954,
+      "step": 1882
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001049553139885507,
+      "loss": 2.3083,
+      "step": 1883
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00010494193536344063,
+      "loss": 2.452,
+      "step": 1884
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00010492855673833054,
+      "loss": 2.41,
+      "step": 1885
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00010491517811322046,
+      "loss": 2.2978,
+      "step": 1886
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00010490179948811039,
+      "loss": 2.1935,
+      "step": 1887
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00010488842086300031,
+      "loss": 2.1917,
+      "step": 1888
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00010487504223789023,
+      "loss": 2.328,
+      "step": 1889
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00010486166361278016,
+      "loss": 2.1383,
+      "step": 1890
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00010484828498767008,
+      "loss": 1.9352,
+      "step": 1891
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00010483490636256,
+      "loss": 2.1248,
+      "step": 1892
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00010482152773744992,
+      "loss": 1.8638,
+      "step": 1893
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00010480814911233984,
+      "loss": 2.2072,
+      "step": 1894
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00010479477048722978,
+      "loss": 1.9768,
+      "step": 1895
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001047813918621197,
+      "loss": 2.0113,
+      "step": 1896
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00010476801323700962,
+      "loss": 1.9968,
+      "step": 1897
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00010475463461189955,
+      "loss": 1.5737,
+      "step": 1898
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00010474125598678947,
+      "loss": 1.6375,
+      "step": 1899
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00010472787736167939,
+      "loss": 1.6864,
+      "step": 1900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00010471449873656931,
+      "loss": 4.1854,
+      "step": 1901
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00010470112011145923,
+      "loss": 3.6107,
+      "step": 1902
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00010468774148634916,
+      "loss": 3.1919,
+      "step": 1903
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00010467436286123908,
+      "loss": 3.0811,
+      "step": 1904
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00010466098423612902,
+      "loss": 2.9468,
+      "step": 1905
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00010464760561101894,
+      "loss": 2.9361,
+      "step": 1906
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00010463422698590886,
+      "loss": 2.9579,
+      "step": 1907
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00010462084836079878,
+      "loss": 3.0893,
+      "step": 1908
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001046074697356887,
+      "loss": 3.0622,
+      "step": 1909
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00010459409111057862,
+      "loss": 2.5811,
+      "step": 1910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00010458071248546855,
+      "loss": 2.8711,
+      "step": 1911
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00010456733386035847,
+      "loss": 2.8706,
+      "step": 1912
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001045539552352484,
+      "loss": 2.6167,
+      "step": 1913
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00010454057661013832,
+      "loss": 2.7985,
+      "step": 1914
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00010452719798502823,
+      "loss": 2.6062,
+      "step": 1915
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00010451381935991816,
+      "loss": 2.743,
+      "step": 1916
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00010450044073480808,
+      "loss": 2.7547,
+      "step": 1917
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00010448706210969802,
+      "loss": 2.5901,
+      "step": 1918
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00010447368348458794,
+      "loss": 2.5076,
+      "step": 1919
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00010446030485947786,
+      "loss": 2.6919,
+      "step": 1920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00010444692623436779,
+      "loss": 2.5198,
+      "step": 1921
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00010443354760925771,
+      "loss": 2.5966,
+      "step": 1922
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00010442016898414762,
+      "loss": 2.5437,
+      "step": 1923
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00010440679035903755,
+      "loss": 2.4078,
+      "step": 1924
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00010439341173392747,
+      "loss": 2.8214,
+      "step": 1925
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001043800331088174,
+      "loss": 2.5112,
+      "step": 1926
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00010436665448370732,
+      "loss": 2.5176,
+      "step": 1927
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00010435327585859724,
+      "loss": 2.5835,
+      "step": 1928
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00010433989723348718,
+      "loss": 2.5275,
+      "step": 1929
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001043265186083771,
+      "loss": 2.2808,
+      "step": 1930
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00010431313998326702,
+      "loss": 2.4742,
+      "step": 1931
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00010429976135815694,
+      "loss": 2.728,
+      "step": 1932
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00010428638273304686,
+      "loss": 2.3206,
+      "step": 1933
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00010427300410793679,
+      "loss": 2.2381,
+      "step": 1934
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00010425962548282671,
+      "loss": 2.4442,
+      "step": 1935
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00010424624685771663,
+      "loss": 2.335,
+      "step": 1936
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00010423286823260657,
+      "loss": 2.3791,
+      "step": 1937
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00010421948960749649,
+      "loss": 2.2449,
+      "step": 1938
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001042061109823864,
+      "loss": 2.404,
+      "step": 1939
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00010419273235727633,
+      "loss": 2.0746,
+      "step": 1940
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00010417935373216625,
+      "loss": 1.9887,
+      "step": 1941
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00010416597510705618,
+      "loss": 2.0557,
+      "step": 1942
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001041525964819461,
+      "loss": 1.9702,
+      "step": 1943
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00010413921785683602,
+      "loss": 1.7484,
+      "step": 1944
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00010412583923172595,
+      "loss": 1.6216,
+      "step": 1945
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00010411246060661586,
+      "loss": 1.8094,
+      "step": 1946
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00010409908198150579,
+      "loss": 1.9698,
+      "step": 1947
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00010408570335639571,
+      "loss": 1.5298,
+      "step": 1948
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00010407232473128563,
+      "loss": 1.7083,
+      "step": 1949
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00010405894610617557,
+      "loss": 1.5954,
+      "step": 1950
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00010404556748106549,
+      "loss": 3.7447,
+      "step": 1951
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00010403218885595542,
+      "loss": 3.8408,
+      "step": 1952
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00010401881023084534,
+      "loss": 3.3518,
+      "step": 1953
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00010400543160573525,
+      "loss": 3.2843,
+      "step": 1954
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00010399205298062518,
+      "loss": 3.1871,
+      "step": 1955
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001039786743555151,
+      "loss": 2.9608,
+      "step": 1956
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00010396529573040502,
+      "loss": 2.7678,
+      "step": 1957
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00010395191710529495,
+      "loss": 2.7634,
+      "step": 1958
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00010393853848018487,
+      "loss": 2.7123,
+      "step": 1959
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00010392515985507481,
+      "loss": 2.5939,
+      "step": 1960
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00010391178122996473,
+      "loss": 3.0235,
+      "step": 1961
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00010389840260485463,
+      "loss": 3.0205,
+      "step": 1962
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00010388502397974457,
+      "loss": 2.8869,
+      "step": 1963
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00010387164535463449,
+      "loss": 2.6262,
+      "step": 1964
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00010385826672952442,
+      "loss": 2.3942,
+      "step": 1965
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00010384488810441434,
+      "loss": 2.765,
+      "step": 1966
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00010383150947930426,
+      "loss": 2.5648,
+      "step": 1967
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001038181308541942,
+      "loss": 2.5882,
+      "step": 1968
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001038047522290841,
+      "loss": 2.6412,
+      "step": 1969
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00010379137360397402,
+      "loss": 2.4716,
+      "step": 1970
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00010377799497886395,
+      "loss": 2.5187,
+      "step": 1971
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00010376461635375387,
+      "loss": 2.5128,
+      "step": 1972
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00010375123772864381,
+      "loss": 2.6028,
+      "step": 1973
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00010373785910353373,
+      "loss": 2.3915,
+      "step": 1974
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00010372448047842365,
+      "loss": 2.6369,
+      "step": 1975
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00010371110185331358,
+      "loss": 2.4212,
+      "step": 1976
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00010369772322820349,
+      "loss": 2.5919,
+      "step": 1977
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00010368434460309341,
+      "loss": 2.6602,
+      "step": 1978
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00010367096597798334,
+      "loss": 2.4194,
+      "step": 1979
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00010365758735287326,
+      "loss": 2.6813,
+      "step": 1980
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001036442087277632,
+      "loss": 2.5461,
+      "step": 1981
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00010363083010265311,
+      "loss": 2.3914,
+      "step": 1982
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00010361745147754303,
+      "loss": 2.2333,
+      "step": 1983
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00010360407285243297,
+      "loss": 2.1446,
+      "step": 1984
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00010359069422732287,
+      "loss": 2.1272,
+      "step": 1985
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00010357731560221281,
+      "loss": 2.3224,
+      "step": 1986
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00010356393697710273,
+      "loss": 2.1324,
+      "step": 1987
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00010355055835199265,
+      "loss": 2.2262,
+      "step": 1988
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00010353717972688258,
+      "loss": 2.2544,
+      "step": 1989
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001035238011017725,
+      "loss": 2.1042,
+      "step": 1990
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00010351042247666242,
+      "loss": 2.0247,
+      "step": 1991
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00010349704385155236,
+      "loss": 1.8581,
+      "step": 1992
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00010348366522644226,
+      "loss": 1.8519,
+      "step": 1993
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001034702866013322,
+      "loss": 1.8112,
+      "step": 1994
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00010345690797622212,
+      "loss": 1.7001,
+      "step": 1995
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00010344352935111204,
+      "loss": 1.9241,
+      "step": 1996
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00010343015072600197,
+      "loss": 1.6179,
+      "step": 1997
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00010341677210089189,
+      "loss": 1.7555,
+      "step": 1998
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00010340339347578182,
+      "loss": 1.4769,
+      "step": 1999
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00010339001485067173,
+      "loss": 1.4558,
+      "step": 2000
+    },
+    {
+      "epoch": 0.62,
+      "eval_bleu": 1.2907875456550126e-13,
+      "eval_loss": 2.82922101020813,
+      "eval_runtime": 2537.9369,
+      "eval_samples_per_second": 5.816,
+      "eval_steps_per_second": 0.727,
+      "step": 2000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00010337663622556165,
+      "loss": 3.891,
+      "step": 2001
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00010336325760045158,
+      "loss": 3.3693,
+      "step": 2002
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001033498789753415,
+      "loss": 3.2042,
+      "step": 2003
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00010333650035023142,
+      "loss": 2.5908,
+      "step": 2004
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00010332312172512136,
+      "loss": 2.5717,
+      "step": 2005
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00010330974310001128,
+      "loss": 2.9881,
+      "step": 2006
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00010329636447490121,
+      "loss": 2.6734,
+      "step": 2007
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00010328298584979112,
+      "loss": 2.6066,
+      "step": 2008
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00010326960722468104,
+      "loss": 2.5777,
+      "step": 2009
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00010325622859957097,
+      "loss": 2.756,
+      "step": 2010
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00010324284997446089,
+      "loss": 2.7246,
+      "step": 2011
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00010322947134935082,
+      "loss": 2.5681,
+      "step": 2012
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00010321609272424074,
+      "loss": 2.6843,
+      "step": 2013
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00010320271409913066,
+      "loss": 2.6264,
+      "step": 2014
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001031893354740206,
+      "loss": 2.7647,
+      "step": 2015
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001031759568489105,
+      "loss": 2.6176,
+      "step": 2016
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00010316257822380042,
+      "loss": 2.4696,
+      "step": 2017
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00010314919959869036,
+      "loss": 2.5587,
+      "step": 2018
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00010313582097358028,
+      "loss": 2.4573,
+      "step": 2019
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00010312244234847021,
+      "loss": 2.5395,
+      "step": 2020
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00010310906372336013,
+      "loss": 2.609,
+      "step": 2021
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00010309568509825005,
+      "loss": 2.5249,
+      "step": 2022
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00010308230647313997,
+      "loss": 2.6191,
+      "step": 2023
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00010306892784802989,
+      "loss": 2.3863,
+      "step": 2024
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00010305554922291981,
+      "loss": 2.4177,
+      "step": 2025
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00010304217059780974,
+      "loss": 2.294,
+      "step": 2026
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00010302879197269966,
+      "loss": 2.1289,
+      "step": 2027
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001030154133475896,
+      "loss": 2.5431,
+      "step": 2028
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00010300203472247952,
+      "loss": 2.3875,
+      "step": 2029
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00010298865609736944,
+      "loss": 2.2829,
+      "step": 2030
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00010297527747225936,
+      "loss": 2.2895,
+      "step": 2031
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00010296189884714928,
+      "loss": 2.2405,
+      "step": 2032
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00010294852022203921,
+      "loss": 2.3542,
+      "step": 2033
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00010293514159692913,
+      "loss": 2.3641,
+      "step": 2034
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00010292176297181905,
+      "loss": 2.3439,
+      "step": 2035
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00010290838434670898,
+      "loss": 2.2611,
+      "step": 2036
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001028950057215989,
+      "loss": 2.1924,
+      "step": 2037
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00010288162709648882,
+      "loss": 1.9669,
+      "step": 2038
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00010286824847137874,
+      "loss": 2.1722,
+      "step": 2039
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00010285486984626866,
+      "loss": 2.0229,
+      "step": 2040
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001028414912211586,
+      "loss": 2.0649,
+      "step": 2041
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00010282811259604852,
+      "loss": 2.0722,
+      "step": 2042
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00010281473397093844,
+      "loss": 2.116,
+      "step": 2043
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00010280135534582837,
+      "loss": 1.74,
+      "step": 2044
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00010278797672071829,
+      "loss": 1.5767,
+      "step": 2045
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00010277459809560821,
+      "loss": 1.5648,
+      "step": 2046
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00010276121947049813,
+      "loss": 1.8609,
+      "step": 2047
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00010274784084538805,
+      "loss": 1.7783,
+      "step": 2048
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00010273446222027799,
+      "loss": 1.5125,
+      "step": 2049
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001027210835951679,
+      "loss": 1.3682,
+      "step": 2050
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00010270770497005783,
+      "loss": 3.751,
+      "step": 2051
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00010269432634494776,
+      "loss": 3.1235,
+      "step": 2052
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00010268094771983768,
+      "loss": 3.1497,
+      "step": 2053
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001026675690947276,
+      "loss": 2.8441,
+      "step": 2054
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00010265419046961752,
+      "loss": 2.8023,
+      "step": 2055
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00010264081184450744,
+      "loss": 2.4402,
+      "step": 2056
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00010262743321939737,
+      "loss": 2.9208,
+      "step": 2057
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00010261405459428729,
+      "loss": 2.9814,
+      "step": 2058
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00010260067596917721,
+      "loss": 2.7028,
+      "step": 2059
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00010258729734406715,
+      "loss": 2.723,
+      "step": 2060
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00010257391871895707,
+      "loss": 2.5056,
+      "step": 2061
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00010256054009384699,
+      "loss": 2.5669,
+      "step": 2062
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001025471614687369,
+      "loss": 2.4831,
+      "step": 2063
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00010253378284362683,
+      "loss": 2.5927,
+      "step": 2064
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00010252040421851676,
+      "loss": 2.7756,
+      "step": 2065
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00010250702559340668,
+      "loss": 2.6431,
+      "step": 2066
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00010249364696829661,
+      "loss": 2.7551,
+      "step": 2067
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00010248026834318653,
+      "loss": 2.7799,
+      "step": 2068
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00010246688971807645,
+      "loss": 2.3062,
+      "step": 2069
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00010245351109296637,
+      "loss": 2.3821,
+      "step": 2070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00010244013246785629,
+      "loss": 2.6688,
+      "step": 2071
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00010242675384274621,
+      "loss": 2.358,
+      "step": 2072
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00010241337521763615,
+      "loss": 2.5297,
+      "step": 2073
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00010239999659252607,
+      "loss": 2.4494,
+      "step": 2074
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000102386617967416,
+      "loss": 2.6141,
+      "step": 2075
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00010237323934230592,
+      "loss": 2.5018,
+      "step": 2076
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00010235986071719583,
+      "loss": 2.4432,
+      "step": 2077
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00010234648209208576,
+      "loss": 2.1726,
+      "step": 2078
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00010233310346697568,
+      "loss": 2.4357,
+      "step": 2079
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00010231972484186561,
+      "loss": 2.3952,
+      "step": 2080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00010230634621675553,
+      "loss": 2.4936,
+      "step": 2081
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00010229296759164545,
+      "loss": 2.1858,
+      "step": 2082
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00010227958896653539,
+      "loss": 2.1995,
+      "step": 2083
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00010226621034142531,
+      "loss": 2.3326,
+      "step": 2084
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00010225283171631521,
+      "loss": 2.2028,
+      "step": 2085
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00010223945309120515,
+      "loss": 2.2521,
+      "step": 2086
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00010222607446609507,
+      "loss": 2.1389,
+      "step": 2087
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000102212695840985,
+      "loss": 2.0834,
+      "step": 2088
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00010219931721587492,
+      "loss": 2.1561,
+      "step": 2089
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00010218593859076484,
+      "loss": 2.3228,
+      "step": 2090
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00010217255996565477,
+      "loss": 2.1386,
+      "step": 2091
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001021591813405447,
+      "loss": 1.8684,
+      "step": 2092
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001021458027154346,
+      "loss": 1.8398,
+      "step": 2093
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00010213242409032453,
+      "loss": 2.0262,
+      "step": 2094
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00010211904546521445,
+      "loss": 1.7202,
+      "step": 2095
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00010210566684010439,
+      "loss": 1.4202,
+      "step": 2096
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00010209228821499431,
+      "loss": 1.6453,
+      "step": 2097
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00010207890958988423,
+      "loss": 1.67,
+      "step": 2098
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00010206553096477416,
+      "loss": 1.5715,
+      "step": 2099
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00010205215233966407,
+      "loss": 1.2944,
+      "step": 2100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000102038773714554,
+      "loss": 3.6327,
+      "step": 2101
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00010202539508944392,
+      "loss": 3.1216,
+      "step": 2102
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00010201201646433384,
+      "loss": 2.9191,
+      "step": 2103
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00010199863783922378,
+      "loss": 2.5907,
+      "step": 2104
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001019852592141137,
+      "loss": 2.9622,
+      "step": 2105
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00010197188058900362,
+      "loss": 2.751,
+      "step": 2106
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00010195850196389355,
+      "loss": 2.7818,
+      "step": 2107
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00010194512333878346,
+      "loss": 2.6743,
+      "step": 2108
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00010193174471367339,
+      "loss": 2.929,
+      "step": 2109
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00010191836608856331,
+      "loss": 2.4374,
+      "step": 2110
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00010190498746345323,
+      "loss": 2.4825,
+      "step": 2111
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00010189160883834316,
+      "loss": 2.7954,
+      "step": 2112
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00010187823021323308,
+      "loss": 2.6872,
+      "step": 2113
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00010186485158812302,
+      "loss": 2.3827,
+      "step": 2114
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00010185147296301294,
+      "loss": 2.6291,
+      "step": 2115
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00010183809433790284,
+      "loss": 2.5012,
+      "step": 2116
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00010182471571279278,
+      "loss": 2.5328,
+      "step": 2117
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001018113370876827,
+      "loss": 2.6024,
+      "step": 2118
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00010179795846257262,
+      "loss": 2.5711,
+      "step": 2119
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00010178457983746255,
+      "loss": 2.4639,
+      "step": 2120
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00010177120121235247,
+      "loss": 2.596,
+      "step": 2121
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001017578225872424,
+      "loss": 2.5169,
+      "step": 2122
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010174444396213232,
+      "loss": 2.4207,
+      "step": 2123
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010173106533702223,
+      "loss": 2.4543,
+      "step": 2124
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010171768671191216,
+      "loss": 2.2572,
+      "step": 2125
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010170430808680208,
+      "loss": 2.2952,
+      "step": 2126
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010169092946169202,
+      "loss": 2.375,
+      "step": 2127
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010167755083658194,
+      "loss": 2.2183,
+      "step": 2128
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010166417221147186,
+      "loss": 2.2133,
+      "step": 2129
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010165079358636179,
+      "loss": 2.2052,
+      "step": 2130
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001016374149612517,
+      "loss": 2.3729,
+      "step": 2131
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010162403633614162,
+      "loss": 2.3784,
+      "step": 2132
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010161065771103155,
+      "loss": 2.2761,
+      "step": 2133
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010159727908592147,
+      "loss": 2.1854,
+      "step": 2134
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001015839004608114,
+      "loss": 2.2206,
+      "step": 2135
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010157052183570132,
+      "loss": 2.345,
+      "step": 2136
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010155714321059124,
+      "loss": 2.2374,
+      "step": 2137
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010154376458548118,
+      "loss": 2.1533,
+      "step": 2138
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010153038596037108,
+      "loss": 1.9275,
+      "step": 2139
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000101517007335261,
+      "loss": 2.0406,
+      "step": 2140
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010150362871015094,
+      "loss": 1.8586,
+      "step": 2141
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010149025008504086,
+      "loss": 2.0109,
+      "step": 2142
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010147687145993079,
+      "loss": 1.9727,
+      "step": 2143
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010146349283482071,
+      "loss": 2.1012,
+      "step": 2144
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010145011420971063,
+      "loss": 1.709,
+      "step": 2145
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010143673558460056,
+      "loss": 1.5442,
+      "step": 2146
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010142335695949047,
+      "loss": 1.7982,
+      "step": 2147
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001014099783343804,
+      "loss": 1.6591,
+      "step": 2148
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010139659970927032,
+      "loss": 1.359,
+      "step": 2149
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010138322108416024,
+      "loss": 1.455,
+      "step": 2150
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010136984245905018,
+      "loss": 3.4178,
+      "step": 2151
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001013564638339401,
+      "loss": 3.0283,
+      "step": 2152
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010134308520883002,
+      "loss": 2.9702,
+      "step": 2153
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010132970658371994,
+      "loss": 2.9417,
+      "step": 2154
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010131632795860986,
+      "loss": 2.6616,
+      "step": 2155
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010130294933349979,
+      "loss": 2.6708,
+      "step": 2156
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010128957070838971,
+      "loss": 2.7093,
+      "step": 2157
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010127619208327963,
+      "loss": 2.7208,
+      "step": 2158
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010126281345816957,
+      "loss": 2.6101,
+      "step": 2159
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010124943483305949,
+      "loss": 2.6849,
+      "step": 2160
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010123605620794942,
+      "loss": 2.4614,
+      "step": 2161
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010122267758283933,
+      "loss": 2.7833,
+      "step": 2162
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010120929895772925,
+      "loss": 2.4865,
+      "step": 2163
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010119592033261918,
+      "loss": 2.7305,
+      "step": 2164
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001011825417075091,
+      "loss": 2.704,
+      "step": 2165
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010116916308239902,
+      "loss": 2.7281,
+      "step": 2166
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010115578445728895,
+      "loss": 2.3781,
+      "step": 2167
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010114240583217887,
+      "loss": 2.3637,
+      "step": 2168
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001011290272070688,
+      "loss": 2.2594,
+      "step": 2169
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010111564858195871,
+      "loss": 2.4512,
+      "step": 2170
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010110226995684863,
+      "loss": 2.3211,
+      "step": 2171
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010108889133173857,
+      "loss": 2.383,
+      "step": 2172
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010107551270662849,
+      "loss": 2.4681,
+      "step": 2173
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001010621340815184,
+      "loss": 2.4699,
+      "step": 2174
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010104875545640834,
+      "loss": 2.3426,
+      "step": 2175
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010103537683129826,
+      "loss": 2.2899,
+      "step": 2176
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010102199820618819,
+      "loss": 2.4156,
+      "step": 2177
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001010086195810781,
+      "loss": 2.1099,
+      "step": 2178
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010099524095596802,
+      "loss": 2.1519,
+      "step": 2179
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010098186233085795,
+      "loss": 2.2174,
+      "step": 2180
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010096848370574787,
+      "loss": 2.1133,
+      "step": 2181
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001009551050806378,
+      "loss": 1.9269,
+      "step": 2182
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010094172645552773,
+      "loss": 2.0325,
+      "step": 2183
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010092834783041765,
+      "loss": 2.1158,
+      "step": 2184
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010091496920530757,
+      "loss": 2.177,
+      "step": 2185
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010090159058019749,
+      "loss": 2.0908,
+      "step": 2186
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001008882119550874,
+      "loss": 2.2268,
+      "step": 2187
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010087483332997734,
+      "loss": 2.2975,
+      "step": 2188
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010086145470486726,
+      "loss": 2.0453,
+      "step": 2189
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001008480760797572,
+      "loss": 2.0367,
+      "step": 2190
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010083469745464711,
+      "loss": 2.0139,
+      "step": 2191
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010082131882953703,
+      "loss": 1.8318,
+      "step": 2192
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010080794020442695,
+      "loss": 1.6144,
+      "step": 2193
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010079456157931687,
+      "loss": 1.7887,
+      "step": 2194
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010078118295420681,
+      "loss": 1.532,
+      "step": 2195
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010076780432909673,
+      "loss": 1.6085,
+      "step": 2196
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010075442570398665,
+      "loss": 1.3803,
+      "step": 2197
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010074104707887658,
+      "loss": 1.6648,
+      "step": 2198
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001007276684537665,
+      "loss": 1.4615,
+      "step": 2199
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010071428982865642,
+      "loss": 1.5222,
+      "step": 2200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010070091120354634,
+      "loss": 3.5915,
+      "step": 2201
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010068753257843626,
+      "loss": 3.5458,
+      "step": 2202
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001006741539533262,
+      "loss": 3.1087,
+      "step": 2203
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010066077532821611,
+      "loss": 2.9701,
+      "step": 2204
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010064739670310603,
+      "loss": 2.7023,
+      "step": 2205
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010063401807799597,
+      "loss": 2.652,
+      "step": 2206
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010062063945288589,
+      "loss": 2.6195,
+      "step": 2207
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001006072608277758,
+      "loss": 2.6375,
+      "step": 2208
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010059388220266573,
+      "loss": 2.5742,
+      "step": 2209
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010058050357755565,
+      "loss": 2.5186,
+      "step": 2210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010056712495244558,
+      "loss": 2.5584,
+      "step": 2211
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001005537463273355,
+      "loss": 2.5267,
+      "step": 2212
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010054036770222542,
+      "loss": 2.525,
+      "step": 2213
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010052698907711535,
+      "loss": 2.5993,
+      "step": 2214
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010051361045200527,
+      "loss": 2.3043,
+      "step": 2215
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001005002318268952,
+      "loss": 2.4013,
+      "step": 2216
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010048685320178511,
+      "loss": 2.5812,
+      "step": 2217
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010047347457667503,
+      "loss": 2.2653,
+      "step": 2218
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010046009595156497,
+      "loss": 2.2498,
+      "step": 2219
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010044671732645489,
+      "loss": 2.1778,
+      "step": 2220
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010043333870134481,
+      "loss": 2.3255,
+      "step": 2221
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010041996007623474,
+      "loss": 2.5121,
+      "step": 2222
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010040658145112466,
+      "loss": 2.6466,
+      "step": 2223
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010039320282601458,
+      "loss": 2.1872,
+      "step": 2224
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001003798242009045,
+      "loss": 2.3792,
+      "step": 2225
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010036644557579442,
+      "loss": 2.2001,
+      "step": 2226
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010035306695068436,
+      "loss": 2.2607,
+      "step": 2227
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010033968832557428,
+      "loss": 2.2754,
+      "step": 2228
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010032630970046421,
+      "loss": 2.3376,
+      "step": 2229
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010031293107535413,
+      "loss": 2.209,
+      "step": 2230
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010029955245024404,
+      "loss": 2.2147,
+      "step": 2231
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010028617382513397,
+      "loss": 2.3754,
+      "step": 2232
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010027279520002389,
+      "loss": 2.2619,
+      "step": 2233
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010025941657491381,
+      "loss": 2.156,
+      "step": 2234
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010024603794980374,
+      "loss": 2.0371,
+      "step": 2235
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010023265932469366,
+      "loss": 2.2698,
+      "step": 2236
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001002192806995836,
+      "loss": 2.1152,
+      "step": 2237
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010020590207447352,
+      "loss": 1.9088,
+      "step": 2238
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010019252344936342,
+      "loss": 1.8773,
+      "step": 2239
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010017914482425336,
+      "loss": 1.9229,
+      "step": 2240
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010016576619914328,
+      "loss": 2.1418,
+      "step": 2241
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010015238757403321,
+      "loss": 1.8994,
+      "step": 2242
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010013900894892313,
+      "loss": 1.8811,
+      "step": 2243
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010012563032381305,
+      "loss": 1.7781,
+      "step": 2244
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010011225169870298,
+      "loss": 1.6895,
+      "step": 2245
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001000988730735929,
+      "loss": 1.7987,
+      "step": 2246
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010008549444848281,
+      "loss": 1.6244,
+      "step": 2247
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010007211582337274,
+      "loss": 1.5277,
+      "step": 2248
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010005873719826266,
+      "loss": 1.473,
+      "step": 2249
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001000453585731526,
+      "loss": 1.6268,
+      "step": 2250
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010003197994804252,
+      "loss": 3.5541,
+      "step": 2251
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010001860132293244,
+      "loss": 3.24,
+      "step": 2252
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010000522269782237,
+      "loss": 2.842,
+      "step": 2253
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.999184407271229e-05,
+      "loss": 2.9304,
+      "step": 2254
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.99784654476022e-05,
+      "loss": 2.7251,
+      "step": 2255
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.996508682249213e-05,
+      "loss": 2.4443,
+      "step": 2256
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.995170819738205e-05,
+      "loss": 2.5852,
+      "step": 2257
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.993832957227198e-05,
+      "loss": 2.812,
+      "step": 2258
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.99249509471619e-05,
+      "loss": 2.4624,
+      "step": 2259
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.991157232205182e-05,
+      "loss": 2.6091,
+      "step": 2260
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.989819369694176e-05,
+      "loss": 2.3945,
+      "step": 2261
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.988481507183166e-05,
+      "loss": 2.4552,
+      "step": 2262
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.98714364467216e-05,
+      "loss": 2.4695,
+      "step": 2263
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.985805782161152e-05,
+      "loss": 2.4973,
+      "step": 2264
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.984467919650144e-05,
+      "loss": 2.3563,
+      "step": 2265
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.983130057139137e-05,
+      "loss": 2.3818,
+      "step": 2266
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.981792194628129e-05,
+      "loss": 2.6779,
+      "step": 2267
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.980454332117121e-05,
+      "loss": 2.4747,
+      "step": 2268
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.979116469606114e-05,
+      "loss": 2.3282,
+      "step": 2269
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.977778607095105e-05,
+      "loss": 2.6941,
+      "step": 2270
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.976440744584098e-05,
+      "loss": 2.3279,
+      "step": 2271
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.97510288207309e-05,
+      "loss": 2.1966,
+      "step": 2272
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.973765019562082e-05,
+      "loss": 2.0546,
+      "step": 2273
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.972427157051076e-05,
+      "loss": 2.1381,
+      "step": 2274
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.971089294540068e-05,
+      "loss": 2.318,
+      "step": 2275
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.969751432029061e-05,
+      "loss": 2.7001,
+      "step": 2276
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.968413569518053e-05,
+      "loss": 2.2746,
+      "step": 2277
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.967075707007044e-05,
+      "loss": 2.256,
+      "step": 2278
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.965737844496037e-05,
+      "loss": 2.2661,
+      "step": 2279
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.964399981985029e-05,
+      "loss": 2.1927,
+      "step": 2280
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.963062119474021e-05,
+      "loss": 2.1476,
+      "step": 2281
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.961724256963015e-05,
+      "loss": 2.3239,
+      "step": 2282
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.960386394452007e-05,
+      "loss": 1.9703,
+      "step": 2283
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.959048531941e-05,
+      "loss": 1.9244,
+      "step": 2284
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.95771066942999e-05,
+      "loss": 2.2505,
+      "step": 2285
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.956372806918983e-05,
+      "loss": 1.936,
+      "step": 2286
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.955034944407976e-05,
+      "loss": 2.1998,
+      "step": 2287
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.953697081896968e-05,
+      "loss": 1.9247,
+      "step": 2288
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.95235921938596e-05,
+      "loss": 1.7647,
+      "step": 2289
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.951021356874953e-05,
+      "loss": 1.7236,
+      "step": 2290
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.949683494363945e-05,
+      "loss": 2.0901,
+      "step": 2291
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.948345631852939e-05,
+      "loss": 1.7958,
+      "step": 2292
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.947007769341929e-05,
+      "loss": 1.9553,
+      "step": 2293
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.945669906830921e-05,
+      "loss": 1.7829,
+      "step": 2294
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.944332044319915e-05,
+      "loss": 1.6376,
+      "step": 2295
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.942994181808907e-05,
+      "loss": 1.6009,
+      "step": 2296
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.9416563192979e-05,
+      "loss": 1.7466,
+      "step": 2297
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.940318456786892e-05,
+      "loss": 1.5189,
+      "step": 2298
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.938980594275884e-05,
+      "loss": 1.4495,
+      "step": 2299
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.937642731764877e-05,
+      "loss": 1.402,
+      "step": 2300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.936304869253868e-05,
+      "loss": 3.4968,
+      "step": 2301
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.93496700674286e-05,
+      "loss": 3.1025,
+      "step": 2302
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.933629144231853e-05,
+      "loss": 2.6056,
+      "step": 2303
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.932291281720845e-05,
+      "loss": 2.5255,
+      "step": 2304
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.930953419209839e-05,
+      "loss": 2.5097,
+      "step": 2305
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.92961555669883e-05,
+      "loss": 2.4855,
+      "step": 2306
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.928277694187823e-05,
+      "loss": 2.8248,
+      "step": 2307
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.926939831676816e-05,
+      "loss": 2.5392,
+      "step": 2308
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.925601969165807e-05,
+      "loss": 2.647,
+      "step": 2309
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.9242641066548e-05,
+      "loss": 2.5401,
+      "step": 2310
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.922926244143792e-05,
+      "loss": 2.5173,
+      "step": 2311
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.921588381632784e-05,
+      "loss": 2.3819,
+      "step": 2312
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.920250519121777e-05,
+      "loss": 2.3113,
+      "step": 2313
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.91891265661077e-05,
+      "loss": 2.5532,
+      "step": 2314
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.917574794099761e-05,
+      "loss": 2.5214,
+      "step": 2315
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.916236931588753e-05,
+      "loss": 2.5522,
+      "step": 2316
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.914899069077745e-05,
+      "loss": 2.2531,
+      "step": 2317
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.913561206566739e-05,
+      "loss": 2.3105,
+      "step": 2318
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.912223344055731e-05,
+      "loss": 2.2662,
+      "step": 2319
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.910885481544723e-05,
+      "loss": 2.1534,
+      "step": 2320
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.909547619033716e-05,
+      "loss": 2.4527,
+      "step": 2321
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.908209756522708e-05,
+      "loss": 2.2645,
+      "step": 2322
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.906871894011701e-05,
+      "loss": 2.3674,
+      "step": 2323
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.905534031500692e-05,
+      "loss": 2.2773,
+      "step": 2324
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.904196168989684e-05,
+      "loss": 2.3099,
+      "step": 2325
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.902858306478677e-05,
+      "loss": 2.218,
+      "step": 2326
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.90152044396767e-05,
+      "loss": 2.1286,
+      "step": 2327
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.900182581456661e-05,
+      "loss": 2.2406,
+      "step": 2328
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.898844718945655e-05,
+      "loss": 2.1438,
+      "step": 2329
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.897506856434647e-05,
+      "loss": 2.2202,
+      "step": 2330
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.89616899392364e-05,
+      "loss": 2.1429,
+      "step": 2331
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.894831131412631e-05,
+      "loss": 2.3079,
+      "step": 2332
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.893493268901623e-05,
+      "loss": 2.2481,
+      "step": 2333
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.892155406390616e-05,
+      "loss": 2.0351,
+      "step": 2334
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.890817543879608e-05,
+      "loss": 2.2121,
+      "step": 2335
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.8894796813686e-05,
+      "loss": 1.9573,
+      "step": 2336
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.888141818857594e-05,
+      "loss": 2.217,
+      "step": 2337
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.886803956346586e-05,
+      "loss": 1.8554,
+      "step": 2338
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.885466093835578e-05,
+      "loss": 1.9912,
+      "step": 2339
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.88412823132457e-05,
+      "loss": 2.1184,
+      "step": 2340
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.882790368813562e-05,
+      "loss": 1.8813,
+      "step": 2341
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.881452506302555e-05,
+      "loss": 1.833,
+      "step": 2342
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.880114643791547e-05,
+      "loss": 1.9308,
+      "step": 2343
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.87877678128054e-05,
+      "loss": 1.9449,
+      "step": 2344
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.877438918769532e-05,
+      "loss": 1.5713,
+      "step": 2345
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.876101056258524e-05,
+      "loss": 1.6625,
+      "step": 2346
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.874763193747516e-05,
+      "loss": 1.5823,
+      "step": 2347
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.873425331236508e-05,
+      "loss": 1.5738,
+      "step": 2348
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.8720874687255e-05,
+      "loss": 1.669,
+      "step": 2349
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.870749606214494e-05,
+      "loss": 1.435,
+      "step": 2350
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.869411743703486e-05,
+      "loss": 3.2427,
+      "step": 2351
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.868073881192479e-05,
+      "loss": 3.0441,
+      "step": 2352
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.866736018681471e-05,
+      "loss": 2.6863,
+      "step": 2353
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.865398156170463e-05,
+      "loss": 2.6372,
+      "step": 2354
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.864060293659455e-05,
+      "loss": 2.6608,
+      "step": 2355
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.862722431148447e-05,
+      "loss": 2.4628,
+      "step": 2356
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.86138456863744e-05,
+      "loss": 2.5322,
+      "step": 2357
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.860046706126432e-05,
+      "loss": 2.4899,
+      "step": 2358
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.858708843615424e-05,
+      "loss": 2.5278,
+      "step": 2359
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.857370981104418e-05,
+      "loss": 2.4028,
+      "step": 2360
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.85603311859341e-05,
+      "loss": 2.3476,
+      "step": 2361
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.8546952560824e-05,
+      "loss": 2.3038,
+      "step": 2362
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.853357393571394e-05,
+      "loss": 2.1547,
+      "step": 2363
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.852019531060386e-05,
+      "loss": 2.4111,
+      "step": 2364
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.850681668549379e-05,
+      "loss": 2.1829,
+      "step": 2365
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.849343806038371e-05,
+      "loss": 2.5159,
+      "step": 2366
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.848005943527363e-05,
+      "loss": 2.3943,
+      "step": 2367
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.846668081016356e-05,
+      "loss": 2.3347,
+      "step": 2368
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.845330218505348e-05,
+      "loss": 2.325,
+      "step": 2369
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.843992355994339e-05,
+      "loss": 2.102,
+      "step": 2370
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.842654493483332e-05,
+      "loss": 2.3212,
+      "step": 2371
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.841316630972324e-05,
+      "loss": 2.2266,
+      "step": 2372
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.839978768461318e-05,
+      "loss": 2.2925,
+      "step": 2373
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.83864090595031e-05,
+      "loss": 2.2424,
+      "step": 2374
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.837303043439302e-05,
+      "loss": 2.1674,
+      "step": 2375
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.835965180928295e-05,
+      "loss": 2.2801,
+      "step": 2376
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.834627318417287e-05,
+      "loss": 2.2104,
+      "step": 2377
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.833289455906279e-05,
+      "loss": 2.2101,
+      "step": 2378
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.831951593395271e-05,
+      "loss": 2.1545,
+      "step": 2379
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.830613730884263e-05,
+      "loss": 2.0586,
+      "step": 2380
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.829275868373256e-05,
+      "loss": 2.1532,
+      "step": 2381
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.827938005862248e-05,
+      "loss": 2.1041,
+      "step": 2382
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.82660014335124e-05,
+      "loss": 2.0338,
+      "step": 2383
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.825262280840234e-05,
+      "loss": 2.2194,
+      "step": 2384
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.823924418329226e-05,
+      "loss": 1.9995,
+      "step": 2385
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.822586555818218e-05,
+      "loss": 2.0621,
+      "step": 2386
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.82124869330721e-05,
+      "loss": 2.0197,
+      "step": 2387
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.819910830796202e-05,
+      "loss": 2.1523,
+      "step": 2388
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.818572968285195e-05,
+      "loss": 1.9023,
+      "step": 2389
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.817235105774187e-05,
+      "loss": 1.9754,
+      "step": 2390
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.81589724326318e-05,
+      "loss": 1.9954,
+      "step": 2391
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.814559380752173e-05,
+      "loss": 1.8349,
+      "step": 2392
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.813221518241163e-05,
+      "loss": 1.919,
+      "step": 2393
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.811883655730157e-05,
+      "loss": 1.6005,
+      "step": 2394
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.810545793219149e-05,
+      "loss": 1.774,
+      "step": 2395
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.80920793070814e-05,
+      "loss": 1.4871,
+      "step": 2396
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.807870068197134e-05,
+      "loss": 1.4411,
+      "step": 2397
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.806532205686126e-05,
+      "loss": 1.6126,
+      "step": 2398
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.805194343175119e-05,
+      "loss": 1.4731,
+      "step": 2399
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.803856480664111e-05,
+      "loss": 1.3396,
+      "step": 2400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.802518618153102e-05,
+      "loss": 3.2735,
+      "step": 2401
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.801180755642095e-05,
+      "loss": 3.0872,
+      "step": 2402
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.799842893131087e-05,
+      "loss": 2.4877,
+      "step": 2403
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.798505030620079e-05,
+      "loss": 2.6551,
+      "step": 2404
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.797167168109073e-05,
+      "loss": 2.6455,
+      "step": 2405
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.795829305598065e-05,
+      "loss": 2.4763,
+      "step": 2406
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.794491443087058e-05,
+      "loss": 2.8287,
+      "step": 2407
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.79315358057605e-05,
+      "loss": 2.2274,
+      "step": 2408
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.79181571806504e-05,
+      "loss": 2.2497,
+      "step": 2409
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.790477855554034e-05,
+      "loss": 2.3337,
+      "step": 2410
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.789139993043026e-05,
+      "loss": 2.6104,
+      "step": 2411
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.787802130532019e-05,
+      "loss": 2.5376,
+      "step": 2412
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.786464268021011e-05,
+      "loss": 2.251,
+      "step": 2413
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.785126405510003e-05,
+      "loss": 2.6062,
+      "step": 2414
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.783788542998997e-05,
+      "loss": 2.3116,
+      "step": 2415
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.782450680487987e-05,
+      "loss": 2.4813,
+      "step": 2416
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.781112817976979e-05,
+      "loss": 2.4336,
+      "step": 2417
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.779774955465973e-05,
+      "loss": 2.3701,
+      "step": 2418
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.778437092954965e-05,
+      "loss": 2.1913,
+      "step": 2419
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.777099230443958e-05,
+      "loss": 2.1685,
+      "step": 2420
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.77576136793295e-05,
+      "loss": 2.1528,
+      "step": 2421
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.774423505421942e-05,
+      "loss": 2.0536,
+      "step": 2422
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.773085642910935e-05,
+      "loss": 2.4047,
+      "step": 2423
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.771747780399926e-05,
+      "loss": 2.302,
+      "step": 2424
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.77040991788892e-05,
+      "loss": 2.132,
+      "step": 2425
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.769072055377911e-05,
+      "loss": 2.0695,
+      "step": 2426
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.767734192866903e-05,
+      "loss": 2.251,
+      "step": 2427
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.766396330355897e-05,
+      "loss": 2.1663,
+      "step": 2428
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.765058467844889e-05,
+      "loss": 2.1467,
+      "step": 2429
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.763720605333881e-05,
+      "loss": 2.1372,
+      "step": 2430
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.762382742822874e-05,
+      "loss": 2.1418,
+      "step": 2431
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.761044880311865e-05,
+      "loss": 2.1671,
+      "step": 2432
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.759707017800858e-05,
+      "loss": 2.0593,
+      "step": 2433
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.75836915528985e-05,
+      "loss": 2.093,
+      "step": 2434
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.757031292778842e-05,
+      "loss": 2.1004,
+      "step": 2435
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.755693430267835e-05,
+      "loss": 2.1181,
+      "step": 2436
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.754355567756827e-05,
+      "loss": 1.9509,
+      "step": 2437
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.753017705245821e-05,
+      "loss": 1.8942,
+      "step": 2438
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.751679842734813e-05,
+      "loss": 1.9656,
+      "step": 2439
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.750341980223803e-05,
+      "loss": 1.7857,
+      "step": 2440
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.749004117712797e-05,
+      "loss": 1.9946,
+      "step": 2441
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.747666255201789e-05,
+      "loss": 1.7434,
+      "step": 2442
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.746328392690781e-05,
+      "loss": 2.1309,
+      "step": 2443
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.744990530179774e-05,
+      "loss": 1.6321,
+      "step": 2444
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.743652667668766e-05,
+      "loss": 1.8773,
+      "step": 2445
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.74231480515776e-05,
+      "loss": 1.5488,
+      "step": 2446
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.74097694264675e-05,
+      "loss": 1.3895,
+      "step": 2447
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.739639080135742e-05,
+      "loss": 1.3516,
+      "step": 2448
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.738301217624735e-05,
+      "loss": 1.4001,
+      "step": 2449
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.736963355113727e-05,
+      "loss": 1.4137,
+      "step": 2450
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.73562549260272e-05,
+      "loss": 3.24,
+      "step": 2451
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.734287630091713e-05,
+      "loss": 2.629,
+      "step": 2452
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.732949767580705e-05,
+      "loss": 2.6996,
+      "step": 2453
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.731611905069698e-05,
+      "loss": 2.4609,
+      "step": 2454
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.730274042558689e-05,
+      "loss": 2.5747,
+      "step": 2455
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.728936180047681e-05,
+      "loss": 2.3715,
+      "step": 2456
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.727598317536674e-05,
+      "loss": 2.4116,
+      "step": 2457
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.726260455025666e-05,
+      "loss": 2.6951,
+      "step": 2458
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.72492259251466e-05,
+      "loss": 2.3431,
+      "step": 2459
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.723584730003652e-05,
+      "loss": 2.4649,
+      "step": 2460
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.722246867492644e-05,
+      "loss": 2.5786,
+      "step": 2461
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.720909004981637e-05,
+      "loss": 2.4794,
+      "step": 2462
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.719571142470628e-05,
+      "loss": 2.1477,
+      "step": 2463
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.71823327995962e-05,
+      "loss": 2.2436,
+      "step": 2464
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.716895417448613e-05,
+      "loss": 2.326,
+      "step": 2465
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.715557554937605e-05,
+      "loss": 2.2337,
+      "step": 2466
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.714219692426598e-05,
+      "loss": 2.419,
+      "step": 2467
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.71288182991559e-05,
+      "loss": 2.3539,
+      "step": 2468
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.711543967404582e-05,
+      "loss": 2.2358,
+      "step": 2469
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.710206104893574e-05,
+      "loss": 2.3301,
+      "step": 2470
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.708868242382566e-05,
+      "loss": 2.4224,
+      "step": 2471
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.70753037987156e-05,
+      "loss": 2.078,
+      "step": 2472
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.706192517360552e-05,
+      "loss": 2.2324,
+      "step": 2473
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.704854654849544e-05,
+      "loss": 2.4353,
+      "step": 2474
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.703516792338537e-05,
+      "loss": 2.0575,
+      "step": 2475
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.702178929827529e-05,
+      "loss": 1.9893,
+      "step": 2476
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.700841067316521e-05,
+      "loss": 1.9766,
+      "step": 2477
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.699503204805513e-05,
+      "loss": 2.196,
+      "step": 2478
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.698165342294505e-05,
+      "loss": 2.1136,
+      "step": 2479
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.696827479783498e-05,
+      "loss": 1.9809,
+      "step": 2480
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.69548961727249e-05,
+      "loss": 1.9094,
+      "step": 2481
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.694151754761482e-05,
+      "loss": 1.8501,
+      "step": 2482
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.692813892250476e-05,
+      "loss": 2.1919,
+      "step": 2483
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.691476029739468e-05,
+      "loss": 2.1218,
+      "step": 2484
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.69013816722846e-05,
+      "loss": 1.8957,
+      "step": 2485
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.688800304717452e-05,
+      "loss": 2.0336,
+      "step": 2486
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.687462442206444e-05,
+      "loss": 1.9839,
+      "step": 2487
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.686124579695437e-05,
+      "loss": 1.9029,
+      "step": 2488
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.684786717184429e-05,
+      "loss": 1.6585,
+      "step": 2489
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.683448854673421e-05,
+      "loss": 1.9972,
+      "step": 2490
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.682110992162414e-05,
+      "loss": 1.7496,
+      "step": 2491
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.680773129651406e-05,
+      "loss": 1.7373,
+      "step": 2492
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.6794352671404e-05,
+      "loss": 1.5797,
+      "step": 2493
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.67809740462939e-05,
+      "loss": 1.7027,
+      "step": 2494
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.676759542118382e-05,
+      "loss": 1.8927,
+      "step": 2495
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.675421679607376e-05,
+      "loss": 1.4484,
+      "step": 2496
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.674083817096368e-05,
+      "loss": 1.6537,
+      "step": 2497
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.67274595458536e-05,
+      "loss": 1.3751,
+      "step": 2498
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.671408092074353e-05,
+      "loss": 1.3657,
+      "step": 2499
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.670070229563345e-05,
+      "loss": 1.2393,
+      "step": 2500
+    },
+    {
+      "epoch": 0.77,
+      "eval_bleu": 1.9167368854093544e-15,
+      "eval_loss": 2.370410919189453,
+      "eval_runtime": 2549.7966,
+      "eval_samples_per_second": 5.789,
+      "eval_steps_per_second": 0.724,
+      "step": 2500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.668732367052337e-05,
+      "loss": 3.3711,
+      "step": 2501
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.667394504541329e-05,
+      "loss": 2.9199,
+      "step": 2502
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.666056642030321e-05,
+      "loss": 2.5418,
+      "step": 2503
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.664718779519314e-05,
+      "loss": 2.5845,
+      "step": 2504
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.663380917008306e-05,
+      "loss": 2.5312,
+      "step": 2505
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.6620430544973e-05,
+      "loss": 2.4165,
+      "step": 2506
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.660705191986292e-05,
+      "loss": 2.3324,
+      "step": 2507
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.659367329475284e-05,
+      "loss": 2.5352,
+      "step": 2508
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.658029466964276e-05,
+      "loss": 2.2513,
+      "step": 2509
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.656691604453268e-05,
+      "loss": 2.5897,
+      "step": 2510
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.65535374194226e-05,
+      "loss": 2.1013,
+      "step": 2511
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.654015879431253e-05,
+      "loss": 2.2351,
+      "step": 2512
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.652678016920245e-05,
+      "loss": 2.1784,
+      "step": 2513
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.651340154409239e-05,
+      "loss": 2.2033,
+      "step": 2514
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.65000229189823e-05,
+      "loss": 2.2299,
+      "step": 2515
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.648664429387223e-05,
+      "loss": 2.212,
+      "step": 2516
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.647326566876215e-05,
+      "loss": 2.4564,
+      "step": 2517
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.645988704365207e-05,
+      "loss": 2.0704,
+      "step": 2518
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.6446508418542e-05,
+      "loss": 2.298,
+      "step": 2519
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.643312979343192e-05,
+      "loss": 2.0293,
+      "step": 2520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.641975116832184e-05,
+      "loss": 2.0532,
+      "step": 2521
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.640637254321177e-05,
+      "loss": 2.2569,
+      "step": 2522
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.639299391810169e-05,
+      "loss": 2.1402,
+      "step": 2523
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.63796152929916e-05,
+      "loss": 2.197,
+      "step": 2524
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.636623666788153e-05,
+      "loss": 2.3433,
+      "step": 2525
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.635285804277145e-05,
+      "loss": 2.1407,
+      "step": 2526
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.633947941766139e-05,
+      "loss": 2.2567,
+      "step": 2527
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.63261007925513e-05,
+      "loss": 2.3939,
+      "step": 2528
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.631272216744123e-05,
+      "loss": 2.2684,
+      "step": 2529
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.629934354233116e-05,
+      "loss": 2.0504,
+      "step": 2530
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.628596491722108e-05,
+      "loss": 2.0188,
+      "step": 2531
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.627258629211099e-05,
+      "loss": 2.1898,
+      "step": 2532
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.625920766700092e-05,
+      "loss": 2.348,
+      "step": 2533
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.624582904189084e-05,
+      "loss": 1.9325,
+      "step": 2534
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.623245041678077e-05,
+      "loss": 2.0181,
+      "step": 2535
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.62190717916707e-05,
+      "loss": 2.1485,
+      "step": 2536
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.620569316656061e-05,
+      "loss": 1.7821,
+      "step": 2537
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.619231454145055e-05,
+      "loss": 1.8564,
+      "step": 2538
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.617893591634047e-05,
+      "loss": 1.9865,
+      "step": 2539
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.616555729123039e-05,
+      "loss": 1.8682,
+      "step": 2540
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.615217866612031e-05,
+      "loss": 1.6597,
+      "step": 2541
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.613880004101023e-05,
+      "loss": 1.809,
+      "step": 2542
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.612542141590016e-05,
+      "loss": 1.6485,
+      "step": 2543
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.611204279079008e-05,
+      "loss": 1.5579,
+      "step": 2544
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.609866416568e-05,
+      "loss": 1.3392,
+      "step": 2545
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.608528554056993e-05,
+      "loss": 1.5768,
+      "step": 2546
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.607190691545985e-05,
+      "loss": 1.3758,
+      "step": 2547
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.605852829034977e-05,
+      "loss": 1.3995,
+      "step": 2548
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.60451496652397e-05,
+      "loss": 1.1176,
+      "step": 2549
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.603177104012961e-05,
+      "loss": 1.2177,
+      "step": 2550
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.601839241501955e-05,
+      "loss": 3.1064,
+      "step": 2551
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.600501378990947e-05,
+      "loss": 2.9763,
+      "step": 2552
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.59916351647994e-05,
+      "loss": 2.7521,
+      "step": 2553
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.597825653968932e-05,
+      "loss": 2.6558,
+      "step": 2554
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.596487791457923e-05,
+      "loss": 2.5223,
+      "step": 2555
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.595149928946916e-05,
+      "loss": 2.5061,
+      "step": 2556
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.593812066435908e-05,
+      "loss": 2.5114,
+      "step": 2557
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.5924742039249e-05,
+      "loss": 2.3016,
+      "step": 2558
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.591136341413893e-05,
+      "loss": 2.6785,
+      "step": 2559
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.589798478902885e-05,
+      "loss": 2.4042,
+      "step": 2560
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.588460616391879e-05,
+      "loss": 2.365,
+      "step": 2561
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.587122753880871e-05,
+      "loss": 2.4077,
+      "step": 2562
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.585784891369861e-05,
+      "loss": 2.185,
+      "step": 2563
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.584447028858855e-05,
+      "loss": 2.3173,
+      "step": 2564
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.583109166347847e-05,
+      "loss": 2.3422,
+      "step": 2565
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.581771303836839e-05,
+      "loss": 2.5247,
+      "step": 2566
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.580433441325832e-05,
+      "loss": 2.3354,
+      "step": 2567
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.579095578814824e-05,
+      "loss": 2.3667,
+      "step": 2568
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.577757716303818e-05,
+      "loss": 2.1518,
+      "step": 2569
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.57641985379281e-05,
+      "loss": 2.3876,
+      "step": 2570
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.5750819912818e-05,
+      "loss": 2.2429,
+      "step": 2571
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.573744128770794e-05,
+      "loss": 2.2869,
+      "step": 2572
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.572406266259786e-05,
+      "loss": 2.3506,
+      "step": 2573
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.571068403748779e-05,
+      "loss": 2.1955,
+      "step": 2574
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.569730541237771e-05,
+      "loss": 2.0731,
+      "step": 2575
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.568392678726763e-05,
+      "loss": 1.9777,
+      "step": 2576
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.567054816215756e-05,
+      "loss": 2.0192,
+      "step": 2577
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.565716953704747e-05,
+      "loss": 1.9019,
+      "step": 2578
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.564379091193739e-05,
+      "loss": 2.0251,
+      "step": 2579
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.563041228682732e-05,
+      "loss": 2.1906,
+      "step": 2580
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.561703366171724e-05,
+      "loss": 2.2793,
+      "step": 2581
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.560365503660718e-05,
+      "loss": 2.1206,
+      "step": 2582
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.55902764114971e-05,
+      "loss": 1.9065,
+      "step": 2583
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.557689778638702e-05,
+      "loss": 1.9886,
+      "step": 2584
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.556351916127695e-05,
+      "loss": 2.0286,
+      "step": 2585
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.555014053616686e-05,
+      "loss": 1.845,
+      "step": 2586
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.553676191105679e-05,
+      "loss": 1.8186,
+      "step": 2587
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.552338328594671e-05,
+      "loss": 1.9245,
+      "step": 2588
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.551000466083663e-05,
+      "loss": 1.8248,
+      "step": 2589
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.549662603572656e-05,
+      "loss": 1.7724,
+      "step": 2590
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.548324741061648e-05,
+      "loss": 1.7075,
+      "step": 2591
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.54698687855064e-05,
+      "loss": 2.0717,
+      "step": 2592
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.545649016039634e-05,
+      "loss": 1.6569,
+      "step": 2593
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.544311153528624e-05,
+      "loss": 1.7146,
+      "step": 2594
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.542973291017618e-05,
+      "loss": 1.5763,
+      "step": 2595
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.54163542850661e-05,
+      "loss": 1.4948,
+      "step": 2596
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.540297565995602e-05,
+      "loss": 1.3791,
+      "step": 2597
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.538959703484595e-05,
+      "loss": 1.5308,
+      "step": 2598
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.537621840973587e-05,
+      "loss": 1.1926,
+      "step": 2599
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.536283978462579e-05,
+      "loss": 1.339,
+      "step": 2600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.534946115951572e-05,
+      "loss": 3.1273,
+      "step": 2601
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.533608253440563e-05,
+      "loss": 2.6912,
+      "step": 2602
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.532270390929556e-05,
+      "loss": 2.5249,
+      "step": 2603
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.530932528418548e-05,
+      "loss": 2.5317,
+      "step": 2604
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.52959466590754e-05,
+      "loss": 2.4175,
+      "step": 2605
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.528256803396534e-05,
+      "loss": 2.3026,
+      "step": 2606
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.526918940885526e-05,
+      "loss": 2.4507,
+      "step": 2607
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.525581078374519e-05,
+      "loss": 2.3932,
+      "step": 2608
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.52424321586351e-05,
+      "loss": 2.477,
+      "step": 2609
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.522905353352502e-05,
+      "loss": 2.3539,
+      "step": 2610
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.521567490841495e-05,
+      "loss": 2.2332,
+      "step": 2611
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.520229628330487e-05,
+      "loss": 2.3863,
+      "step": 2612
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.518891765819479e-05,
+      "loss": 2.3624,
+      "step": 2613
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.517553903308472e-05,
+      "loss": 2.2031,
+      "step": 2614
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.516216040797464e-05,
+      "loss": 2.1093,
+      "step": 2615
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.514878178286458e-05,
+      "loss": 2.2936,
+      "step": 2616
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.513540315775448e-05,
+      "loss": 2.1475,
+      "step": 2617
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.51220245326444e-05,
+      "loss": 1.9967,
+      "step": 2618
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.510864590753434e-05,
+      "loss": 2.2041,
+      "step": 2619
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.509526728242426e-05,
+      "loss": 2.1556,
+      "step": 2620
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.508188865731419e-05,
+      "loss": 2.241,
+      "step": 2621
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.506851003220411e-05,
+      "loss": 2.1939,
+      "step": 2622
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.505513140709403e-05,
+      "loss": 2.3144,
+      "step": 2623
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.504175278198397e-05,
+      "loss": 1.9301,
+      "step": 2624
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.502837415687387e-05,
+      "loss": 1.9496,
+      "step": 2625
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.501499553176379e-05,
+      "loss": 2.3036,
+      "step": 2626
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.500161690665373e-05,
+      "loss": 2.0395,
+      "step": 2627
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.498823828154365e-05,
+      "loss": 2.0255,
+      "step": 2628
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.497485965643358e-05,
+      "loss": 1.8616,
+      "step": 2629
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.49614810313235e-05,
+      "loss": 1.9472,
+      "step": 2630
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.494810240621342e-05,
+      "loss": 2.0251,
+      "step": 2631
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.493472378110334e-05,
+      "loss": 2.1157,
+      "step": 2632
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.492134515599326e-05,
+      "loss": 2.0769,
+      "step": 2633
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.490796653088319e-05,
+      "loss": 1.8908,
+      "step": 2634
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.489458790577311e-05,
+      "loss": 1.8591,
+      "step": 2635
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.488120928066303e-05,
+      "loss": 1.9502,
+      "step": 2636
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.486783065555297e-05,
+      "loss": 1.7102,
+      "step": 2637
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.485445203044289e-05,
+      "loss": 1.8732,
+      "step": 2638
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.48410734053328e-05,
+      "loss": 1.8493,
+      "step": 2639
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.482769478022273e-05,
+      "loss": 1.9648,
+      "step": 2640
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.481431615511265e-05,
+      "loss": 1.7764,
+      "step": 2641
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.480093753000258e-05,
+      "loss": 1.7692,
+      "step": 2642
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.47875589048925e-05,
+      "loss": 1.803,
+      "step": 2643
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.477418027978242e-05,
+      "loss": 1.5559,
+      "step": 2644
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.476080165467235e-05,
+      "loss": 1.4966,
+      "step": 2645
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.474742302956227e-05,
+      "loss": 1.4351,
+      "step": 2646
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.473404440445219e-05,
+      "loss": 1.7413,
+      "step": 2647
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.472066577934211e-05,
+      "loss": 1.4002,
+      "step": 2648
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.470728715423203e-05,
+      "loss": 1.0536,
+      "step": 2649
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.469390852912197e-05,
+      "loss": 1.2574,
+      "step": 2650
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.468052990401189e-05,
+      "loss": 3.0361,
+      "step": 2651
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.46671512789018e-05,
+      "loss": 2.6678,
+      "step": 2652
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.465377265379174e-05,
+      "loss": 2.4621,
+      "step": 2653
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.464039402868166e-05,
+      "loss": 2.4119,
+      "step": 2654
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.46270154035716e-05,
+      "loss": 2.3802,
+      "step": 2655
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.46136367784615e-05,
+      "loss": 2.3279,
+      "step": 2656
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.460025815335142e-05,
+      "loss": 2.3087,
+      "step": 2657
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.458687952824135e-05,
+      "loss": 2.4305,
+      "step": 2658
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.457350090313127e-05,
+      "loss": 2.4724,
+      "step": 2659
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.45601222780212e-05,
+      "loss": 2.5346,
+      "step": 2660
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.454674365291113e-05,
+      "loss": 2.398,
+      "step": 2661
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.453336502780105e-05,
+      "loss": 2.2904,
+      "step": 2662
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.451998640269097e-05,
+      "loss": 2.1139,
+      "step": 2663
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.450660777758089e-05,
+      "loss": 2.1992,
+      "step": 2664
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.449322915247081e-05,
+      "loss": 2.3049,
+      "step": 2665
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.447985052736074e-05,
+      "loss": 2.185,
+      "step": 2666
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.446647190225066e-05,
+      "loss": 2.5899,
+      "step": 2667
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.44530932771406e-05,
+      "loss": 2.0983,
+      "step": 2668
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.443971465203051e-05,
+      "loss": 2.0947,
+      "step": 2669
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.442633602692043e-05,
+      "loss": 2.2436,
+      "step": 2670
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.441295740181035e-05,
+      "loss": 1.9785,
+      "step": 2671
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.439957877670027e-05,
+      "loss": 2.1451,
+      "step": 2672
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.43862001515902e-05,
+      "loss": 2.2217,
+      "step": 2673
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.437282152648013e-05,
+      "loss": 2.1114,
+      "step": 2674
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.435944290137005e-05,
+      "loss": 2.009,
+      "step": 2675
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.434606427625998e-05,
+      "loss": 2.2394,
+      "step": 2676
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.43326856511499e-05,
+      "loss": 2.209,
+      "step": 2677
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.431930702603982e-05,
+      "loss": 1.9819,
+      "step": 2678
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.430592840092974e-05,
+      "loss": 1.8707,
+      "step": 2679
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.429254977581966e-05,
+      "loss": 1.9536,
+      "step": 2680
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.427917115070958e-05,
+      "loss": 2.0242,
+      "step": 2681
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.426579252559952e-05,
+      "loss": 1.9379,
+      "step": 2682
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.425241390048944e-05,
+      "loss": 1.9147,
+      "step": 2683
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.423903527537937e-05,
+      "loss": 1.9583,
+      "step": 2684
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.422565665026929e-05,
+      "loss": 1.7119,
+      "step": 2685
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.42122780251592e-05,
+      "loss": 2.0564,
+      "step": 2686
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.419889940004913e-05,
+      "loss": 2.0259,
+      "step": 2687
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.418552077493905e-05,
+      "loss": 1.8462,
+      "step": 2688
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.417214214982898e-05,
+      "loss": 1.8706,
+      "step": 2689
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.41587635247189e-05,
+      "loss": 1.9105,
+      "step": 2690
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.414538489960882e-05,
+      "loss": 1.6353,
+      "step": 2691
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.413200627449876e-05,
+      "loss": 1.6546,
+      "step": 2692
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.411862764938868e-05,
+      "loss": 1.7602,
+      "step": 2693
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.410524902427858e-05,
+      "loss": 1.8897,
+      "step": 2694
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.409187039916852e-05,
+      "loss": 1.3021,
+      "step": 2695
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.407849177405844e-05,
+      "loss": 1.4498,
+      "step": 2696
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.406511314894837e-05,
+      "loss": 1.3699,
+      "step": 2697
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.405173452383829e-05,
+      "loss": 1.2618,
+      "step": 2698
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.403835589872821e-05,
+      "loss": 1.4112,
+      "step": 2699
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.402497727361814e-05,
+      "loss": 1.5256,
+      "step": 2700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.401159864850806e-05,
+      "loss": 3.0629,
+      "step": 2701
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.399822002339798e-05,
+      "loss": 2.8399,
+      "step": 2702
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.39848413982879e-05,
+      "loss": 2.5116,
+      "step": 2703
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.397146277317782e-05,
+      "loss": 2.2257,
+      "step": 2704
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.395808414806776e-05,
+      "loss": 2.4394,
+      "step": 2705
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.394470552295768e-05,
+      "loss": 2.3912,
+      "step": 2706
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.39313268978476e-05,
+      "loss": 2.3322,
+      "step": 2707
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.391794827273753e-05,
+      "loss": 2.2611,
+      "step": 2708
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.390456964762744e-05,
+      "loss": 2.2616,
+      "step": 2709
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.389119102251737e-05,
+      "loss": 2.3186,
+      "step": 2710
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.387781239740729e-05,
+      "loss": 2.5568,
+      "step": 2711
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.386443377229721e-05,
+      "loss": 2.2381,
+      "step": 2712
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.385105514718714e-05,
+      "loss": 2.3743,
+      "step": 2713
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.383767652207706e-05,
+      "loss": 2.5084,
+      "step": 2714
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.3824297896967e-05,
+      "loss": 2.2575,
+      "step": 2715
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.381091927185692e-05,
+      "loss": 2.1456,
+      "step": 2716
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.379754064674682e-05,
+      "loss": 2.1502,
+      "step": 2717
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.378416202163676e-05,
+      "loss": 2.2227,
+      "step": 2718
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.377078339652668e-05,
+      "loss": 2.1263,
+      "step": 2719
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.37574047714166e-05,
+      "loss": 2.1342,
+      "step": 2720
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.374402614630653e-05,
+      "loss": 2.1598,
+      "step": 2721
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.373064752119645e-05,
+      "loss": 1.9899,
+      "step": 2722
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.371726889608638e-05,
+      "loss": 2.2867,
+      "step": 2723
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.37038902709763e-05,
+      "loss": 2.3309,
+      "step": 2724
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.369051164586621e-05,
+      "loss": 1.9753,
+      "step": 2725
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.367713302075614e-05,
+      "loss": 2.1222,
+      "step": 2726
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.366375439564606e-05,
+      "loss": 1.8591,
+      "step": 2727
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.365037577053598e-05,
+      "loss": 2.0728,
+      "step": 2728
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.363699714542592e-05,
+      "loss": 1.9494,
+      "step": 2729
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.362361852031584e-05,
+      "loss": 2.0685,
+      "step": 2730
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.361023989520577e-05,
+      "loss": 2.0003,
+      "step": 2731
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.359686127009569e-05,
+      "loss": 1.6434,
+      "step": 2732
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.35834826449856e-05,
+      "loss": 1.885,
+      "step": 2733
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.357010401987553e-05,
+      "loss": 1.8194,
+      "step": 2734
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.355672539476545e-05,
+      "loss": 1.7787,
+      "step": 2735
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.354334676965538e-05,
+      "loss": 1.7339,
+      "step": 2736
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.35299681445453e-05,
+      "loss": 1.7723,
+      "step": 2737
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.351658951943522e-05,
+      "loss": 1.7503,
+      "step": 2738
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.350321089432516e-05,
+      "loss": 1.6969,
+      "step": 2739
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.348983226921506e-05,
+      "loss": 1.7606,
+      "step": 2740
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.347645364410498e-05,
+      "loss": 1.836,
+      "step": 2741
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.346307501899492e-05,
+      "loss": 1.6664,
+      "step": 2742
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.344969639388484e-05,
+      "loss": 1.6148,
+      "step": 2743
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.343631776877477e-05,
+      "loss": 1.6589,
+      "step": 2744
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.342293914366469e-05,
+      "loss": 1.3222,
+      "step": 2745
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.340956051855461e-05,
+      "loss": 1.3923,
+      "step": 2746
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.339618189344455e-05,
+      "loss": 1.5243,
+      "step": 2747
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.338280326833445e-05,
+      "loss": 1.2581,
+      "step": 2748
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.336942464322439e-05,
+      "loss": 1.2581,
+      "step": 2749
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.33560460181143e-05,
+      "loss": 1.2197,
+      "step": 2750
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.334266739300423e-05,
+      "loss": 3.3854,
+      "step": 2751
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.332928876789416e-05,
+      "loss": 2.6228,
+      "step": 2752
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.331591014278408e-05,
+      "loss": 2.6136,
+      "step": 2753
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.3302531517674e-05,
+      "loss": 2.3999,
+      "step": 2754
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.328915289256393e-05,
+      "loss": 2.0262,
+      "step": 2755
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.327577426745384e-05,
+      "loss": 2.4526,
+      "step": 2756
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.326239564234377e-05,
+      "loss": 2.2624,
+      "step": 2757
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.324901701723369e-05,
+      "loss": 2.3995,
+      "step": 2758
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.323563839212361e-05,
+      "loss": 2.275,
+      "step": 2759
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.322225976701355e-05,
+      "loss": 2.1136,
+      "step": 2760
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.320888114190347e-05,
+      "loss": 2.127,
+      "step": 2761
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.319550251679339e-05,
+      "loss": 2.1204,
+      "step": 2762
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.31821238916833e-05,
+      "loss": 2.4167,
+      "step": 2763
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.316874526657323e-05,
+      "loss": 2.111,
+      "step": 2764
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.315536664146316e-05,
+      "loss": 2.2573,
+      "step": 2765
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.314198801635308e-05,
+      "loss": 2.0865,
+      "step": 2766
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.3128609391243e-05,
+      "loss": 2.3279,
+      "step": 2767
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.311523076613293e-05,
+      "loss": 2.2153,
+      "step": 2768
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.310185214102285e-05,
+      "loss": 2.0513,
+      "step": 2769
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.308847351591279e-05,
+      "loss": 2.186,
+      "step": 2770
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.30750948908027e-05,
+      "loss": 2.164,
+      "step": 2771
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.306171626569261e-05,
+      "loss": 2.1787,
+      "step": 2772
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.304833764058255e-05,
+      "loss": 2.0801,
+      "step": 2773
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.303495901547247e-05,
+      "loss": 2.0746,
+      "step": 2774
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.302158039036239e-05,
+      "loss": 1.9242,
+      "step": 2775
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.300820176525232e-05,
+      "loss": 2.1465,
+      "step": 2776
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.299482314014224e-05,
+      "loss": 1.9602,
+      "step": 2777
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.298144451503217e-05,
+      "loss": 1.9702,
+      "step": 2778
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.296806588992208e-05,
+      "loss": 1.8038,
+      "step": 2779
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.2954687264812e-05,
+      "loss": 2.03,
+      "step": 2780
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.294130863970193e-05,
+      "loss": 2.1515,
+      "step": 2781
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.292793001459185e-05,
+      "loss": 2.0113,
+      "step": 2782
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.291455138948179e-05,
+      "loss": 2.009,
+      "step": 2783
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.290117276437171e-05,
+      "loss": 1.6897,
+      "step": 2784
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.288779413926163e-05,
+      "loss": 1.9397,
+      "step": 2785
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.287441551415156e-05,
+      "loss": 1.8338,
+      "step": 2786
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.286103688904147e-05,
+      "loss": 1.7565,
+      "step": 2787
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.284765826393139e-05,
+      "loss": 1.6405,
+      "step": 2788
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.283427963882132e-05,
+      "loss": 1.8513,
+      "step": 2789
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.282090101371124e-05,
+      "loss": 1.9468,
+      "step": 2790
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.280752238860117e-05,
+      "loss": 1.7869,
+      "step": 2791
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.27941437634911e-05,
+      "loss": 1.6661,
+      "step": 2792
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.278076513838101e-05,
+      "loss": 1.4998,
+      "step": 2793
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.276738651327093e-05,
+      "loss": 1.5981,
+      "step": 2794
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.275400788816085e-05,
+      "loss": 1.4995,
+      "step": 2795
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.274062926305077e-05,
+      "loss": 1.4727,
+      "step": 2796
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.272725063794071e-05,
+      "loss": 1.3251,
+      "step": 2797
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.271387201283063e-05,
+      "loss": 1.3289,
+      "step": 2798
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.270049338772056e-05,
+      "loss": 1.236,
+      "step": 2799
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.268711476261048e-05,
+      "loss": 1.2464,
+      "step": 2800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.26737361375004e-05,
+      "loss": 2.8418,
+      "step": 2801
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.266035751239032e-05,
+      "loss": 2.5714,
+      "step": 2802
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.264697888728024e-05,
+      "loss": 2.3758,
+      "step": 2803
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.263360026217018e-05,
+      "loss": 2.275,
+      "step": 2804
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.26202216370601e-05,
+      "loss": 2.1588,
+      "step": 2805
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.260684301195002e-05,
+      "loss": 2.2644,
+      "step": 2806
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.259346438683995e-05,
+      "loss": 2.5308,
+      "step": 2807
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.258008576172987e-05,
+      "loss": 2.3242,
+      "step": 2808
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.256670713661979e-05,
+      "loss": 2.3522,
+      "step": 2809
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.255332851150971e-05,
+      "loss": 2.1922,
+      "step": 2810
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.253994988639963e-05,
+      "loss": 2.1181,
+      "step": 2811
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.252657126128956e-05,
+      "loss": 2.2906,
+      "step": 2812
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.251319263617948e-05,
+      "loss": 2.0022,
+      "step": 2813
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.24998140110694e-05,
+      "loss": 1.9936,
+      "step": 2814
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.248643538595934e-05,
+      "loss": 2.2499,
+      "step": 2815
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.247305676084926e-05,
+      "loss": 2.1879,
+      "step": 2816
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.245967813573918e-05,
+      "loss": 2.2076,
+      "step": 2817
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.24462995106291e-05,
+      "loss": 2.3427,
+      "step": 2818
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.243292088551902e-05,
+      "loss": 2.0679,
+      "step": 2819
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.241954226040895e-05,
+      "loss": 2.3778,
+      "step": 2820
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.240616363529887e-05,
+      "loss": 2.0078,
+      "step": 2821
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.239278501018879e-05,
+      "loss": 1.9981,
+      "step": 2822
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.237940638507872e-05,
+      "loss": 2.1903,
+      "step": 2823
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.236602775996864e-05,
+      "loss": 1.9468,
+      "step": 2824
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.235264913485856e-05,
+      "loss": 1.9319,
+      "step": 2825
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.233927050974848e-05,
+      "loss": 2.1314,
+      "step": 2826
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.23258918846384e-05,
+      "loss": 1.9645,
+      "step": 2827
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.231251325952834e-05,
+      "loss": 2.1398,
+      "step": 2828
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.229913463441826e-05,
+      "loss": 2.179,
+      "step": 2829
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.228575600930819e-05,
+      "loss": 2.2451,
+      "step": 2830
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.227237738419811e-05,
+      "loss": 1.9601,
+      "step": 2831
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.225899875908803e-05,
+      "loss": 1.7915,
+      "step": 2832
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.224562013397795e-05,
+      "loss": 2.0505,
+      "step": 2833
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.223224150886787e-05,
+      "loss": 1.9512,
+      "step": 2834
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.221886288375779e-05,
+      "loss": 2.0206,
+      "step": 2835
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.220548425864772e-05,
+      "loss": 1.9152,
+      "step": 2836
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.219210563353764e-05,
+      "loss": 1.7862,
+      "step": 2837
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.217872700842758e-05,
+      "loss": 1.7689,
+      "step": 2838
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.21653483833175e-05,
+      "loss": 1.8736,
+      "step": 2839
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.21519697582074e-05,
+      "loss": 1.6086,
+      "step": 2840
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.213859113309734e-05,
+      "loss": 1.6838,
+      "step": 2841
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.212521250798726e-05,
+      "loss": 1.6371,
+      "step": 2842
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.211183388287718e-05,
+      "loss": 1.7309,
+      "step": 2843
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.209845525776711e-05,
+      "loss": 1.478,
+      "step": 2844
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.208507663265703e-05,
+      "loss": 1.6577,
+      "step": 2845
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.207169800754696e-05,
+      "loss": 1.4321,
+      "step": 2846
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.205831938243688e-05,
+      "loss": 1.3348,
+      "step": 2847
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.204494075732679e-05,
+      "loss": 1.3976,
+      "step": 2848
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.203156213221672e-05,
+      "loss": 1.1783,
+      "step": 2849
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.201818350710664e-05,
+      "loss": 1.2994,
+      "step": 2850
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.200480488199658e-05,
+      "loss": 3.0412,
+      "step": 2851
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.19914262568865e-05,
+      "loss": 2.6308,
+      "step": 2852
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.197804763177642e-05,
+      "loss": 2.6347,
+      "step": 2853
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.196466900666635e-05,
+      "loss": 2.3214,
+      "step": 2854
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.195129038155627e-05,
+      "loss": 2.3307,
+      "step": 2855
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.193791175644618e-05,
+      "loss": 2.2855,
+      "step": 2856
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.192453313133611e-05,
+      "loss": 1.9935,
+      "step": 2857
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.191115450622603e-05,
+      "loss": 2.1779,
+      "step": 2858
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.189777588111597e-05,
+      "loss": 2.2189,
+      "step": 2859
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.188439725600589e-05,
+      "loss": 2.0062,
+      "step": 2860
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.18710186308958e-05,
+      "loss": 2.154,
+      "step": 2861
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.185764000578574e-05,
+      "loss": 2.2793,
+      "step": 2862
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.184426138067566e-05,
+      "loss": 2.2867,
+      "step": 2863
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.183088275556558e-05,
+      "loss": 2.1728,
+      "step": 2864
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.18175041304555e-05,
+      "loss": 2.0544,
+      "step": 2865
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.180412550534542e-05,
+      "loss": 2.182,
+      "step": 2866
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.179074688023535e-05,
+      "loss": 2.1304,
+      "step": 2867
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.177736825512527e-05,
+      "loss": 2.1312,
+      "step": 2868
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.176398963001519e-05,
+      "loss": 2.1274,
+      "step": 2869
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.175061100490513e-05,
+      "loss": 1.9501,
+      "step": 2870
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.173723237979503e-05,
+      "loss": 1.8777,
+      "step": 2871
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.172385375468497e-05,
+      "loss": 2.012,
+      "step": 2872
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.171047512957489e-05,
+      "loss": 2.0903,
+      "step": 2873
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.16970965044648e-05,
+      "loss": 2.0235,
+      "step": 2874
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.168371787935474e-05,
+      "loss": 2.2288,
+      "step": 2875
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.167033925424466e-05,
+      "loss": 1.92,
+      "step": 2876
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.165696062913458e-05,
+      "loss": 1.9403,
+      "step": 2877
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.164358200402451e-05,
+      "loss": 2.0654,
+      "step": 2878
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.163020337891442e-05,
+      "loss": 1.8867,
+      "step": 2879
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.161682475380435e-05,
+      "loss": 1.8693,
+      "step": 2880
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.160344612869427e-05,
+      "loss": 1.9047,
+      "step": 2881
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.159006750358419e-05,
+      "loss": 1.7887,
+      "step": 2882
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.157668887847413e-05,
+      "loss": 1.7976,
+      "step": 2883
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.156331025336405e-05,
+      "loss": 2.1551,
+      "step": 2884
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.154993162825398e-05,
+      "loss": 1.9932,
+      "step": 2885
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.15365530031439e-05,
+      "loss": 2.022,
+      "step": 2886
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.15231743780338e-05,
+      "loss": 1.639,
+      "step": 2887
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.150979575292374e-05,
+      "loss": 1.7547,
+      "step": 2888
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.149641712781366e-05,
+      "loss": 1.5472,
+      "step": 2889
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.148303850270358e-05,
+      "loss": 1.6834,
+      "step": 2890
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.146965987759351e-05,
+      "loss": 1.5966,
+      "step": 2891
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.145628125248343e-05,
+      "loss": 1.5981,
+      "step": 2892
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.144290262737337e-05,
+      "loss": 1.6098,
+      "step": 2893
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.142952400226327e-05,
+      "loss": 1.5291,
+      "step": 2894
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.14161453771532e-05,
+      "loss": 1.591,
+      "step": 2895
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.140276675204313e-05,
+      "loss": 1.5466,
+      "step": 2896
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.138938812693305e-05,
+      "loss": 1.3136,
+      "step": 2897
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.137600950182298e-05,
+      "loss": 1.4878,
+      "step": 2898
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.13626308767129e-05,
+      "loss": 1.0774,
+      "step": 2899
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.134925225160282e-05,
+      "loss": 1.265,
+      "step": 2900
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.133587362649275e-05,
+      "loss": 3.0134,
+      "step": 2901
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.132249500138266e-05,
+      "loss": 2.5394,
+      "step": 2902
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.130911637627258e-05,
+      "loss": 2.4225,
+      "step": 2903
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.129573775116251e-05,
+      "loss": 2.2468,
+      "step": 2904
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.128235912605243e-05,
+      "loss": 2.0914,
+      "step": 2905
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.126898050094237e-05,
+      "loss": 2.3132,
+      "step": 2906
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.125560187583229e-05,
+      "loss": 2.2168,
+      "step": 2907
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.124222325072221e-05,
+      "loss": 2.3089,
+      "step": 2908
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.122884462561214e-05,
+      "loss": 2.185,
+      "step": 2909
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.121546600050205e-05,
+      "loss": 2.112,
+      "step": 2910
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.120208737539197e-05,
+      "loss": 1.971,
+      "step": 2911
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.11887087502819e-05,
+      "loss": 2.2383,
+      "step": 2912
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.117533012517182e-05,
+      "loss": 2.1168,
+      "step": 2913
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.116195150006176e-05,
+      "loss": 1.9906,
+      "step": 2914
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.114857287495168e-05,
+      "loss": 2.2395,
+      "step": 2915
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.11351942498416e-05,
+      "loss": 2.1378,
+      "step": 2916
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.112181562473153e-05,
+      "loss": 1.9101,
+      "step": 2917
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.110843699962144e-05,
+      "loss": 2.0678,
+      "step": 2918
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.109505837451137e-05,
+      "loss": 1.9739,
+      "step": 2919
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.108167974940129e-05,
+      "loss": 2.0804,
+      "step": 2920
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.106830112429121e-05,
+      "loss": 1.9364,
+      "step": 2921
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.105492249918114e-05,
+      "loss": 2.1633,
+      "step": 2922
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.104154387407106e-05,
+      "loss": 1.9859,
+      "step": 2923
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.102816524896098e-05,
+      "loss": 2.1752,
+      "step": 2924
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.10147866238509e-05,
+      "loss": 2.1094,
+      "step": 2925
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.100140799874082e-05,
+      "loss": 1.8868,
+      "step": 2926
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.098802937363076e-05,
+      "loss": 2.1019,
+      "step": 2927
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.097465074852068e-05,
+      "loss": 2.0346,
+      "step": 2928
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.09612721234106e-05,
+      "loss": 2.1355,
+      "step": 2929
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.094789349830053e-05,
+      "loss": 2.0379,
+      "step": 2930
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.093451487319045e-05,
+      "loss": 1.8775,
+      "step": 2931
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.092113624808038e-05,
+      "loss": 1.9303,
+      "step": 2932
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.090775762297029e-05,
+      "loss": 1.7491,
+      "step": 2933
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.089437899786021e-05,
+      "loss": 1.7733,
+      "step": 2934
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.088100037275014e-05,
+      "loss": 1.5588,
+      "step": 2935
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.086762174764006e-05,
+      "loss": 1.789,
+      "step": 2936
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.085424312252998e-05,
+      "loss": 1.7953,
+      "step": 2937
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.084086449741992e-05,
+      "loss": 1.7193,
+      "step": 2938
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.082748587230984e-05,
+      "loss": 1.5204,
+      "step": 2939
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.081410724719977e-05,
+      "loss": 1.7999,
+      "step": 2940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.080072862208968e-05,
+      "loss": 1.7858,
+      "step": 2941
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.07873499969796e-05,
+      "loss": 1.6926,
+      "step": 2942
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.077397137186953e-05,
+      "loss": 1.6842,
+      "step": 2943
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.076059274675945e-05,
+      "loss": 1.6048,
+      "step": 2944
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.074721412164938e-05,
+      "loss": 1.5865,
+      "step": 2945
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.07338354965393e-05,
+      "loss": 1.3463,
+      "step": 2946
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.072045687142922e-05,
+      "loss": 1.2797,
+      "step": 2947
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.070707824631914e-05,
+      "loss": 1.2904,
+      "step": 2948
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.069369962120906e-05,
+      "loss": 1.1099,
+      "step": 2949
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.068032099609898e-05,
+      "loss": 1.2692,
+      "step": 2950
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.066694237098892e-05,
+      "loss": 2.9031,
+      "step": 2951
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.065356374587884e-05,
+      "loss": 2.5559,
+      "step": 2952
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.064018512076877e-05,
+      "loss": 2.3015,
+      "step": 2953
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.062680649565869e-05,
+      "loss": 2.1154,
+      "step": 2954
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.061342787054861e-05,
+      "loss": 2.1141,
+      "step": 2955
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.060004924543853e-05,
+      "loss": 2.1293,
+      "step": 2956
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.058667062032845e-05,
+      "loss": 2.2,
+      "step": 2957
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.057329199521837e-05,
+      "loss": 2.1751,
+      "step": 2958
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.05599133701083e-05,
+      "loss": 2.0411,
+      "step": 2959
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.054653474499822e-05,
+      "loss": 2.2911,
+      "step": 2960
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.053315611988816e-05,
+      "loss": 2.0404,
+      "step": 2961
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.051977749477808e-05,
+      "loss": 2.2754,
+      "step": 2962
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.0506398869668e-05,
+      "loss": 1.9898,
+      "step": 2963
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.049302024455792e-05,
+      "loss": 1.7801,
+      "step": 2964
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.047964161944784e-05,
+      "loss": 2.1171,
+      "step": 2965
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.046626299433777e-05,
+      "loss": 2.048,
+      "step": 2966
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.045288436922769e-05,
+      "loss": 2.049,
+      "step": 2967
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.043950574411761e-05,
+      "loss": 2.0942,
+      "step": 2968
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.042612711900754e-05,
+      "loss": 2.246,
+      "step": 2969
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.041274849389746e-05,
+      "loss": 2.2959,
+      "step": 2970
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.039936986878737e-05,
+      "loss": 2.1461,
+      "step": 2971
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.03859912436773e-05,
+      "loss": 2.0121,
+      "step": 2972
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.037261261856722e-05,
+      "loss": 2.1276,
+      "step": 2973
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.035923399345716e-05,
+      "loss": 2.0186,
+      "step": 2974
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.034585536834708e-05,
+      "loss": 1.961,
+      "step": 2975
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.0332476743237e-05,
+      "loss": 2.0439,
+      "step": 2976
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.031909811812693e-05,
+      "loss": 1.9318,
+      "step": 2977
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.030571949301685e-05,
+      "loss": 1.9989,
+      "step": 2978
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.029234086790677e-05,
+      "loss": 1.8855,
+      "step": 2979
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.027896224279669e-05,
+      "loss": 1.8618,
+      "step": 2980
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.026558361768661e-05,
+      "loss": 1.646,
+      "step": 2981
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.025220499257655e-05,
+      "loss": 1.8677,
+      "step": 2982
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.023882636746647e-05,
+      "loss": 1.6623,
+      "step": 2983
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.022544774235639e-05,
+      "loss": 1.9259,
+      "step": 2984
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.021206911724632e-05,
+      "loss": 1.8806,
+      "step": 2985
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.019869049213624e-05,
+      "loss": 1.7267,
+      "step": 2986
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.018531186702616e-05,
+      "loss": 1.6521,
+      "step": 2987
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.017193324191608e-05,
+      "loss": 1.8011,
+      "step": 2988
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.0158554616806e-05,
+      "loss": 1.8187,
+      "step": 2989
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.014517599169593e-05,
+      "loss": 1.4386,
+      "step": 2990
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.013179736658585e-05,
+      "loss": 1.441,
+      "step": 2991
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.011841874147577e-05,
+      "loss": 1.4715,
+      "step": 2992
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.01050401163657e-05,
+      "loss": 1.5036,
+      "step": 2993
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.009166149125563e-05,
+      "loss": 1.5913,
+      "step": 2994
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.007828286614555e-05,
+      "loss": 1.5118,
+      "step": 2995
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.006490424103547e-05,
+      "loss": 1.585,
+      "step": 2996
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.005152561592539e-05,
+      "loss": 1.1989,
+      "step": 2997
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.003814699081532e-05,
+      "loss": 1.3055,
+      "step": 2998
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.002476836570524e-05,
+      "loss": 1.4213,
+      "step": 2999
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.001138974059517e-05,
+      "loss": 1.3315,
+      "step": 3000
+    },
+    {
+      "epoch": 0.93,
+      "eval_bleu": 1.5460941713493133e-18,
+      "eval_loss": 2.3100783824920654,
+      "eval_runtime": 2255.2098,
+      "eval_samples_per_second": 6.545,
+      "eval_steps_per_second": 0.818,
+      "step": 3000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.99980111154851e-05,
+      "loss": 2.8218,
+      "step": 3001
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.9984632490375e-05,
+      "loss": 2.7045,
+      "step": 3002
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.997125386526493e-05,
+      "loss": 2.5341,
+      "step": 3003
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.995787524015485e-05,
+      "loss": 2.5728,
+      "step": 3004
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.994449661504477e-05,
+      "loss": 2.2582,
+      "step": 3005
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.993111798993471e-05,
+      "loss": 2.1951,
+      "step": 3006
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.991773936482463e-05,
+      "loss": 2.403,
+      "step": 3007
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.990436073971456e-05,
+      "loss": 2.2342,
+      "step": 3008
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.989098211460448e-05,
+      "loss": 2.4197,
+      "step": 3009
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.987760348949439e-05,
+      "loss": 2.0284,
+      "step": 3010
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.986422486438432e-05,
+      "loss": 2.0281,
+      "step": 3011
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.985084623927424e-05,
+      "loss": 2.2045,
+      "step": 3012
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.983746761416417e-05,
+      "loss": 2.0293,
+      "step": 3013
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.98240889890541e-05,
+      "loss": 2.0389,
+      "step": 3014
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.981071036394401e-05,
+      "loss": 1.9268,
+      "step": 3015
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.979733173883395e-05,
+      "loss": 2.3066,
+      "step": 3016
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.978395311372387e-05,
+      "loss": 1.9293,
+      "step": 3017
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.977057448861377e-05,
+      "loss": 2.402,
+      "step": 3018
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.975719586350371e-05,
+      "loss": 2.0964,
+      "step": 3019
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.974381723839363e-05,
+      "loss": 2.0038,
+      "step": 3020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.973043861328356e-05,
+      "loss": 1.8441,
+      "step": 3021
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.971705998817348e-05,
+      "loss": 2.0104,
+      "step": 3022
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.97036813630634e-05,
+      "loss": 2.054,
+      "step": 3023
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.969030273795333e-05,
+      "loss": 2.0597,
+      "step": 3024
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.967692411284324e-05,
+      "loss": 1.8892,
+      "step": 3025
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.966354548773317e-05,
+      "loss": 2.0396,
+      "step": 3026
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.96501668626231e-05,
+      "loss": 1.8961,
+      "step": 3027
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.963678823751301e-05,
+      "loss": 2.0708,
+      "step": 3028
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.962340961240295e-05,
+      "loss": 1.9785,
+      "step": 3029
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.961003098729287e-05,
+      "loss": 1.7634,
+      "step": 3030
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.959665236218279e-05,
+      "loss": 1.8375,
+      "step": 3031
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.958327373707272e-05,
+      "loss": 1.9504,
+      "step": 3032
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.956989511196263e-05,
+      "loss": 1.6759,
+      "step": 3033
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.955651648685256e-05,
+      "loss": 1.8173,
+      "step": 3034
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.954313786174248e-05,
+      "loss": 1.6195,
+      "step": 3035
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.95297592366324e-05,
+      "loss": 1.8411,
+      "step": 3036
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.951638061152234e-05,
+      "loss": 1.7753,
+      "step": 3037
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.950300198641226e-05,
+      "loss": 1.9761,
+      "step": 3038
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.948962336130218e-05,
+      "loss": 1.6252,
+      "step": 3039
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.947624473619211e-05,
+      "loss": 1.6587,
+      "step": 3040
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.946286611108202e-05,
+      "loss": 1.7748,
+      "step": 3041
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.944948748597195e-05,
+      "loss": 1.617,
+      "step": 3042
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.943610886086187e-05,
+      "loss": 1.4652,
+      "step": 3043
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.942273023575179e-05,
+      "loss": 1.4379,
+      "step": 3044
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.940935161064172e-05,
+      "loss": 1.5572,
+      "step": 3045
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.939597298553164e-05,
+      "loss": 1.3431,
+      "step": 3046
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.938259436042158e-05,
+      "loss": 1.2502,
+      "step": 3047
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.93692157353115e-05,
+      "loss": 1.2921,
+      "step": 3048
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.93558371102014e-05,
+      "loss": 1.2573,
+      "step": 3049
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.934245848509134e-05,
+      "loss": 1.1872,
+      "step": 3050
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.932907985998126e-05,
+      "loss": 2.8393,
+      "step": 3051
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.931570123487118e-05,
+      "loss": 2.6963,
+      "step": 3052
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.930232260976111e-05,
+      "loss": 2.3484,
+      "step": 3053
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.928894398465103e-05,
+      "loss": 2.2785,
+      "step": 3054
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.927556535954096e-05,
+      "loss": 2.356,
+      "step": 3055
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.926218673443087e-05,
+      "loss": 2.1585,
+      "step": 3056
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.924880810932079e-05,
+      "loss": 2.4368,
+      "step": 3057
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.923542948421072e-05,
+      "loss": 2.419,
+      "step": 3058
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.922205085910064e-05,
+      "loss": 2.0664,
+      "step": 3059
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.920867223399058e-05,
+      "loss": 2.2705,
+      "step": 3060
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.91952936088805e-05,
+      "loss": 2.1606,
+      "step": 3061
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.918191498377042e-05,
+      "loss": 2.2808,
+      "step": 3062
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.916853635866035e-05,
+      "loss": 2.136,
+      "step": 3063
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.915515773355026e-05,
+      "loss": 2.01,
+      "step": 3064
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.914177910844018e-05,
+      "loss": 1.8596,
+      "step": 3065
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.912840048333011e-05,
+      "loss": 1.8623,
+      "step": 3066
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.911502185822003e-05,
+      "loss": 1.9013,
+      "step": 3067
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.910164323310996e-05,
+      "loss": 2.048,
+      "step": 3068
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.908826460799988e-05,
+      "loss": 1.94,
+      "step": 3069
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.90748859828898e-05,
+      "loss": 2.0695,
+      "step": 3070
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.906150735777974e-05,
+      "loss": 2.3136,
+      "step": 3071
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.904812873266964e-05,
+      "loss": 2.0808,
+      "step": 3072
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.903475010755956e-05,
+      "loss": 1.9789,
+      "step": 3073
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.90213714824495e-05,
+      "loss": 1.9764,
+      "step": 3074
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.900799285733942e-05,
+      "loss": 2.0882,
+      "step": 3075
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.899461423222935e-05,
+      "loss": 2.0722,
+      "step": 3076
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.898123560711927e-05,
+      "loss": 2.0232,
+      "step": 3077
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.896785698200919e-05,
+      "loss": 1.8178,
+      "step": 3078
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.895447835689911e-05,
+      "loss": 1.7876,
+      "step": 3079
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.894109973178903e-05,
+      "loss": 2.1755,
+      "step": 3080
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.892772110667896e-05,
+      "loss": 2.1362,
+      "step": 3081
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.891434248156888e-05,
+      "loss": 1.9302,
+      "step": 3082
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.89009638564588e-05,
+      "loss": 1.8466,
+      "step": 3083
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.888758523134874e-05,
+      "loss": 1.8298,
+      "step": 3084
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.887420660623866e-05,
+      "loss": 1.5566,
+      "step": 3085
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.886082798112858e-05,
+      "loss": 1.7556,
+      "step": 3086
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.88474493560185e-05,
+      "loss": 1.8389,
+      "step": 3087
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.883407073090842e-05,
+      "loss": 1.6387,
+      "step": 3088
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.882069210579835e-05,
+      "loss": 1.3414,
+      "step": 3089
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.880731348068827e-05,
+      "loss": 1.5627,
+      "step": 3090
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.879393485557819e-05,
+      "loss": 1.4989,
+      "step": 3091
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.878055623046813e-05,
+      "loss": 1.5728,
+      "step": 3092
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.876717760535805e-05,
+      "loss": 1.5778,
+      "step": 3093
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.875379898024798e-05,
+      "loss": 1.4592,
+      "step": 3094
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.874042035513789e-05,
+      "loss": 1.4706,
+      "step": 3095
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.87270417300278e-05,
+      "loss": 1.4656,
+      "step": 3096
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.871366310491774e-05,
+      "loss": 1.2652,
+      "step": 3097
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.870028447980766e-05,
+      "loss": 1.3656,
+      "step": 3098
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.868690585469758e-05,
+      "loss": 1.2134,
+      "step": 3099
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.867352722958751e-05,
+      "loss": 1.1417,
+      "step": 3100
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.866014860447743e-05,
+      "loss": 2.7865,
+      "step": 3101
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.864676997936737e-05,
+      "loss": 2.455,
+      "step": 3102
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.863339135425727e-05,
+      "loss": 2.5106,
+      "step": 3103
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.862001272914719e-05,
+      "loss": 2.3741,
+      "step": 3104
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.860663410403713e-05,
+      "loss": 2.3406,
+      "step": 3105
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.859325547892705e-05,
+      "loss": 2.2665,
+      "step": 3106
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.857987685381697e-05,
+      "loss": 2.1459,
+      "step": 3107
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.85664982287069e-05,
+      "loss": 2.2166,
+      "step": 3108
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.855311960359682e-05,
+      "loss": 2.1239,
+      "step": 3109
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.853974097848674e-05,
+      "loss": 1.8732,
+      "step": 3110
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.852636235337666e-05,
+      "loss": 2.0395,
+      "step": 3111
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.851298372826658e-05,
+      "loss": 2.0423,
+      "step": 3112
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.849960510315651e-05,
+      "loss": 1.9966,
+      "step": 3113
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.848622647804643e-05,
+      "loss": 2.2917,
+      "step": 3114
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.847284785293637e-05,
+      "loss": 1.9531,
+      "step": 3115
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.845946922782629e-05,
+      "loss": 1.9659,
+      "step": 3116
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.84460906027162e-05,
+      "loss": 2.0778,
+      "step": 3117
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.843271197760613e-05,
+      "loss": 2.2429,
+      "step": 3118
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.841933335249605e-05,
+      "loss": 2.0014,
+      "step": 3119
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.840595472738597e-05,
+      "loss": 2.3287,
+      "step": 3120
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.83925761022759e-05,
+      "loss": 1.9515,
+      "step": 3121
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.837919747716582e-05,
+      "loss": 1.8973,
+      "step": 3122
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.836581885205575e-05,
+      "loss": 2.0701,
+      "step": 3123
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.835244022694567e-05,
+      "loss": 2.0848,
+      "step": 3124
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.83390616018356e-05,
+      "loss": 1.8035,
+      "step": 3125
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.832568297672551e-05,
+      "loss": 2.0733,
+      "step": 3126
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.831230435161543e-05,
+      "loss": 1.8923,
+      "step": 3127
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.829892572650537e-05,
+      "loss": 1.9816,
+      "step": 3128
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.828554710139529e-05,
+      "loss": 1.8883,
+      "step": 3129
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.827216847628521e-05,
+      "loss": 1.6424,
+      "step": 3130
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.825878985117514e-05,
+      "loss": 2.029,
+      "step": 3131
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.824541122606506e-05,
+      "loss": 1.6896,
+      "step": 3132
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.823203260095497e-05,
+      "loss": 1.6729,
+      "step": 3133
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.82186539758449e-05,
+      "loss": 1.9219,
+      "step": 3134
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.820527535073482e-05,
+      "loss": 1.6505,
+      "step": 3135
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.819189672562475e-05,
+      "loss": 1.8621,
+      "step": 3136
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.817851810051467e-05,
+      "loss": 1.6977,
+      "step": 3137
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.81651394754046e-05,
+      "loss": 1.5603,
+      "step": 3138
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.815176085029453e-05,
+      "loss": 1.7713,
+      "step": 3139
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.813838222518445e-05,
+      "loss": 1.6139,
+      "step": 3140
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.812500360007437e-05,
+      "loss": 1.7116,
+      "step": 3141
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.811162497496429e-05,
+      "loss": 1.454,
+      "step": 3142
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.809824634985421e-05,
+      "loss": 1.3827,
+      "step": 3143
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.808486772474414e-05,
+      "loss": 1.5542,
+      "step": 3144
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.807148909963406e-05,
+      "loss": 1.1608,
+      "step": 3145
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.805811047452398e-05,
+      "loss": 1.2162,
+      "step": 3146
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.804473184941392e-05,
+      "loss": 1.3131,
+      "step": 3147
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.803135322430384e-05,
+      "loss": 1.0222,
+      "step": 3148
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.801797459919376e-05,
+      "loss": 1.0897,
+      "step": 3149
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.800459597408368e-05,
+      "loss": 1.4381,
+      "step": 3150
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.79912173489736e-05,
+      "loss": 2.8572,
+      "step": 3151
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.797783872386353e-05,
+      "loss": 2.4274,
+      "step": 3152
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.796446009875345e-05,
+      "loss": 2.2154,
+      "step": 3153
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.795108147364337e-05,
+      "loss": 2.1212,
+      "step": 3154
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.79377028485333e-05,
+      "loss": 2.0574,
+      "step": 3155
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.792432422342322e-05,
+      "loss": 2.2667,
+      "step": 3156
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.791094559831314e-05,
+      "loss": 2.1861,
+      "step": 3157
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.789756697320306e-05,
+      "loss": 2.1517,
+      "step": 3158
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.788418834809298e-05,
+      "loss": 2.4323,
+      "step": 3159
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.787080972298292e-05,
+      "loss": 2.1588,
+      "step": 3160
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.785743109787284e-05,
+      "loss": 2.2814,
+      "step": 3161
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.784405247276277e-05,
+      "loss": 2.2388,
+      "step": 3162
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.783067384765269e-05,
+      "loss": 2.1205,
+      "step": 3163
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.78172952225426e-05,
+      "loss": 2.0766,
+      "step": 3164
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.780391659743253e-05,
+      "loss": 2.1664,
+      "step": 3165
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.779053797232245e-05,
+      "loss": 1.8835,
+      "step": 3166
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.777715934721237e-05,
+      "loss": 1.8944,
+      "step": 3167
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.77637807221023e-05,
+      "loss": 1.9177,
+      "step": 3168
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.775040209699222e-05,
+      "loss": 1.9792,
+      "step": 3169
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.773702347188216e-05,
+      "loss": 2.1647,
+      "step": 3170
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.772364484677208e-05,
+      "loss": 2.1926,
+      "step": 3171
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.771026622166198e-05,
+      "loss": 1.9336,
+      "step": 3172
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.769688759655192e-05,
+      "loss": 1.8098,
+      "step": 3173
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.768350897144184e-05,
+      "loss": 1.898,
+      "step": 3174
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.767013034633177e-05,
+      "loss": 1.9173,
+      "step": 3175
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.765675172122169e-05,
+      "loss": 1.8899,
+      "step": 3176
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.764337309611161e-05,
+      "loss": 1.9246,
+      "step": 3177
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.762999447100154e-05,
+      "loss": 1.8543,
+      "step": 3178
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.761661584589146e-05,
+      "loss": 1.7991,
+      "step": 3179
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.760323722078137e-05,
+      "loss": 1.6822,
+      "step": 3180
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.75898585956713e-05,
+      "loss": 1.5594,
+      "step": 3181
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.757647997056122e-05,
+      "loss": 1.8877,
+      "step": 3182
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.756310134545116e-05,
+      "loss": 1.7317,
+      "step": 3183
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.754972272034108e-05,
+      "loss": 1.8524,
+      "step": 3184
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.7536344095231e-05,
+      "loss": 1.5392,
+      "step": 3185
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.752296547012093e-05,
+      "loss": 1.9974,
+      "step": 3186
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.750958684501084e-05,
+      "loss": 1.6806,
+      "step": 3187
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.749620821990076e-05,
+      "loss": 1.7131,
+      "step": 3188
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.748282959479069e-05,
+      "loss": 1.5006,
+      "step": 3189
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.746945096968061e-05,
+      "loss": 1.5644,
+      "step": 3190
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.745607234457054e-05,
+      "loss": 1.7061,
+      "step": 3191
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.744269371946046e-05,
+      "loss": 1.6635,
+      "step": 3192
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.742931509435038e-05,
+      "loss": 1.7089,
+      "step": 3193
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.741593646924032e-05,
+      "loss": 1.5743,
+      "step": 3194
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.740255784413022e-05,
+      "loss": 1.6174,
+      "step": 3195
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.738917921902016e-05,
+      "loss": 1.1489,
+      "step": 3196
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.737580059391008e-05,
+      "loss": 1.2183,
+      "step": 3197
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.73624219688e-05,
+      "loss": 1.2503,
+      "step": 3198
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.734904334368993e-05,
+      "loss": 1.0498,
+      "step": 3199
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.733566471857985e-05,
+      "loss": 1.271,
+      "step": 3200
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.732228609346977e-05,
+      "loss": 2.7099,
+      "step": 3201
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.73089074683597e-05,
+      "loss": 2.6479,
+      "step": 3202
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.729552884324961e-05,
+      "loss": 2.1954,
+      "step": 3203
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.728215021813955e-05,
+      "loss": 2.1332,
+      "step": 3204
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.726877159302947e-05,
+      "loss": 2.0733,
+      "step": 3205
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.725539296791939e-05,
+      "loss": 2.19,
+      "step": 3206
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.724201434280932e-05,
+      "loss": 2.2291,
+      "step": 3207
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.722863571769924e-05,
+      "loss": 2.1006,
+      "step": 3208
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.721525709258917e-05,
+      "loss": 2.0775,
+      "step": 3209
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.720187846747909e-05,
+      "loss": 2.1052,
+      "step": 3210
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.7188499842369e-05,
+      "loss": 2.2995,
+      "step": 3211
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.717512121725893e-05,
+      "loss": 1.8738,
+      "step": 3212
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.716174259214885e-05,
+      "loss": 1.9656,
+      "step": 3213
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.714836396703877e-05,
+      "loss": 2.0262,
+      "step": 3214
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.71349853419287e-05,
+      "loss": 2.0363,
+      "step": 3215
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.712160671681863e-05,
+      "loss": 2.1399,
+      "step": 3216
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.710822809170856e-05,
+      "loss": 2.0125,
+      "step": 3217
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.709484946659847e-05,
+      "loss": 2.0448,
+      "step": 3218
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.708147084148839e-05,
+      "loss": 1.9814,
+      "step": 3219
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.706809221637832e-05,
+      "loss": 2.0265,
+      "step": 3220
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.705471359126824e-05,
+      "loss": 1.8186,
+      "step": 3221
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.704133496615817e-05,
+      "loss": 2.0267,
+      "step": 3222
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.702795634104809e-05,
+      "loss": 1.9065,
+      "step": 3223
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.701457771593801e-05,
+      "loss": 1.9334,
+      "step": 3224
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.700119909082795e-05,
+      "loss": 1.8154,
+      "step": 3225
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.698782046571785e-05,
+      "loss": 1.8005,
+      "step": 3226
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.697444184060777e-05,
+      "loss": 1.805,
+      "step": 3227
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.69610632154977e-05,
+      "loss": 1.7867,
+      "step": 3228
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.694768459038763e-05,
+      "loss": 1.6077,
+      "step": 3229
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.693430596527756e-05,
+      "loss": 1.7036,
+      "step": 3230
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.692092734016748e-05,
+      "loss": 1.6364,
+      "step": 3231
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.69075487150574e-05,
+      "loss": 1.7775,
+      "step": 3232
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.689417008994733e-05,
+      "loss": 1.5539,
+      "step": 3233
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.688079146483724e-05,
+      "loss": 1.5594,
+      "step": 3234
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.686741283972716e-05,
+      "loss": 1.7378,
+      "step": 3235
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.68540342146171e-05,
+      "loss": 1.5749,
+      "step": 3236
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.684065558950701e-05,
+      "loss": 1.275,
+      "step": 3237
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.682727696439695e-05,
+      "loss": 1.33,
+      "step": 3238
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.681389833928687e-05,
+      "loss": 1.1281,
+      "step": 3239
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.680051971417679e-05,
+      "loss": 1.2243,
+      "step": 3240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.678714108906671e-05,
+      "loss": 2.8776,
+      "step": 3241
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.677376246395663e-05,
+      "loss": 2.3469,
+      "step": 3242
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.676038383884656e-05,
+      "loss": 2.0925,
+      "step": 3243
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.674700521373648e-05,
+      "loss": 2.1499,
+      "step": 3244
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.67336265886264e-05,
+      "loss": 1.9854,
+      "step": 3245
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.672024796351633e-05,
+      "loss": 1.9881,
+      "step": 3246
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.670686933840625e-05,
+      "loss": 1.7776,
+      "step": 3247
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.669349071329617e-05,
+      "loss": 1.9587,
+      "step": 3248
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.66801120881861e-05,
+      "loss": 1.9636,
+      "step": 3249
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.666673346307601e-05,
+      "loss": 1.8845,
+      "step": 3250
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.665335483796595e-05,
+      "loss": 1.8064,
+      "step": 3251
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.663997621285587e-05,
+      "loss": 1.711,
+      "step": 3252
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.662659758774579e-05,
+      "loss": 1.7904,
+      "step": 3253
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.661321896263572e-05,
+      "loss": 1.7235,
+      "step": 3254
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.659984033752564e-05,
+      "loss": 1.7833,
+      "step": 3255
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.658646171241557e-05,
+      "loss": 1.6687,
+      "step": 3256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.657308308730548e-05,
+      "loss": 1.7951,
+      "step": 3257
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.65597044621954e-05,
+      "loss": 1.669,
+      "step": 3258
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.654632583708533e-05,
+      "loss": 1.4248,
+      "step": 3259
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.653294721197525e-05,
+      "loss": 1.7651,
+      "step": 3260
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.651956858686517e-05,
+      "loss": 1.5805,
+      "step": 3261
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.650618996175511e-05,
+      "loss": 1.6218,
+      "step": 3262
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.649281133664503e-05,
+      "loss": 1.6703,
+      "step": 3263
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.647943271153496e-05,
+      "loss": 1.6292,
+      "step": 3264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.646605408642487e-05,
+      "loss": 1.6436,
+      "step": 3265
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.645267546131479e-05,
+      "loss": 1.7488,
+      "step": 3266
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.643929683620472e-05,
+      "loss": 1.6605,
+      "step": 3267
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.642591821109464e-05,
+      "loss": 1.5919,
+      "step": 3268
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.641253958598456e-05,
+      "loss": 1.6709,
+      "step": 3269
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.63991609608745e-05,
+      "loss": 1.5963,
+      "step": 3270
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.638578233576442e-05,
+      "loss": 1.507,
+      "step": 3271
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.637240371065434e-05,
+      "loss": 1.4709,
+      "step": 3272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.635902508554426e-05,
+      "loss": 1.495,
+      "step": 3273
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.634564646043418e-05,
+      "loss": 1.3283,
+      "step": 3274
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.633226783532411e-05,
+      "loss": 1.4856,
+      "step": 3275
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.631888921021403e-05,
+      "loss": 1.3789,
+      "step": 3276
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.630551058510396e-05,
+      "loss": 1.5656,
+      "step": 3277
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.629213195999388e-05,
+      "loss": 1.4759,
+      "step": 3278
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.62787533348838e-05,
+      "loss": 1.4147,
+      "step": 3279
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.626537470977372e-05,
+      "loss": 1.3661,
+      "step": 3280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.625199608466364e-05,
+      "loss": 1.4823,
+      "step": 3281
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.623861745955356e-05,
+      "loss": 1.487,
+      "step": 3282
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.62252388344435e-05,
+      "loss": 1.2992,
+      "step": 3283
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.621186020933342e-05,
+      "loss": 1.2368,
+      "step": 3284
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.619848158422335e-05,
+      "loss": 1.1219,
+      "step": 3285
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.618510295911327e-05,
+      "loss": 1.2933,
+      "step": 3286
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.617172433400319e-05,
+      "loss": 1.0904,
+      "step": 3287
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.615834570889311e-05,
+      "loss": 1.1612,
+      "step": 3288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.614496708378303e-05,
+      "loss": 1.1799,
+      "step": 3289
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.613158845867296e-05,
+      "loss": 1.0113,
+      "step": 3290
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.611820983356288e-05,
+      "loss": 2.4722,
+      "step": 3291
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.61048312084528e-05,
+      "loss": 2.2872,
+      "step": 3292
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.609145258334274e-05,
+      "loss": 2.1149,
+      "step": 3293
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.607807395823266e-05,
+      "loss": 2.1181,
+      "step": 3294
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.606469533312256e-05,
+      "loss": 1.8649,
+      "step": 3295
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.60513167080125e-05,
+      "loss": 1.7229,
+      "step": 3296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.603793808290242e-05,
+      "loss": 1.9108,
+      "step": 3297
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.602455945779235e-05,
+      "loss": 1.8204,
+      "step": 3298
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.601118083268227e-05,
+      "loss": 1.9331,
+      "step": 3299
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.599780220757219e-05,
+      "loss": 1.9223,
+      "step": 3300
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.598442358246212e-05,
+      "loss": 1.7493,
+      "step": 3301
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.597104495735204e-05,
+      "loss": 1.6404,
+      "step": 3302
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.595766633224195e-05,
+      "loss": 1.9073,
+      "step": 3303
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.594428770713188e-05,
+      "loss": 1.7675,
+      "step": 3304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.59309090820218e-05,
+      "loss": 1.8226,
+      "step": 3305
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.591753045691174e-05,
+      "loss": 1.7019,
+      "step": 3306
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.590415183180166e-05,
+      "loss": 1.8587,
+      "step": 3307
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.589077320669158e-05,
+      "loss": 1.5548,
+      "step": 3308
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.587739458158151e-05,
+      "loss": 1.7112,
+      "step": 3309
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.586401595647143e-05,
+      "loss": 1.7475,
+      "step": 3310
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.585063733136135e-05,
+      "loss": 1.6008,
+      "step": 3311
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.583725870625127e-05,
+      "loss": 1.5654,
+      "step": 3312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.582388008114119e-05,
+      "loss": 1.5575,
+      "step": 3313
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.581050145603112e-05,
+      "loss": 1.8465,
+      "step": 3314
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.579712283092104e-05,
+      "loss": 1.6709,
+      "step": 3315
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.578374420581096e-05,
+      "loss": 1.6695,
+      "step": 3316
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.57703655807009e-05,
+      "loss": 1.5795,
+      "step": 3317
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.57569869555908e-05,
+      "loss": 1.5738,
+      "step": 3318
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.574360833048074e-05,
+      "loss": 1.4624,
+      "step": 3319
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.573022970537066e-05,
+      "loss": 1.6294,
+      "step": 3320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.571685108026058e-05,
+      "loss": 1.608,
+      "step": 3321
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.570347245515051e-05,
+      "loss": 1.6662,
+      "step": 3322
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.569009383004043e-05,
+      "loss": 1.5956,
+      "step": 3323
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.567671520493037e-05,
+      "loss": 1.6215,
+      "step": 3324
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.566333657982029e-05,
+      "loss": 1.5084,
+      "step": 3325
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.564995795471019e-05,
+      "loss": 1.4742,
+      "step": 3326
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.563657932960013e-05,
+      "loss": 1.2467,
+      "step": 3327
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.562320070449005e-05,
+      "loss": 1.4826,
+      "step": 3328
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.560982207937997e-05,
+      "loss": 1.404,
+      "step": 3329
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.55964434542699e-05,
+      "loss": 1.5169,
+      "step": 3330
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.558306482915982e-05,
+      "loss": 1.4562,
+      "step": 3331
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.556968620404975e-05,
+      "loss": 1.3518,
+      "step": 3332
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.555630757893967e-05,
+      "loss": 1.2445,
+      "step": 3333
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.554292895382958e-05,
+      "loss": 1.3067,
+      "step": 3334
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.552955032871951e-05,
+      "loss": 1.082,
+      "step": 3335
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.551617170360943e-05,
+      "loss": 0.9913,
+      "step": 3336
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.550279307849937e-05,
+      "loss": 1.0545,
+      "step": 3337
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.548941445338929e-05,
+      "loss": 1.1517,
+      "step": 3338
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.54760358282792e-05,
+      "loss": 0.926,
+      "step": 3339
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.546265720316914e-05,
+      "loss": 1.1249,
+      "step": 3340
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.544927857805906e-05,
+      "loss": 2.6649,
+      "step": 3341
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.543589995294897e-05,
+      "loss": 2.266,
+      "step": 3342
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.54225213278389e-05,
+      "loss": 2.1494,
+      "step": 3343
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.540914270272882e-05,
+      "loss": 2.1282,
+      "step": 3344
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.539576407761875e-05,
+      "loss": 1.939,
+      "step": 3345
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.538238545250867e-05,
+      "loss": 1.92,
+      "step": 3346
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.536900682739859e-05,
+      "loss": 1.8326,
+      "step": 3347
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.535562820228853e-05,
+      "loss": 1.9571,
+      "step": 3348
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.534224957717843e-05,
+      "loss": 1.8329,
+      "step": 3349
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.532887095206835e-05,
+      "loss": 1.6408,
+      "step": 3350
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.531549232695829e-05,
+      "loss": 1.7899,
+      "step": 3351
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.53021137018482e-05,
+      "loss": 1.9002,
+      "step": 3352
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.528873507673814e-05,
+      "loss": 1.6449,
+      "step": 3353
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.527535645162806e-05,
+      "loss": 1.8235,
+      "step": 3354
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.526197782651798e-05,
+      "loss": 1.7896,
+      "step": 3355
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.524859920140791e-05,
+      "loss": 1.756,
+      "step": 3356
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.523522057629782e-05,
+      "loss": 1.4531,
+      "step": 3357
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.522184195118775e-05,
+      "loss": 1.8985,
+      "step": 3358
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.520846332607767e-05,
+      "loss": 1.7736,
+      "step": 3359
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.51950847009676e-05,
+      "loss": 1.6225,
+      "step": 3360
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.518170607585753e-05,
+      "loss": 1.7558,
+      "step": 3361
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.516832745074745e-05,
+      "loss": 1.5106,
+      "step": 3362
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.515494882563737e-05,
+      "loss": 1.8595,
+      "step": 3363
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.51415702005273e-05,
+      "loss": 1.77,
+      "step": 3364
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.512819157541721e-05,
+      "loss": 1.5502,
+      "step": 3365
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.511481295030714e-05,
+      "loss": 1.7253,
+      "step": 3366
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.510143432519706e-05,
+      "loss": 1.7264,
+      "step": 3367
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.508805570008698e-05,
+      "loss": 1.5481,
+      "step": 3368
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.507467707497691e-05,
+      "loss": 1.4994,
+      "step": 3369
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.506129844986683e-05,
+      "loss": 1.7147,
+      "step": 3370
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.504791982475677e-05,
+      "loss": 1.5869,
+      "step": 3371
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.503454119964667e-05,
+      "loss": 1.3196,
+      "step": 3372
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.50211625745366e-05,
+      "loss": 1.493,
+      "step": 3373
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.500778394942653e-05,
+      "loss": 1.3579,
+      "step": 3374
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.499440532431645e-05,
+      "loss": 1.4568,
+      "step": 3375
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.498102669920637e-05,
+      "loss": 1.5575,
+      "step": 3376
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.49676480740963e-05,
+      "loss": 1.3821,
+      "step": 3377
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.495426944898622e-05,
+      "loss": 1.4516,
+      "step": 3378
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.494089082387616e-05,
+      "loss": 1.4673,
+      "step": 3379
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.492751219876606e-05,
+      "loss": 1.4237,
+      "step": 3380
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.491413357365598e-05,
+      "loss": 1.3515,
+      "step": 3381
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.490075494854592e-05,
+      "loss": 1.189,
+      "step": 3382
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.488737632343584e-05,
+      "loss": 1.2884,
+      "step": 3383
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.487399769832576e-05,
+      "loss": 1.2341,
+      "step": 3384
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.486061907321569e-05,
+      "loss": 1.1543,
+      "step": 3385
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.484724044810561e-05,
+      "loss": 1.1333,
+      "step": 3386
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.483386182299554e-05,
+      "loss": 1.0052,
+      "step": 3387
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.482048319788545e-05,
+      "loss": 1.0078,
+      "step": 3388
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.480710457277537e-05,
+      "loss": 1.2198,
+      "step": 3389
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.47937259476653e-05,
+      "loss": 0.9277,
+      "step": 3390
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.478034732255522e-05,
+      "loss": 2.3788,
+      "step": 3391
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.476696869744516e-05,
+      "loss": 2.1255,
+      "step": 3392
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.475359007233508e-05,
+      "loss": 2.0486,
+      "step": 3393
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.4740211447225e-05,
+      "loss": 1.983,
+      "step": 3394
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.472683282211493e-05,
+      "loss": 1.9868,
+      "step": 3395
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.471345419700484e-05,
+      "loss": 1.8588,
+      "step": 3396
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.470007557189476e-05,
+      "loss": 1.9275,
+      "step": 3397
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.468669694678469e-05,
+      "loss": 1.908,
+      "step": 3398
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.467331832167461e-05,
+      "loss": 1.9783,
+      "step": 3399
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.465993969656454e-05,
+      "loss": 1.728,
+      "step": 3400
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.464656107145446e-05,
+      "loss": 1.6039,
+      "step": 3401
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.463318244634438e-05,
+      "loss": 1.8724,
+      "step": 3402
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.46198038212343e-05,
+      "loss": 1.6504,
+      "step": 3403
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.460642519612422e-05,
+      "loss": 1.9848,
+      "step": 3404
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.459304657101416e-05,
+      "loss": 1.8855,
+      "step": 3405
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.457966794590408e-05,
+      "loss": 1.6972,
+      "step": 3406
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.4566289320794e-05,
+      "loss": 1.7446,
+      "step": 3407
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.455291069568393e-05,
+      "loss": 1.6748,
+      "step": 3408
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.453953207057385e-05,
+      "loss": 1.714,
+      "step": 3409
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.452615344546377e-05,
+      "loss": 1.6891,
+      "step": 3410
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.451277482035369e-05,
+      "loss": 1.8487,
+      "step": 3411
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.449939619524361e-05,
+      "loss": 1.6617,
+      "step": 3412
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.448601757013354e-05,
+      "loss": 1.7305,
+      "step": 3413
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.447263894502346e-05,
+      "loss": 1.5377,
+      "step": 3414
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.445926031991338e-05,
+      "loss": 1.6639,
+      "step": 3415
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.444588169480332e-05,
+      "loss": 1.8163,
+      "step": 3416
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.443250306969324e-05,
+      "loss": 1.5518,
+      "step": 3417
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.441912444458317e-05,
+      "loss": 1.5805,
+      "step": 3418
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.440574581947308e-05,
+      "loss": 1.669,
+      "step": 3419
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.4392367194363e-05,
+      "loss": 1.7413,
+      "step": 3420
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.437898856925293e-05,
+      "loss": 1.5641,
+      "step": 3421
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.436560994414285e-05,
+      "loss": 1.4335,
+      "step": 3422
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.435223131903277e-05,
+      "loss": 1.5071,
+      "step": 3423
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.43388526939227e-05,
+      "loss": 1.4218,
+      "step": 3424
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.432547406881262e-05,
+      "loss": 1.4582,
+      "step": 3425
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.431209544370254e-05,
+      "loss": 1.3045,
+      "step": 3426
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.429871681859246e-05,
+      "loss": 1.3833,
+      "step": 3427
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.428533819348238e-05,
+      "loss": 1.2697,
+      "step": 3428
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.427195956837232e-05,
+      "loss": 1.1695,
+      "step": 3429
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.425858094326224e-05,
+      "loss": 1.2625,
+      "step": 3430
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.424520231815216e-05,
+      "loss": 1.1171,
+      "step": 3431
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.423182369304209e-05,
+      "loss": 1.2983,
+      "step": 3432
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.421844506793201e-05,
+      "loss": 1.3083,
+      "step": 3433
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.420506644282193e-05,
+      "loss": 1.2198,
+      "step": 3434
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.419168781771185e-05,
+      "loss": 1.2506,
+      "step": 3435
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.417830919260177e-05,
+      "loss": 1.1471,
+      "step": 3436
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.41649305674917e-05,
+      "loss": 1.1827,
+      "step": 3437
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.415155194238163e-05,
+      "loss": 1.1344,
+      "step": 3438
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.413817331727156e-05,
+      "loss": 0.9086,
+      "step": 3439
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.412479469216148e-05,
+      "loss": 1.1477,
+      "step": 3440
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.41114160670514e-05,
+      "loss": 2.619,
+      "step": 3441
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.409803744194132e-05,
+      "loss": 2.3334,
+      "step": 3442
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.408465881683124e-05,
+      "loss": 2.1262,
+      "step": 3443
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.407128019172116e-05,
+      "loss": 1.7561,
+      "step": 3444
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.405790156661109e-05,
+      "loss": 1.9779,
+      "step": 3445
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.404452294150101e-05,
+      "loss": 1.9287,
+      "step": 3446
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.403114431639095e-05,
+      "loss": 2.0059,
+      "step": 3447
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.401776569128087e-05,
+      "loss": 1.9936,
+      "step": 3448
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.400438706617077e-05,
+      "loss": 1.7907,
+      "step": 3449
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.39910084410607e-05,
+      "loss": 1.8031,
+      "step": 3450
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.397762981595063e-05,
+      "loss": 1.6212,
+      "step": 3451
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.396425119084056e-05,
+      "loss": 1.7395,
+      "step": 3452
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.395087256573048e-05,
+      "loss": 2.0102,
+      "step": 3453
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.39374939406204e-05,
+      "loss": 1.6331,
+      "step": 3454
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.392411531551033e-05,
+      "loss": 1.8809,
+      "step": 3455
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.391073669040025e-05,
+      "loss": 1.7055,
+      "step": 3456
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.389735806529016e-05,
+      "loss": 1.7058,
+      "step": 3457
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.388397944018009e-05,
+      "loss": 1.8001,
+      "step": 3458
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.387060081507001e-05,
+      "loss": 1.7015,
+      "step": 3459
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.385722218995995e-05,
+      "loss": 1.8675,
+      "step": 3460
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.384384356484987e-05,
+      "loss": 1.4552,
+      "step": 3461
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.383046493973979e-05,
+      "loss": 1.6344,
+      "step": 3462
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.381708631462972e-05,
+      "loss": 1.5181,
+      "step": 3463
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.380370768951964e-05,
+      "loss": 1.6179,
+      "step": 3464
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.379032906440955e-05,
+      "loss": 1.6601,
+      "step": 3465
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.377695043929948e-05,
+      "loss": 1.4852,
+      "step": 3466
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.37635718141894e-05,
+      "loss": 1.6701,
+      "step": 3467
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.375019318907933e-05,
+      "loss": 1.6389,
+      "step": 3468
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.373681456396925e-05,
+      "loss": 1.7533,
+      "step": 3469
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.372343593885917e-05,
+      "loss": 1.5685,
+      "step": 3470
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.371005731374911e-05,
+      "loss": 1.5853,
+      "step": 3471
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.369667868863903e-05,
+      "loss": 1.6359,
+      "step": 3472
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.368330006352895e-05,
+      "loss": 1.6083,
+      "step": 3473
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.366992143841887e-05,
+      "loss": 1.3056,
+      "step": 3474
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.365654281330879e-05,
+      "loss": 1.368,
+      "step": 3475
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.364316418819872e-05,
+      "loss": 1.3634,
+      "step": 3476
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.362978556308864e-05,
+      "loss": 1.5304,
+      "step": 3477
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.361640693797856e-05,
+      "loss": 1.3203,
+      "step": 3478
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.36030283128685e-05,
+      "loss": 1.3069,
+      "step": 3479
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.35896496877584e-05,
+      "loss": 1.1076,
+      "step": 3480
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.357627106264833e-05,
+      "loss": 1.2318,
+      "step": 3481
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.356289243753825e-05,
+      "loss": 1.2316,
+      "step": 3482
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.354951381242817e-05,
+      "loss": 1.3811,
+      "step": 3483
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.353613518731811e-05,
+      "loss": 1.0581,
+      "step": 3484
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.352275656220803e-05,
+      "loss": 1.0946,
+      "step": 3485
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.350937793709796e-05,
+      "loss": 0.9879,
+      "step": 3486
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.349599931198788e-05,
+      "loss": 1.2301,
+      "step": 3487
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.348262068687779e-05,
+      "loss": 1.006,
+      "step": 3488
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.346924206176772e-05,
+      "loss": 1.0431,
+      "step": 3489
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.345586343665764e-05,
+      "loss": 0.9274,
+      "step": 3490
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.344248481154756e-05,
+      "loss": 2.289,
+      "step": 3491
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.34291061864375e-05,
+      "loss": 1.9655,
+      "step": 3492
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.341572756132741e-05,
+      "loss": 1.9582,
+      "step": 3493
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.340234893621735e-05,
+      "loss": 1.961,
+      "step": 3494
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.338897031110727e-05,
+      "loss": 1.8933,
+      "step": 3495
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.337559168599717e-05,
+      "loss": 1.7523,
+      "step": 3496
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.336221306088711e-05,
+      "loss": 1.8745,
+      "step": 3497
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.334883443577703e-05,
+      "loss": 1.8881,
+      "step": 3498
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.333545581066695e-05,
+      "loss": 1.9361,
+      "step": 3499
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.332207718555688e-05,
+      "loss": 1.8476,
+      "step": 3500
+    },
+    {
+      "epoch": 1.08,
+      "eval_bleu": 6.422611977338095e-18,
+      "eval_loss": 1.9935623407363892,
+      "eval_runtime": 2522.5708,
+      "eval_samples_per_second": 5.851,
+      "eval_steps_per_second": 0.731,
+      "step": 3500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.33086985604468e-05,
+      "loss": 1.625,
+      "step": 3501
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.329531993533674e-05,
+      "loss": 1.8765,
+      "step": 3502
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.328194131022664e-05,
+      "loss": 2.0112,
+      "step": 3503
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.326856268511656e-05,
+      "loss": 1.7005,
+      "step": 3504
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.32551840600065e-05,
+      "loss": 1.83,
+      "step": 3505
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.324180543489642e-05,
+      "loss": 1.5362,
+      "step": 3506
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.322842680978635e-05,
+      "loss": 1.7156,
+      "step": 3507
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.321504818467627e-05,
+      "loss": 1.953,
+      "step": 3508
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.320166955956619e-05,
+      "loss": 1.5919,
+      "step": 3509
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.318829093445612e-05,
+      "loss": 1.654,
+      "step": 3510
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.317491230934603e-05,
+      "loss": 1.5903,
+      "step": 3511
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.316153368423595e-05,
+      "loss": 1.6345,
+      "step": 3512
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.314815505912588e-05,
+      "loss": 1.6496,
+      "step": 3513
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.31347764340158e-05,
+      "loss": 1.545,
+      "step": 3514
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.312139780890574e-05,
+      "loss": 1.3926,
+      "step": 3515
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.310801918379566e-05,
+      "loss": 1.4143,
+      "step": 3516
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.309464055868558e-05,
+      "loss": 1.7044,
+      "step": 3517
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.308126193357551e-05,
+      "loss": 1.5066,
+      "step": 3518
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.306788330846542e-05,
+      "loss": 1.3736,
+      "step": 3519
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.305450468335535e-05,
+      "loss": 1.7009,
+      "step": 3520
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.304112605824527e-05,
+      "loss": 1.599,
+      "step": 3521
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.302774743313519e-05,
+      "loss": 1.5302,
+      "step": 3522
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.301436880802512e-05,
+      "loss": 1.465,
+      "step": 3523
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.300099018291504e-05,
+      "loss": 1.2783,
+      "step": 3524
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.298761155780496e-05,
+      "loss": 1.4374,
+      "step": 3525
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.29742329326949e-05,
+      "loss": 1.3912,
+      "step": 3526
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.29608543075848e-05,
+      "loss": 1.4083,
+      "step": 3527
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.294747568247474e-05,
+      "loss": 1.3609,
+      "step": 3528
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.293409705736466e-05,
+      "loss": 1.4112,
+      "step": 3529
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.292071843225458e-05,
+      "loss": 1.5141,
+      "step": 3530
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.290733980714451e-05,
+      "loss": 1.3093,
+      "step": 3531
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.289396118203443e-05,
+      "loss": 1.1888,
+      "step": 3532
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.288058255692436e-05,
+      "loss": 1.306,
+      "step": 3533
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.286720393181427e-05,
+      "loss": 1.3324,
+      "step": 3534
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.285382530670419e-05,
+      "loss": 1.1351,
+      "step": 3535
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.284044668159412e-05,
+      "loss": 1.3246,
+      "step": 3536
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.282706805648404e-05,
+      "loss": 1.1246,
+      "step": 3537
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.281368943137396e-05,
+      "loss": 1.1955,
+      "step": 3538
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.28003108062639e-05,
+      "loss": 0.9425,
+      "step": 3539
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.278693218115382e-05,
+      "loss": 1.0072,
+      "step": 3540
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.277355355604375e-05,
+      "loss": 2.595,
+      "step": 3541
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.276017493093366e-05,
+      "loss": 2.2134,
+      "step": 3542
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.274679630582358e-05,
+      "loss": 1.9827,
+      "step": 3543
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.273341768071351e-05,
+      "loss": 1.9936,
+      "step": 3544
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.272003905560343e-05,
+      "loss": 1.8675,
+      "step": 3545
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.270666043049335e-05,
+      "loss": 1.827,
+      "step": 3546
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.269328180538328e-05,
+      "loss": 1.8879,
+      "step": 3547
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.26799031802732e-05,
+      "loss": 1.7471,
+      "step": 3548
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.266652455516314e-05,
+      "loss": 1.7287,
+      "step": 3549
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.265314593005304e-05,
+      "loss": 1.8962,
+      "step": 3550
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.263976730494296e-05,
+      "loss": 1.847,
+      "step": 3551
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.26263886798329e-05,
+      "loss": 1.7126,
+      "step": 3552
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.261301005472282e-05,
+      "loss": 1.7955,
+      "step": 3553
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.259963142961275e-05,
+      "loss": 1.5637,
+      "step": 3554
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.258625280450267e-05,
+      "loss": 1.7568,
+      "step": 3555
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.257287417939259e-05,
+      "loss": 1.8505,
+      "step": 3556
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.255949555428251e-05,
+      "loss": 1.6135,
+      "step": 3557
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.254611692917243e-05,
+      "loss": 1.831,
+      "step": 3558
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.253273830406235e-05,
+      "loss": 1.798,
+      "step": 3559
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.251935967895229e-05,
+      "loss": 1.5696,
+      "step": 3560
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.25059810538422e-05,
+      "loss": 1.8642,
+      "step": 3561
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.249260242873214e-05,
+      "loss": 1.5776,
+      "step": 3562
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.247922380362206e-05,
+      "loss": 1.5093,
+      "step": 3563
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.246584517851198e-05,
+      "loss": 1.7214,
+      "step": 3564
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.24524665534019e-05,
+      "loss": 1.5011,
+      "step": 3565
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.243908792829182e-05,
+      "loss": 1.4839,
+      "step": 3566
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.242570930318175e-05,
+      "loss": 1.4926,
+      "step": 3567
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.241233067807167e-05,
+      "loss": 1.6961,
+      "step": 3568
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.239895205296159e-05,
+      "loss": 1.5327,
+      "step": 3569
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.238557342785153e-05,
+      "loss": 1.4287,
+      "step": 3570
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.237219480274145e-05,
+      "loss": 1.7331,
+      "step": 3571
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.235881617763137e-05,
+      "loss": 1.6544,
+      "step": 3572
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.234543755252129e-05,
+      "loss": 1.4158,
+      "step": 3573
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.23320589274112e-05,
+      "loss": 1.2452,
+      "step": 3574
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.231868030230114e-05,
+      "loss": 1.3434,
+      "step": 3575
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.230530167719106e-05,
+      "loss": 1.6862,
+      "step": 3576
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.229192305208098e-05,
+      "loss": 1.5695,
+      "step": 3577
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.227854442697091e-05,
+      "loss": 1.4592,
+      "step": 3578
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.226516580186083e-05,
+      "loss": 1.3379,
+      "step": 3579
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.225178717675075e-05,
+      "loss": 1.2623,
+      "step": 3580
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.223840855164067e-05,
+      "loss": 1.411,
+      "step": 3581
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.22250299265306e-05,
+      "loss": 1.2722,
+      "step": 3582
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.221165130142053e-05,
+      "loss": 1.1336,
+      "step": 3583
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.219827267631045e-05,
+      "loss": 1.1535,
+      "step": 3584
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.218489405120037e-05,
+      "loss": 1.1142,
+      "step": 3585
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.21715154260903e-05,
+      "loss": 1.0074,
+      "step": 3586
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.215813680098022e-05,
+      "loss": 1.0652,
+      "step": 3587
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.214475817587014e-05,
+      "loss": 1.0091,
+      "step": 3588
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.213137955076006e-05,
+      "loss": 1.0737,
+      "step": 3589
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.211800092564998e-05,
+      "loss": 1.2744,
+      "step": 3590
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.210462230053991e-05,
+      "loss": 2.4603,
+      "step": 3591
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.209124367542983e-05,
+      "loss": 2.0742,
+      "step": 3592
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.207786505031975e-05,
+      "loss": 2.1763,
+      "step": 3593
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.206448642520969e-05,
+      "loss": 1.9028,
+      "step": 3594
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.205110780009961e-05,
+      "loss": 1.8183,
+      "step": 3595
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.203772917498953e-05,
+      "loss": 2.0166,
+      "step": 3596
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.202435054987945e-05,
+      "loss": 1.8952,
+      "step": 3597
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.201097192476937e-05,
+      "loss": 1.8261,
+      "step": 3598
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.19975932996593e-05,
+      "loss": 1.8574,
+      "step": 3599
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.198421467454922e-05,
+      "loss": 1.6776,
+      "step": 3600
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.197083604943915e-05,
+      "loss": 1.8186,
+      "step": 3601
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.195745742432907e-05,
+      "loss": 1.5105,
+      "step": 3602
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.1944078799219e-05,
+      "loss": 1.923,
+      "step": 3603
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.193070017410891e-05,
+      "loss": 1.6954,
+      "step": 3604
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.191732154899883e-05,
+      "loss": 1.5828,
+      "step": 3605
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.190394292388875e-05,
+      "loss": 1.87,
+      "step": 3606
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.189056429877869e-05,
+      "loss": 1.5826,
+      "step": 3607
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.187718567366861e-05,
+      "loss": 1.8964,
+      "step": 3608
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.186380704855854e-05,
+      "loss": 1.8973,
+      "step": 3609
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.185042842344846e-05,
+      "loss": 1.8006,
+      "step": 3610
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.183704979833837e-05,
+      "loss": 1.3384,
+      "step": 3611
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.18236711732283e-05,
+      "loss": 1.543,
+      "step": 3612
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.181029254811822e-05,
+      "loss": 1.749,
+      "step": 3613
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.179691392300814e-05,
+      "loss": 1.5932,
+      "step": 3614
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.178353529789808e-05,
+      "loss": 1.6013,
+      "step": 3615
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.1770156672788e-05,
+      "loss": 1.6547,
+      "step": 3616
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.175677804767793e-05,
+      "loss": 1.8099,
+      "step": 3617
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.174339942256785e-05,
+      "loss": 1.4859,
+      "step": 3618
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.173002079745776e-05,
+      "loss": 1.5235,
+      "step": 3619
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.171664217234769e-05,
+      "loss": 1.4643,
+      "step": 3620
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.170326354723761e-05,
+      "loss": 1.4373,
+      "step": 3621
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.168988492212754e-05,
+      "loss": 1.6264,
+      "step": 3622
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.167650629701746e-05,
+      "loss": 1.5845,
+      "step": 3623
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.166312767190738e-05,
+      "loss": 1.3335,
+      "step": 3624
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.164974904679732e-05,
+      "loss": 1.5507,
+      "step": 3625
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.163637042168724e-05,
+      "loss": 1.372,
+      "step": 3626
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.162299179657714e-05,
+      "loss": 1.3533,
+      "step": 3627
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.160961317146708e-05,
+      "loss": 1.3862,
+      "step": 3628
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.1596234546357e-05,
+      "loss": 1.4147,
+      "step": 3629
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.158285592124693e-05,
+      "loss": 1.2665,
+      "step": 3630
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.156947729613685e-05,
+      "loss": 1.2688,
+      "step": 3631
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.155609867102677e-05,
+      "loss": 1.5247,
+      "step": 3632
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.15427200459167e-05,
+      "loss": 1.2387,
+      "step": 3633
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.152934142080661e-05,
+      "loss": 1.15,
+      "step": 3634
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.151596279569654e-05,
+      "loss": 1.0845,
+      "step": 3635
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.150258417058646e-05,
+      "loss": 0.9647,
+      "step": 3636
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.148920554547638e-05,
+      "loss": 1.0552,
+      "step": 3637
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.147582692036632e-05,
+      "loss": 1.0103,
+      "step": 3638
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.146244829525624e-05,
+      "loss": 0.9823,
+      "step": 3639
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.144906967014616e-05,
+      "loss": 1.0174,
+      "step": 3640
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.143569104503609e-05,
+      "loss": 2.7076,
+      "step": 3641
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.1422312419926e-05,
+      "loss": 2.4586,
+      "step": 3642
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.140893379481593e-05,
+      "loss": 1.8432,
+      "step": 3643
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.139555516970585e-05,
+      "loss": 1.7708,
+      "step": 3644
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.138217654459577e-05,
+      "loss": 2.0358,
+      "step": 3645
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.13687979194857e-05,
+      "loss": 1.9686,
+      "step": 3646
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.135541929437562e-05,
+      "loss": 1.9742,
+      "step": 3647
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.134204066926556e-05,
+      "loss": 1.9708,
+      "step": 3648
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.132866204415548e-05,
+      "loss": 1.9245,
+      "step": 3649
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.131528341904538e-05,
+      "loss": 1.7436,
+      "step": 3650
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.130190479393532e-05,
+      "loss": 1.7869,
+      "step": 3651
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.128852616882524e-05,
+      "loss": 1.8676,
+      "step": 3652
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.127514754371516e-05,
+      "loss": 1.9246,
+      "step": 3653
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.126176891860509e-05,
+      "loss": 1.8199,
+      "step": 3654
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.124839029349501e-05,
+      "loss": 1.6085,
+      "step": 3655
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.123501166838494e-05,
+      "loss": 1.7649,
+      "step": 3656
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.122163304327486e-05,
+      "loss": 1.7278,
+      "step": 3657
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.120825441816477e-05,
+      "loss": 1.7385,
+      "step": 3658
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.11948757930547e-05,
+      "loss": 1.6192,
+      "step": 3659
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.118149716794462e-05,
+      "loss": 1.5882,
+      "step": 3660
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.116811854283454e-05,
+      "loss": 1.6911,
+      "step": 3661
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.115473991772448e-05,
+      "loss": 1.459,
+      "step": 3662
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.11413612926144e-05,
+      "loss": 1.7016,
+      "step": 3663
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.112798266750433e-05,
+      "loss": 1.6119,
+      "step": 3664
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.111460404239424e-05,
+      "loss": 1.631,
+      "step": 3665
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.110122541728416e-05,
+      "loss": 1.6168,
+      "step": 3666
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.108784679217409e-05,
+      "loss": 1.5913,
+      "step": 3667
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.107446816706401e-05,
+      "loss": 1.8855,
+      "step": 3668
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.106108954195395e-05,
+      "loss": 1.3873,
+      "step": 3669
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.104771091684387e-05,
+      "loss": 1.4289,
+      "step": 3670
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.103433229173379e-05,
+      "loss": 1.3713,
+      "step": 3671
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.102095366662372e-05,
+      "loss": 1.5665,
+      "step": 3672
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.100757504151363e-05,
+      "loss": 1.6158,
+      "step": 3673
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.099419641640355e-05,
+      "loss": 1.6136,
+      "step": 3674
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.098081779129348e-05,
+      "loss": 1.4531,
+      "step": 3675
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.09674391661834e-05,
+      "loss": 1.3973,
+      "step": 3676
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.095406054107333e-05,
+      "loss": 1.2648,
+      "step": 3677
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.094068191596325e-05,
+      "loss": 1.2391,
+      "step": 3678
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.092730329085317e-05,
+      "loss": 1.2409,
+      "step": 3679
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.09139246657431e-05,
+      "loss": 1.4762,
+      "step": 3680
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.090054604063301e-05,
+      "loss": 1.2626,
+      "step": 3681
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.088716741552295e-05,
+      "loss": 1.3126,
+      "step": 3682
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.087378879041287e-05,
+      "loss": 1.1945,
+      "step": 3683
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.086041016530279e-05,
+      "loss": 1.3115,
+      "step": 3684
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.084703154019272e-05,
+      "loss": 1.1147,
+      "step": 3685
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.083365291508264e-05,
+      "loss": 1.1051,
+      "step": 3686
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.082027428997256e-05,
+      "loss": 1.208,
+      "step": 3687
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.080689566486248e-05,
+      "loss": 1.054,
+      "step": 3688
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.07935170397524e-05,
+      "loss": 1.0181,
+      "step": 3689
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.078013841464233e-05,
+      "loss": 1.1164,
+      "step": 3690
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.076675978953225e-05,
+      "loss": 2.5391,
+      "step": 3691
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.075338116442217e-05,
+      "loss": 2.1871,
+      "step": 3692
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.07400025393121e-05,
+      "loss": 1.9153,
+      "step": 3693
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.072662391420203e-05,
+      "loss": 1.8091,
+      "step": 3694
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.071324528909195e-05,
+      "loss": 1.8233,
+      "step": 3695
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.069986666398187e-05,
+      "loss": 1.7684,
+      "step": 3696
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.068648803887179e-05,
+      "loss": 1.9564,
+      "step": 3697
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.067310941376172e-05,
+      "loss": 1.8686,
+      "step": 3698
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.065973078865164e-05,
+      "loss": 1.7469,
+      "step": 3699
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.064635216354156e-05,
+      "loss": 1.8676,
+      "step": 3700
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.06329735384315e-05,
+      "loss": 1.9391,
+      "step": 3701
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.061959491332141e-05,
+      "loss": 1.5794,
+      "step": 3702
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.060621628821135e-05,
+      "loss": 2.0026,
+      "step": 3703
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.059283766310125e-05,
+      "loss": 1.8837,
+      "step": 3704
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.057945903799117e-05,
+      "loss": 1.9495,
+      "step": 3705
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.056608041288111e-05,
+      "loss": 1.7796,
+      "step": 3706
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.055270178777103e-05,
+      "loss": 1.8742,
+      "step": 3707
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.053932316266095e-05,
+      "loss": 1.7422,
+      "step": 3708
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.052594453755088e-05,
+      "loss": 1.5089,
+      "step": 3709
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.05125659124408e-05,
+      "loss": 1.6087,
+      "step": 3710
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.049918728733073e-05,
+      "loss": 1.7807,
+      "step": 3711
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.048580866222064e-05,
+      "loss": 1.7116,
+      "step": 3712
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.047243003711056e-05,
+      "loss": 1.5187,
+      "step": 3713
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.04590514120005e-05,
+      "loss": 1.6112,
+      "step": 3714
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.044567278689041e-05,
+      "loss": 1.4249,
+      "step": 3715
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.043229416178035e-05,
+      "loss": 1.5797,
+      "step": 3716
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.041891553667027e-05,
+      "loss": 1.738,
+      "step": 3717
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.040553691156019e-05,
+      "loss": 1.3908,
+      "step": 3718
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.039215828645011e-05,
+      "loss": 1.474,
+      "step": 3719
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.037877966134003e-05,
+      "loss": 1.4325,
+      "step": 3720
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.036540103622995e-05,
+      "loss": 1.5445,
+      "step": 3721
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.035202241111988e-05,
+      "loss": 1.4659,
+      "step": 3722
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.03386437860098e-05,
+      "loss": 1.75,
+      "step": 3723
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.032526516089974e-05,
+      "loss": 1.5944,
+      "step": 3724
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.031188653578966e-05,
+      "loss": 1.5564,
+      "step": 3725
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.029850791067958e-05,
+      "loss": 1.5431,
+      "step": 3726
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.02851292855695e-05,
+      "loss": 1.2906,
+      "step": 3727
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.027175066045942e-05,
+      "loss": 1.3058,
+      "step": 3728
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.025837203534935e-05,
+      "loss": 1.1231,
+      "step": 3729
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.024499341023927e-05,
+      "loss": 1.2091,
+      "step": 3730
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.023161478512919e-05,
+      "loss": 1.1848,
+      "step": 3731
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.021823616001912e-05,
+      "loss": 1.2071,
+      "step": 3732
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.020485753490904e-05,
+      "loss": 1.1301,
+      "step": 3733
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.019147890979896e-05,
+      "loss": 1.2707,
+      "step": 3734
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.017810028468888e-05,
+      "loss": 1.0099,
+      "step": 3735
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.01647216595788e-05,
+      "loss": 0.9885,
+      "step": 3736
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.015134303446874e-05,
+      "loss": 1.0641,
+      "step": 3737
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.013796440935866e-05,
+      "loss": 1.125,
+      "step": 3738
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.012458578424858e-05,
+      "loss": 0.9535,
+      "step": 3739
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.011120715913851e-05,
+      "loss": 0.8442,
+      "step": 3740
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.009782853402843e-05,
+      "loss": 2.467,
+      "step": 3741
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.008444990891834e-05,
+      "loss": 2.1439,
+      "step": 3742
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.007107128380827e-05,
+      "loss": 2.0769,
+      "step": 3743
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.005769265869819e-05,
+      "loss": 1.8689,
+      "step": 3744
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.004431403358812e-05,
+      "loss": 1.9501,
+      "step": 3745
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.003093540847804e-05,
+      "loss": 1.8577,
+      "step": 3746
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.001755678336796e-05,
+      "loss": 1.9013,
+      "step": 3747
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.00041781582579e-05,
+      "loss": 1.8259,
+      "step": 3748
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.999079953314782e-05,
+      "loss": 2.0053,
+      "step": 3749
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.997742090803774e-05,
+      "loss": 1.9099,
+      "step": 3750
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.996404228292766e-05,
+      "loss": 1.9532,
+      "step": 3751
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.995066365781758e-05,
+      "loss": 1.7262,
+      "step": 3752
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.993728503270751e-05,
+      "loss": 1.8179,
+      "step": 3753
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.992390640759743e-05,
+      "loss": 1.8129,
+      "step": 3754
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.991052778248735e-05,
+      "loss": 1.9287,
+      "step": 3755
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.989714915737728e-05,
+      "loss": 1.4705,
+      "step": 3756
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.98837705322672e-05,
+      "loss": 1.7783,
+      "step": 3757
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.987039190715712e-05,
+      "loss": 1.713,
+      "step": 3758
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.985701328204704e-05,
+      "loss": 1.8053,
+      "step": 3759
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.984363465693696e-05,
+      "loss": 1.707,
+      "step": 3760
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.98302560318269e-05,
+      "loss": 1.6951,
+      "step": 3761
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.981687740671682e-05,
+      "loss": 1.8989,
+      "step": 3762
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.980349878160675e-05,
+      "loss": 1.6564,
+      "step": 3763
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.979012015649667e-05,
+      "loss": 1.6415,
+      "step": 3764
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.977674153138659e-05,
+      "loss": 1.808,
+      "step": 3765
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.976336290627651e-05,
+      "loss": 1.5876,
+      "step": 3766
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.974998428116643e-05,
+      "loss": 1.4909,
+      "step": 3767
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.973660565605635e-05,
+      "loss": 1.6669,
+      "step": 3768
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.972322703094628e-05,
+      "loss": 1.5924,
+      "step": 3769
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.97098484058362e-05,
+      "loss": 1.6164,
+      "step": 3770
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.969646978072614e-05,
+      "loss": 1.4926,
+      "step": 3771
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.968309115561606e-05,
+      "loss": 1.4484,
+      "step": 3772
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.966971253050596e-05,
+      "loss": 1.4279,
+      "step": 3773
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.96563339053959e-05,
+      "loss": 1.6125,
+      "step": 3774
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.964295528028582e-05,
+      "loss": 1.3927,
+      "step": 3775
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.962957665517574e-05,
+      "loss": 1.4135,
+      "step": 3776
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.961619803006567e-05,
+      "loss": 1.4861,
+      "step": 3777
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.960281940495559e-05,
+      "loss": 1.4735,
+      "step": 3778
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.958944077984552e-05,
+      "loss": 1.2555,
+      "step": 3779
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.957606215473544e-05,
+      "loss": 1.395,
+      "step": 3780
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.956268352962535e-05,
+      "loss": 1.3336,
+      "step": 3781
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.954930490451528e-05,
+      "loss": 1.2661,
+      "step": 3782
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.95359262794052e-05,
+      "loss": 1.2704,
+      "step": 3783
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.952254765429514e-05,
+      "loss": 1.063,
+      "step": 3784
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.950916902918506e-05,
+      "loss": 1.034,
+      "step": 3785
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.949579040407498e-05,
+      "loss": 1.1,
+      "step": 3786
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.948241177896491e-05,
+      "loss": 1.0238,
+      "step": 3787
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.946903315385483e-05,
+      "loss": 1.2642,
+      "step": 3788
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.945565452874474e-05,
+      "loss": 0.9894,
+      "step": 3789
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.944227590363467e-05,
+      "loss": 0.9747,
+      "step": 3790
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.942889727852459e-05,
+      "loss": 2.2325,
+      "step": 3791
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.941551865341453e-05,
+      "loss": 2.2696,
+      "step": 3792
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.940214002830445e-05,
+      "loss": 2.0414,
+      "step": 3793
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.938876140319437e-05,
+      "loss": 1.7688,
+      "step": 3794
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.93753827780843e-05,
+      "loss": 1.9779,
+      "step": 3795
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.93620041529742e-05,
+      "loss": 1.8752,
+      "step": 3796
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.934862552786414e-05,
+      "loss": 1.8092,
+      "step": 3797
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.933524690275406e-05,
+      "loss": 1.8866,
+      "step": 3798
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.932186827764398e-05,
+      "loss": 1.7143,
+      "step": 3799
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.930848965253391e-05,
+      "loss": 1.6322,
+      "step": 3800
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.929511102742383e-05,
+      "loss": 1.8098,
+      "step": 3801
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.928173240231375e-05,
+      "loss": 1.7849,
+      "step": 3802
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.926835377720369e-05,
+      "loss": 1.5298,
+      "step": 3803
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.925497515209359e-05,
+      "loss": 1.6735,
+      "step": 3804
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.924159652698353e-05,
+      "loss": 1.8573,
+      "step": 3805
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.922821790187345e-05,
+      "loss": 1.6489,
+      "step": 3806
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.921483927676337e-05,
+      "loss": 1.4462,
+      "step": 3807
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.92014606516533e-05,
+      "loss": 1.5522,
+      "step": 3808
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.918808202654322e-05,
+      "loss": 1.6666,
+      "step": 3809
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.917470340143314e-05,
+      "loss": 1.6363,
+      "step": 3810
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.916132477632307e-05,
+      "loss": 1.6282,
+      "step": 3811
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.914794615121298e-05,
+      "loss": 1.5037,
+      "step": 3812
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.913456752610291e-05,
+      "loss": 1.6278,
+      "step": 3813
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.912118890099283e-05,
+      "loss": 1.4632,
+      "step": 3814
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.910781027588275e-05,
+      "loss": 1.7531,
+      "step": 3815
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.909443165077269e-05,
+      "loss": 1.3725,
+      "step": 3816
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.908105302566261e-05,
+      "loss": 1.6424,
+      "step": 3817
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.906767440055254e-05,
+      "loss": 1.549,
+      "step": 3818
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.905429577544246e-05,
+      "loss": 1.6238,
+      "step": 3819
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.904091715033237e-05,
+      "loss": 1.5274,
+      "step": 3820
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.90275385252223e-05,
+      "loss": 1.5699,
+      "step": 3821
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.901415990011222e-05,
+      "loss": 1.3907,
+      "step": 3822
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.900078127500214e-05,
+      "loss": 1.5397,
+      "step": 3823
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.898740264989207e-05,
+      "loss": 1.3786,
+      "step": 3824
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.8974024024782e-05,
+      "loss": 1.2796,
+      "step": 3825
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.896064539967193e-05,
+      "loss": 1.247,
+      "step": 3826
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.894726677456183e-05,
+      "loss": 1.3213,
+      "step": 3827
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.893388814945175e-05,
+      "loss": 1.2086,
+      "step": 3828
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.892050952434169e-05,
+      "loss": 1.262,
+      "step": 3829
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.890713089923161e-05,
+      "loss": 1.219,
+      "step": 3830
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.889375227412154e-05,
+      "loss": 1.3084,
+      "step": 3831
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.888037364901146e-05,
+      "loss": 1.3866,
+      "step": 3832
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.886699502390138e-05,
+      "loss": 1.0628,
+      "step": 3833
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.885361639879131e-05,
+      "loss": 1.1281,
+      "step": 3834
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.884023777368122e-05,
+      "loss": 1.1998,
+      "step": 3835
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.882685914857114e-05,
+      "loss": 0.9696,
+      "step": 3836
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.881348052346107e-05,
+      "loss": 0.9658,
+      "step": 3837
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.8800101898351e-05,
+      "loss": 0.958,
+      "step": 3838
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.878672327324093e-05,
+      "loss": 1.0633,
+      "step": 3839
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.877334464813085e-05,
+      "loss": 0.985,
+      "step": 3840
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.875996602302077e-05,
+      "loss": 2.3924,
+      "step": 3841
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.87465873979107e-05,
+      "loss": 2.167,
+      "step": 3842
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.873320877280061e-05,
+      "loss": 1.7807,
+      "step": 3843
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.871983014769054e-05,
+      "loss": 2.044,
+      "step": 3844
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.870645152258046e-05,
+      "loss": 1.7524,
+      "step": 3845
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.869307289747038e-05,
+      "loss": 1.7919,
+      "step": 3846
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.867969427236032e-05,
+      "loss": 1.7413,
+      "step": 3847
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.866631564725024e-05,
+      "loss": 2.0428,
+      "step": 3848
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.865293702214016e-05,
+      "loss": 1.6582,
+      "step": 3849
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.863955839703008e-05,
+      "loss": 1.5705,
+      "step": 3850
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.862617977192e-05,
+      "loss": 1.624,
+      "step": 3851
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.861280114680993e-05,
+      "loss": 1.7364,
+      "step": 3852
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.859942252169985e-05,
+      "loss": 1.6044,
+      "step": 3853
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.858604389658977e-05,
+      "loss": 1.5125,
+      "step": 3854
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.85726652714797e-05,
+      "loss": 1.7713,
+      "step": 3855
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.855928664636962e-05,
+      "loss": 1.7616,
+      "step": 3856
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.854590802125954e-05,
+      "loss": 1.7061,
+      "step": 3857
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.853252939614946e-05,
+      "loss": 1.6212,
+      "step": 3858
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.851915077103938e-05,
+      "loss": 1.557,
+      "step": 3859
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.850577214592932e-05,
+      "loss": 1.6632,
+      "step": 3860
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.849239352081924e-05,
+      "loss": 1.511,
+      "step": 3861
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.847901489570916e-05,
+      "loss": 1.5483,
+      "step": 3862
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.846563627059909e-05,
+      "loss": 1.512,
+      "step": 3863
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.845225764548901e-05,
+      "loss": 1.4184,
+      "step": 3864
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.843887902037894e-05,
+      "loss": 1.6158,
+      "step": 3865
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.842550039526885e-05,
+      "loss": 1.463,
+      "step": 3866
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.841212177015877e-05,
+      "loss": 1.4899,
+      "step": 3867
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.83987431450487e-05,
+      "loss": 1.5327,
+      "step": 3868
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.838536451993862e-05,
+      "loss": 1.5547,
+      "step": 3869
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.837198589482854e-05,
+      "loss": 1.5479,
+      "step": 3870
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.835860726971848e-05,
+      "loss": 1.4727,
+      "step": 3871
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.83452286446084e-05,
+      "loss": 1.4835,
+      "step": 3872
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.833185001949833e-05,
+      "loss": 1.4421,
+      "step": 3873
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.831847139438824e-05,
+      "loss": 1.5104,
+      "step": 3874
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.830509276927816e-05,
+      "loss": 1.4124,
+      "step": 3875
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.829171414416809e-05,
+      "loss": 1.2925,
+      "step": 3876
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.827833551905801e-05,
+      "loss": 1.2607,
+      "step": 3877
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.826495689394794e-05,
+      "loss": 1.3899,
+      "step": 3878
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.825157826883786e-05,
+      "loss": 1.4125,
+      "step": 3879
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.823819964372778e-05,
+      "loss": 1.2584,
+      "step": 3880
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.82248210186177e-05,
+      "loss": 1.235,
+      "step": 3881
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.821144239350762e-05,
+      "loss": 1.2769,
+      "step": 3882
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.819806376839754e-05,
+      "loss": 1.3414,
+      "step": 3883
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.818468514328748e-05,
+      "loss": 1.2036,
+      "step": 3884
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.81713065181774e-05,
+      "loss": 1.0108,
+      "step": 3885
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.815792789306733e-05,
+      "loss": 1.1061,
+      "step": 3886
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.814454926795725e-05,
+      "loss": 0.9796,
+      "step": 3887
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.813117064284717e-05,
+      "loss": 0.9196,
+      "step": 3888
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.811779201773709e-05,
+      "loss": 0.9343,
+      "step": 3889
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.810441339262701e-05,
+      "loss": 0.8839,
+      "step": 3890
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.809103476751693e-05,
+      "loss": 2.3564,
+      "step": 3891
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.807765614240686e-05,
+      "loss": 2.3787,
+      "step": 3892
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.806427751729678e-05,
+      "loss": 1.889,
+      "step": 3893
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.805089889218672e-05,
+      "loss": 1.9103,
+      "step": 3894
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.803752026707664e-05,
+      "loss": 1.8241,
+      "step": 3895
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.802414164196656e-05,
+      "loss": 1.6395,
+      "step": 3896
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.801076301685648e-05,
+      "loss": 1.6861,
+      "step": 3897
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.79973843917464e-05,
+      "loss": 1.6706,
+      "step": 3898
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.798400576663633e-05,
+      "loss": 1.6571,
+      "step": 3899
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.797062714152625e-05,
+      "loss": 1.698,
+      "step": 3900
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.795724851641617e-05,
+      "loss": 1.8322,
+      "step": 3901
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.79438698913061e-05,
+      "loss": 1.8502,
+      "step": 3902
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.793049126619603e-05,
+      "loss": 1.6716,
+      "step": 3903
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.791711264108593e-05,
+      "loss": 1.4884,
+      "step": 3904
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.790373401597587e-05,
+      "loss": 1.6964,
+      "step": 3905
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.789035539086579e-05,
+      "loss": 1.5954,
+      "step": 3906
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.787697676575572e-05,
+      "loss": 1.7879,
+      "step": 3907
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.786359814064564e-05,
+      "loss": 1.5695,
+      "step": 3908
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.785021951553556e-05,
+      "loss": 1.6168,
+      "step": 3909
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.783684089042549e-05,
+      "loss": 1.6079,
+      "step": 3910
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.782346226531541e-05,
+      "loss": 1.5896,
+      "step": 3911
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.781008364020533e-05,
+      "loss": 1.6001,
+      "step": 3912
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.779670501509525e-05,
+      "loss": 1.4912,
+      "step": 3913
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.778332638998517e-05,
+      "loss": 1.5004,
+      "step": 3914
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.77699477648751e-05,
+      "loss": 1.5304,
+      "step": 3915
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.775656913976503e-05,
+      "loss": 1.4953,
+      "step": 3916
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.774319051465495e-05,
+      "loss": 1.7326,
+      "step": 3917
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.772981188954488e-05,
+      "loss": 1.6504,
+      "step": 3918
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.77164332644348e-05,
+      "loss": 1.4203,
+      "step": 3919
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.770305463932472e-05,
+      "loss": 1.2771,
+      "step": 3920
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.768967601421464e-05,
+      "loss": 1.4349,
+      "step": 3921
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.767629738910456e-05,
+      "loss": 1.4878,
+      "step": 3922
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.766291876399449e-05,
+      "loss": 1.235,
+      "step": 3923
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.764954013888441e-05,
+      "loss": 1.4049,
+      "step": 3924
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.763616151377435e-05,
+      "loss": 1.373,
+      "step": 3925
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.762278288866427e-05,
+      "loss": 1.3888,
+      "step": 3926
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.760940426355417e-05,
+      "loss": 1.3634,
+      "step": 3927
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.75960256384441e-05,
+      "loss": 1.411,
+      "step": 3928
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.758264701333403e-05,
+      "loss": 1.3665,
+      "step": 3929
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.756926838822395e-05,
+      "loss": 1.1057,
+      "step": 3930
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.755588976311388e-05,
+      "loss": 1.2614,
+      "step": 3931
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.75425111380038e-05,
+      "loss": 1.165,
+      "step": 3932
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.752913251289373e-05,
+      "loss": 1.1367,
+      "step": 3933
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.751575388778365e-05,
+      "loss": 1.0102,
+      "step": 3934
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.750237526267356e-05,
+      "loss": 1.1651,
+      "step": 3935
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.74889966375635e-05,
+      "loss": 1.1283,
+      "step": 3936
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.747561801245341e-05,
+      "loss": 1.0509,
+      "step": 3937
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.746223938734333e-05,
+      "loss": 1.0293,
+      "step": 3938
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.744886076223327e-05,
+      "loss": 0.9486,
+      "step": 3939
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.743548213712319e-05,
+      "loss": 1.0651,
+      "step": 3940
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.742210351201312e-05,
+      "loss": 2.3708,
+      "step": 3941
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.740872488690304e-05,
+      "loss": 2.2129,
+      "step": 3942
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.739534626179295e-05,
+      "loss": 1.9224,
+      "step": 3943
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.738196763668288e-05,
+      "loss": 1.8532,
+      "step": 3944
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.73685890115728e-05,
+      "loss": 1.799,
+      "step": 3945
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.735521038646273e-05,
+      "loss": 1.8451,
+      "step": 3946
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.734183176135265e-05,
+      "loss": 1.6765,
+      "step": 3947
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.732845313624257e-05,
+      "loss": 1.5864,
+      "step": 3948
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.731507451113251e-05,
+      "loss": 1.6457,
+      "step": 3949
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.730169588602243e-05,
+      "loss": 1.6969,
+      "step": 3950
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.728831726091233e-05,
+      "loss": 1.9347,
+      "step": 3951
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.727493863580227e-05,
+      "loss": 1.7076,
+      "step": 3952
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.726156001069219e-05,
+      "loss": 1.7398,
+      "step": 3953
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.724818138558212e-05,
+      "loss": 1.841,
+      "step": 3954
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.723480276047204e-05,
+      "loss": 1.6118,
+      "step": 3955
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.722142413536196e-05,
+      "loss": 1.8025,
+      "step": 3956
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.72080455102519e-05,
+      "loss": 1.7037,
+      "step": 3957
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.71946668851418e-05,
+      "loss": 1.7646,
+      "step": 3958
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.718128826003174e-05,
+      "loss": 1.5907,
+      "step": 3959
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.716790963492166e-05,
+      "loss": 1.4613,
+      "step": 3960
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.715453100981158e-05,
+      "loss": 1.748,
+      "step": 3961
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.714115238470151e-05,
+      "loss": 1.5065,
+      "step": 3962
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.712777375959143e-05,
+      "loss": 1.6127,
+      "step": 3963
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.711439513448135e-05,
+      "loss": 1.4782,
+      "step": 3964
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.710101650937128e-05,
+      "loss": 1.5867,
+      "step": 3965
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.708763788426119e-05,
+      "loss": 1.5873,
+      "step": 3966
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.707425925915112e-05,
+      "loss": 1.7518,
+      "step": 3967
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.706088063404104e-05,
+      "loss": 1.6074,
+      "step": 3968
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.704750200893096e-05,
+      "loss": 1.569,
+      "step": 3969
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.70341233838209e-05,
+      "loss": 1.4372,
+      "step": 3970
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.702074475871082e-05,
+      "loss": 1.5637,
+      "step": 3971
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.700736613360074e-05,
+      "loss": 1.5151,
+      "step": 3972
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.699398750849067e-05,
+      "loss": 1.1357,
+      "step": 3973
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.698060888338058e-05,
+      "loss": 1.5766,
+      "step": 3974
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.696723025827051e-05,
+      "loss": 1.3439,
+      "step": 3975
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.695385163316043e-05,
+      "loss": 1.271,
+      "step": 3976
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.694047300805035e-05,
+      "loss": 1.3477,
+      "step": 3977
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.692709438294028e-05,
+      "loss": 1.2862,
+      "step": 3978
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.69137157578302e-05,
+      "loss": 1.2262,
+      "step": 3979
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.690033713272014e-05,
+      "loss": 1.2263,
+      "step": 3980
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.688695850761004e-05,
+      "loss": 1.4287,
+      "step": 3981
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.687357988249996e-05,
+      "loss": 1.291,
+      "step": 3982
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.68602012573899e-05,
+      "loss": 1.4482,
+      "step": 3983
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.684682263227982e-05,
+      "loss": 1.2709,
+      "step": 3984
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.683344400716974e-05,
+      "loss": 1.0399,
+      "step": 3985
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.682006538205967e-05,
+      "loss": 1.0677,
+      "step": 3986
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.680668675694959e-05,
+      "loss": 1.0082,
+      "step": 3987
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.679330813183952e-05,
+      "loss": 0.8435,
+      "step": 3988
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.677992950672943e-05,
+      "loss": 1.0882,
+      "step": 3989
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.676655088161935e-05,
+      "loss": 0.8359,
+      "step": 3990
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.675317225650928e-05,
+      "loss": 2.3178,
+      "step": 3991
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.67397936313992e-05,
+      "loss": 1.8805,
+      "step": 3992
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.672641500628914e-05,
+      "loss": 2.0383,
+      "step": 3993
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.671303638117906e-05,
+      "loss": 1.8813,
+      "step": 3994
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.669965775606898e-05,
+      "loss": 1.9402,
+      "step": 3995
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.668627913095891e-05,
+      "loss": 1.7933,
+      "step": 3996
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.667290050584882e-05,
+      "loss": 1.8508,
+      "step": 3997
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.665952188073874e-05,
+      "loss": 1.7501,
+      "step": 3998
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.664614325562867e-05,
+      "loss": 1.5654,
+      "step": 3999
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.663276463051859e-05,
+      "loss": 1.683,
+      "step": 4000
+    },
+    {
+      "epoch": 1.23,
+      "eval_bleu": 1.7228539083708808e-14,
+      "eval_loss": 1.9307533502578735,
+      "eval_runtime": 2539.8971,
+      "eval_samples_per_second": 5.811,
+      "eval_steps_per_second": 0.726,
+      "step": 4000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.661938600540852e-05,
+      "loss": 1.91,
+      "step": 4001
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.660600738029844e-05,
+      "loss": 1.6979,
+      "step": 4002
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.659262875518836e-05,
+      "loss": 1.7522,
+      "step": 4003
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.65792501300783e-05,
+      "loss": 1.6142,
+      "step": 4004
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.65658715049682e-05,
+      "loss": 1.7319,
+      "step": 4005
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.655249287985812e-05,
+      "loss": 1.5604,
+      "step": 4006
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.653911425474806e-05,
+      "loss": 1.6501,
+      "step": 4007
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.652573562963798e-05,
+      "loss": 1.7592,
+      "step": 4008
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.651235700452791e-05,
+      "loss": 1.5616,
+      "step": 4009
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.649897837941783e-05,
+      "loss": 1.4783,
+      "step": 4010
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.648559975430775e-05,
+      "loss": 1.5287,
+      "step": 4011
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.647222112919767e-05,
+      "loss": 1.6315,
+      "step": 4012
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.645884250408759e-05,
+      "loss": 1.6212,
+      "step": 4013
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.644546387897752e-05,
+      "loss": 1.5257,
+      "step": 4014
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.643208525386744e-05,
+      "loss": 1.7221,
+      "step": 4015
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.641870662875736e-05,
+      "loss": 1.5826,
+      "step": 4016
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.64053280036473e-05,
+      "loss": 1.5176,
+      "step": 4017
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.639194937853722e-05,
+      "loss": 1.4722,
+      "step": 4018
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.637857075342714e-05,
+      "loss": 1.5526,
+      "step": 4019
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.636519212831706e-05,
+      "loss": 1.3557,
+      "step": 4020
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.635181350320698e-05,
+      "loss": 1.4141,
+      "step": 4021
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.633843487809691e-05,
+      "loss": 1.3863,
+      "step": 4022
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.632505625298683e-05,
+      "loss": 1.6019,
+      "step": 4023
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.631167762787675e-05,
+      "loss": 1.3207,
+      "step": 4024
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.629829900276669e-05,
+      "loss": 1.6311,
+      "step": 4025
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.62849203776566e-05,
+      "loss": 1.2906,
+      "step": 4026
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.627154175254654e-05,
+      "loss": 1.4251,
+      "step": 4027
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.625816312743645e-05,
+      "loss": 1.2107,
+      "step": 4028
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.624478450232637e-05,
+      "loss": 1.4011,
+      "step": 4029
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.62314058772163e-05,
+      "loss": 1.4961,
+      "step": 4030
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.621802725210622e-05,
+      "loss": 1.1617,
+      "step": 4031
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.620464862699614e-05,
+      "loss": 1.416,
+      "step": 4032
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.619127000188607e-05,
+      "loss": 1.1827,
+      "step": 4033
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.617789137677599e-05,
+      "loss": 0.9819,
+      "step": 4034
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.616451275166591e-05,
+      "loss": 1.116,
+      "step": 4035
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.615113412655583e-05,
+      "loss": 1.2508,
+      "step": 4036
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.613775550144575e-05,
+      "loss": 0.9572,
+      "step": 4037
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.612437687633569e-05,
+      "loss": 1.0164,
+      "step": 4038
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.61109982512256e-05,
+      "loss": 0.9047,
+      "step": 4039
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.609761962611554e-05,
+      "loss": 1.0158,
+      "step": 4040
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.608424100100546e-05,
+      "loss": 2.3612,
+      "step": 4041
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.607086237589538e-05,
+      "loss": 2.0558,
+      "step": 4042
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.60574837507853e-05,
+      "loss": 1.8025,
+      "step": 4043
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.604410512567522e-05,
+      "loss": 1.843,
+      "step": 4044
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.603072650056514e-05,
+      "loss": 1.6825,
+      "step": 4045
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.601734787545507e-05,
+      "loss": 1.8358,
+      "step": 4046
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.6003969250345e-05,
+      "loss": 1.903,
+      "step": 4047
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.599059062523493e-05,
+      "loss": 1.6736,
+      "step": 4048
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.597721200012485e-05,
+      "loss": 1.8015,
+      "step": 4049
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.596383337501477e-05,
+      "loss": 1.6598,
+      "step": 4050
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.595045474990469e-05,
+      "loss": 1.6612,
+      "step": 4051
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.593707612479461e-05,
+      "loss": 1.4611,
+      "step": 4052
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.592369749968453e-05,
+      "loss": 1.7458,
+      "step": 4053
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.591031887457446e-05,
+      "loss": 1.4953,
+      "step": 4054
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.589694024946438e-05,
+      "loss": 1.6348,
+      "step": 4055
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.588356162435431e-05,
+      "loss": 1.7904,
+      "step": 4056
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.587018299924423e-05,
+      "loss": 1.5364,
+      "step": 4057
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.585680437413414e-05,
+      "loss": 1.5813,
+      "step": 4058
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.584342574902407e-05,
+      "loss": 1.6283,
+      "step": 4059
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.5830047123914e-05,
+      "loss": 1.4862,
+      "step": 4060
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.581666849880393e-05,
+      "loss": 1.8633,
+      "step": 4061
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.580328987369385e-05,
+      "loss": 1.6007,
+      "step": 4062
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.578991124858377e-05,
+      "loss": 1.6407,
+      "step": 4063
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.57765326234737e-05,
+      "loss": 1.4967,
+      "step": 4064
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.576315399836362e-05,
+      "loss": 1.4901,
+      "step": 4065
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.574977537325353e-05,
+      "loss": 1.5051,
+      "step": 4066
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.573639674814346e-05,
+      "loss": 1.3081,
+      "step": 4067
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.572301812303338e-05,
+      "loss": 1.5294,
+      "step": 4068
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.570963949792331e-05,
+      "loss": 1.4254,
+      "step": 4069
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.569626087281323e-05,
+      "loss": 1.468,
+      "step": 4070
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.568288224770315e-05,
+      "loss": 1.4161,
+      "step": 4071
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.566950362259309e-05,
+      "loss": 1.3824,
+      "step": 4072
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.565612499748301e-05,
+      "loss": 1.2145,
+      "step": 4073
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.564274637237293e-05,
+      "loss": 1.4141,
+      "step": 4074
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.562936774726285e-05,
+      "loss": 1.2711,
+      "step": 4075
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.561598912215277e-05,
+      "loss": 1.2518,
+      "step": 4076
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.56026104970427e-05,
+      "loss": 1.3444,
+      "step": 4077
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.558923187193262e-05,
+      "loss": 1.3576,
+      "step": 4078
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.557585324682254e-05,
+      "loss": 1.2544,
+      "step": 4079
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.556247462171248e-05,
+      "loss": 1.1841,
+      "step": 4080
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.55490959966024e-05,
+      "loss": 1.1293,
+      "step": 4081
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.553571737149232e-05,
+      "loss": 1.3587,
+      "step": 4082
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.552233874638224e-05,
+      "loss": 1.1468,
+      "step": 4083
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.550896012127216e-05,
+      "loss": 1.1807,
+      "step": 4084
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.549558149616209e-05,
+      "loss": 1.0224,
+      "step": 4085
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.548220287105201e-05,
+      "loss": 0.9985,
+      "step": 4086
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.546882424594193e-05,
+      "loss": 0.9469,
+      "step": 4087
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.545544562083186e-05,
+      "loss": 0.9892,
+      "step": 4088
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.544206699572177e-05,
+      "loss": 0.9838,
+      "step": 4089
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.54286883706117e-05,
+      "loss": 0.9369,
+      "step": 4090
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.541530974550162e-05,
+      "loss": 2.5169,
+      "step": 4091
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.540193112039154e-05,
+      "loss": 1.9999,
+      "step": 4092
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.538855249528148e-05,
+      "loss": 2.1512,
+      "step": 4093
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.53751738701714e-05,
+      "loss": 1.6732,
+      "step": 4094
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.536179524506133e-05,
+      "loss": 1.9017,
+      "step": 4095
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.534841661995125e-05,
+      "loss": 1.6424,
+      "step": 4096
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.533503799484116e-05,
+      "loss": 1.9656,
+      "step": 4097
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.532165936973109e-05,
+      "loss": 1.9348,
+      "step": 4098
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.530828074462101e-05,
+      "loss": 1.7863,
+      "step": 4099
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.529490211951093e-05,
+      "loss": 1.7675,
+      "step": 4100
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.528152349440086e-05,
+      "loss": 1.669,
+      "step": 4101
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.526814486929078e-05,
+      "loss": 1.8839,
+      "step": 4102
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.525476624418072e-05,
+      "loss": 1.6682,
+      "step": 4103
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.524138761907064e-05,
+      "loss": 1.5641,
+      "step": 4104
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.522800899396054e-05,
+      "loss": 1.8094,
+      "step": 4105
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.521463036885048e-05,
+      "loss": 1.8443,
+      "step": 4106
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.52012517437404e-05,
+      "loss": 1.5659,
+      "step": 4107
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.518787311863033e-05,
+      "loss": 1.6828,
+      "step": 4108
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.517449449352025e-05,
+      "loss": 1.4783,
+      "step": 4109
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.516111586841017e-05,
+      "loss": 1.5782,
+      "step": 4110
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.51477372433001e-05,
+      "loss": 1.4291,
+      "step": 4111
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.513435861819001e-05,
+      "loss": 1.7153,
+      "step": 4112
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.512097999307993e-05,
+      "loss": 1.5311,
+      "step": 4113
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.510760136796986e-05,
+      "loss": 1.5728,
+      "step": 4114
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.509422274285978e-05,
+      "loss": 1.5614,
+      "step": 4115
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.508084411774972e-05,
+      "loss": 1.2749,
+      "step": 4116
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.506746549263964e-05,
+      "loss": 1.4289,
+      "step": 4117
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.505408686752956e-05,
+      "loss": 1.4894,
+      "step": 4118
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.504070824241949e-05,
+      "loss": 1.4968,
+      "step": 4119
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.50273296173094e-05,
+      "loss": 1.5888,
+      "step": 4120
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.501395099219933e-05,
+      "loss": 1.4093,
+      "step": 4121
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.500057236708925e-05,
+      "loss": 1.5147,
+      "step": 4122
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.498719374197917e-05,
+      "loss": 1.3923,
+      "step": 4123
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.49738151168691e-05,
+      "loss": 1.2444,
+      "step": 4124
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.496043649175902e-05,
+      "loss": 1.3498,
+      "step": 4125
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.494705786664894e-05,
+      "loss": 1.2719,
+      "step": 4126
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.493367924153888e-05,
+      "loss": 1.261,
+      "step": 4127
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.492030061642878e-05,
+      "loss": 1.4253,
+      "step": 4128
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.490692199131872e-05,
+      "loss": 1.1519,
+      "step": 4129
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.489354336620864e-05,
+      "loss": 1.4172,
+      "step": 4130
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.488016474109856e-05,
+      "loss": 1.0563,
+      "step": 4131
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.486678611598849e-05,
+      "loss": 1.2608,
+      "step": 4132
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.485340749087841e-05,
+      "loss": 1.0823,
+      "step": 4133
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.484002886576833e-05,
+      "loss": 1.0919,
+      "step": 4134
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.482665024065827e-05,
+      "loss": 0.9559,
+      "step": 4135
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.481327161554817e-05,
+      "loss": 1.1058,
+      "step": 4136
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.47998929904381e-05,
+      "loss": 0.8589,
+      "step": 4137
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.478651436532803e-05,
+      "loss": 1.0432,
+      "step": 4138
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.477313574021795e-05,
+      "loss": 0.8884,
+      "step": 4139
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.475975711510788e-05,
+      "loss": 0.8893,
+      "step": 4140
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.47463784899978e-05,
+      "loss": 2.1718,
+      "step": 4141
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.473299986488773e-05,
+      "loss": 2.1863,
+      "step": 4142
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.471962123977764e-05,
+      "loss": 2.0254,
+      "step": 4143
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.470624261466756e-05,
+      "loss": 1.9119,
+      "step": 4144
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.469286398955749e-05,
+      "loss": 1.6827,
+      "step": 4145
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.467948536444741e-05,
+      "loss": 1.621,
+      "step": 4146
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.466610673933733e-05,
+      "loss": 1.7313,
+      "step": 4147
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.465272811422727e-05,
+      "loss": 1.7256,
+      "step": 4148
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.463934948911719e-05,
+      "loss": 1.9502,
+      "step": 4149
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.462597086400712e-05,
+      "loss": 1.839,
+      "step": 4150
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.461259223889703e-05,
+      "loss": 1.4576,
+      "step": 4151
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.459921361378695e-05,
+      "loss": 1.7068,
+      "step": 4152
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.458583498867688e-05,
+      "loss": 1.5543,
+      "step": 4153
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.45724563635668e-05,
+      "loss": 1.7982,
+      "step": 4154
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.455907773845673e-05,
+      "loss": 1.7725,
+      "step": 4155
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.454569911334665e-05,
+      "loss": 1.5522,
+      "step": 4156
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.453232048823657e-05,
+      "loss": 1.3626,
+      "step": 4157
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.45189418631265e-05,
+      "loss": 1.8835,
+      "step": 4158
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.450556323801641e-05,
+      "loss": 1.6252,
+      "step": 4159
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.449218461290633e-05,
+      "loss": 1.5221,
+      "step": 4160
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.447880598779627e-05,
+      "loss": 1.6361,
+      "step": 4161
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.446542736268619e-05,
+      "loss": 1.6942,
+      "step": 4162
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.445204873757612e-05,
+      "loss": 1.4064,
+      "step": 4163
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.443867011246604e-05,
+      "loss": 1.5987,
+      "step": 4164
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.442529148735596e-05,
+      "loss": 1.6474,
+      "step": 4165
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.441191286224588e-05,
+      "loss": 1.6248,
+      "step": 4166
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.43985342371358e-05,
+      "loss": 1.5026,
+      "step": 4167
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.438515561202572e-05,
+      "loss": 1.3564,
+      "step": 4168
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.437177698691565e-05,
+      "loss": 1.3951,
+      "step": 4169
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.435839836180557e-05,
+      "loss": 1.5188,
+      "step": 4170
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.434501973669551e-05,
+      "loss": 1.396,
+      "step": 4171
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.433164111158543e-05,
+      "loss": 1.3595,
+      "step": 4172
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.431826248647535e-05,
+      "loss": 1.277,
+      "step": 4173
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.430488386136527e-05,
+      "loss": 1.4881,
+      "step": 4174
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.429150523625519e-05,
+      "loss": 1.3318,
+      "step": 4175
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.427812661114512e-05,
+      "loss": 1.2848,
+      "step": 4176
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.426474798603504e-05,
+      "loss": 1.323,
+      "step": 4177
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.425136936092496e-05,
+      "loss": 1.3095,
+      "step": 4178
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.42379907358149e-05,
+      "loss": 1.3433,
+      "step": 4179
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.422461211070481e-05,
+      "loss": 1.2151,
+      "step": 4180
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.421123348559473e-05,
+      "loss": 1.0652,
+      "step": 4181
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.419785486048465e-05,
+      "loss": 1.3325,
+      "step": 4182
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.418447623537457e-05,
+      "loss": 1.164,
+      "step": 4183
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.417109761026451e-05,
+      "loss": 0.8567,
+      "step": 4184
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.415771898515443e-05,
+      "loss": 0.9705,
+      "step": 4185
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.414434036004435e-05,
+      "loss": 1.0048,
+      "step": 4186
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.413096173493428e-05,
+      "loss": 1.1,
+      "step": 4187
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.41175831098242e-05,
+      "loss": 1.0314,
+      "step": 4188
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.410420448471414e-05,
+      "loss": 0.9848,
+      "step": 4189
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.409082585960404e-05,
+      "loss": 0.9268,
+      "step": 4190
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.407744723449396e-05,
+      "loss": 2.4671,
+      "step": 4191
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.40640686093839e-05,
+      "loss": 2.0396,
+      "step": 4192
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.405068998427382e-05,
+      "loss": 1.8168,
+      "step": 4193
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.403731135916374e-05,
+      "loss": 1.8152,
+      "step": 4194
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.402393273405367e-05,
+      "loss": 1.7182,
+      "step": 4195
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.401055410894359e-05,
+      "loss": 1.793,
+      "step": 4196
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.399717548383351e-05,
+      "loss": 1.8249,
+      "step": 4197
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.398379685872343e-05,
+      "loss": 1.8629,
+      "step": 4198
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.397041823361335e-05,
+      "loss": 1.673,
+      "step": 4199
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.395703960850328e-05,
+      "loss": 1.8551,
+      "step": 4200
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.39436609833932e-05,
+      "loss": 1.7996,
+      "step": 4201
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.393028235828312e-05,
+      "loss": 1.8196,
+      "step": 4202
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.391690373317306e-05,
+      "loss": 1.7224,
+      "step": 4203
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.390352510806298e-05,
+      "loss": 1.5642,
+      "step": 4204
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.38901464829529e-05,
+      "loss": 1.6854,
+      "step": 4205
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.387676785784282e-05,
+      "loss": 1.5895,
+      "step": 4206
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.386338923273274e-05,
+      "loss": 1.6929,
+      "step": 4207
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.385001060762267e-05,
+      "loss": 1.5724,
+      "step": 4208
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.383663198251259e-05,
+      "loss": 1.7155,
+      "step": 4209
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.382325335740252e-05,
+      "loss": 1.5853,
+      "step": 4210
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.380987473229244e-05,
+      "loss": 1.5339,
+      "step": 4211
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.379649610718236e-05,
+      "loss": 1.5553,
+      "step": 4212
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.378311748207228e-05,
+      "loss": 1.4813,
+      "step": 4213
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.37697388569622e-05,
+      "loss": 1.4196,
+      "step": 4214
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.375636023185212e-05,
+      "loss": 1.392,
+      "step": 4215
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.374298160674206e-05,
+      "loss": 1.7339,
+      "step": 4216
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.372960298163198e-05,
+      "loss": 1.5719,
+      "step": 4217
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.371622435652191e-05,
+      "loss": 1.4838,
+      "step": 4218
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.370284573141183e-05,
+      "loss": 1.3264,
+      "step": 4219
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.368946710630174e-05,
+      "loss": 1.5419,
+      "step": 4220
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.367608848119167e-05,
+      "loss": 1.3075,
+      "step": 4221
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.366270985608159e-05,
+      "loss": 1.4749,
+      "step": 4222
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.364933123097152e-05,
+      "loss": 1.2959,
+      "step": 4223
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.363595260586144e-05,
+      "loss": 1.4177,
+      "step": 4224
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.362257398075136e-05,
+      "loss": 1.2956,
+      "step": 4225
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.36091953556413e-05,
+      "loss": 1.4291,
+      "step": 4226
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.359581673053122e-05,
+      "loss": 1.4335,
+      "step": 4227
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.358243810542112e-05,
+      "loss": 1.3665,
+      "step": 4228
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.356905948031106e-05,
+      "loss": 1.3993,
+      "step": 4229
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.355568085520098e-05,
+      "loss": 1.2231,
+      "step": 4230
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.354230223009091e-05,
+      "loss": 1.051,
+      "step": 4231
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.352892360498083e-05,
+      "loss": 1.3276,
+      "step": 4232
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.351554497987075e-05,
+      "loss": 1.2523,
+      "step": 4233
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.350216635476068e-05,
+      "loss": 1.1486,
+      "step": 4234
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.34887877296506e-05,
+      "loss": 1.23,
+      "step": 4235
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.347540910454052e-05,
+      "loss": 1.1596,
+      "step": 4236
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.346203047943044e-05,
+      "loss": 1.0545,
+      "step": 4237
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.344865185432036e-05,
+      "loss": 0.927,
+      "step": 4238
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.34352732292103e-05,
+      "loss": 0.8279,
+      "step": 4239
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.342189460410022e-05,
+      "loss": 1.0032,
+      "step": 4240
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.340851597899014e-05,
+      "loss": 2.1524,
+      "step": 4241
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.339513735388007e-05,
+      "loss": 1.9022,
+      "step": 4242
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.338175872876999e-05,
+      "loss": 1.85,
+      "step": 4243
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.336838010365991e-05,
+      "loss": 1.6113,
+      "step": 4244
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.335500147854983e-05,
+      "loss": 1.7112,
+      "step": 4245
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.334162285343975e-05,
+      "loss": 1.7958,
+      "step": 4246
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.332824422832969e-05,
+      "loss": 1.8314,
+      "step": 4247
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.33148656032196e-05,
+      "loss": 1.6462,
+      "step": 4248
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.330148697810953e-05,
+      "loss": 1.6582,
+      "step": 4249
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.328810835299946e-05,
+      "loss": 1.7708,
+      "step": 4250
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.327472972788937e-05,
+      "loss": 1.6131,
+      "step": 4251
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.32613511027793e-05,
+      "loss": 1.6238,
+      "step": 4252
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.324797247766922e-05,
+      "loss": 1.4735,
+      "step": 4253
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.323459385255914e-05,
+      "loss": 1.7066,
+      "step": 4254
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.322121522744907e-05,
+      "loss": 1.7619,
+      "step": 4255
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.320783660233899e-05,
+      "loss": 1.675,
+      "step": 4256
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.319445797722893e-05,
+      "loss": 1.4322,
+      "step": 4257
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.318107935211885e-05,
+      "loss": 1.547,
+      "step": 4258
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.316770072700875e-05,
+      "loss": 1.6918,
+      "step": 4259
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.315432210189869e-05,
+      "loss": 1.6435,
+      "step": 4260
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.31409434767886e-05,
+      "loss": 1.586,
+      "step": 4261
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.312756485167853e-05,
+      "loss": 1.5123,
+      "step": 4262
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.311418622656846e-05,
+      "loss": 1.6891,
+      "step": 4263
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.310080760145838e-05,
+      "loss": 1.5541,
+      "step": 4264
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.308742897634831e-05,
+      "loss": 1.509,
+      "step": 4265
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.307405035123823e-05,
+      "loss": 1.6127,
+      "step": 4266
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.306067172612814e-05,
+      "loss": 1.4835,
+      "step": 4267
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.304729310101807e-05,
+      "loss": 1.5373,
+      "step": 4268
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.303391447590799e-05,
+      "loss": 1.4546,
+      "step": 4269
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.302053585079793e-05,
+      "loss": 1.5482,
+      "step": 4270
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.300715722568785e-05,
+      "loss": 1.5601,
+      "step": 4271
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.299377860057777e-05,
+      "loss": 1.4532,
+      "step": 4272
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.29803999754677e-05,
+      "loss": 1.4446,
+      "step": 4273
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.29670213503576e-05,
+      "loss": 1.6224,
+      "step": 4274
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.295364272524753e-05,
+      "loss": 1.3368,
+      "step": 4275
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.294026410013746e-05,
+      "loss": 1.2832,
+      "step": 4276
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.292688547502738e-05,
+      "loss": 1.2094,
+      "step": 4277
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.291350684991731e-05,
+      "loss": 1.258,
+      "step": 4278
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.290012822480723e-05,
+      "loss": 1.1763,
+      "step": 4279
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.288674959969715e-05,
+      "loss": 1.1574,
+      "step": 4280
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.287337097458709e-05,
+      "loss": 1.2194,
+      "step": 4281
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.2859992349477e-05,
+      "loss": 1.2187,
+      "step": 4282
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.284661372436691e-05,
+      "loss": 1.2988,
+      "step": 4283
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.283323509925685e-05,
+      "loss": 1.0381,
+      "step": 4284
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.281985647414677e-05,
+      "loss": 0.967,
+      "step": 4285
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.28064778490367e-05,
+      "loss": 1.115,
+      "step": 4286
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.279309922392662e-05,
+      "loss": 0.9473,
+      "step": 4287
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.277972059881654e-05,
+      "loss": 0.897,
+      "step": 4288
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.276634197370647e-05,
+      "loss": 0.883,
+      "step": 4289
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.275296334859638e-05,
+      "loss": 0.9409,
+      "step": 4290
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.273958472348631e-05,
+      "loss": 2.2921,
+      "step": 4291
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.272620609837623e-05,
+      "loss": 2.1465,
+      "step": 4292
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.271282747326615e-05,
+      "loss": 1.775,
+      "step": 4293
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.269944884815609e-05,
+      "loss": 1.7827,
+      "step": 4294
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.268607022304601e-05,
+      "loss": 1.7288,
+      "step": 4295
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.267269159793593e-05,
+      "loss": 1.672,
+      "step": 4296
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.265931297282586e-05,
+      "loss": 1.818,
+      "step": 4297
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.264593434771577e-05,
+      "loss": 1.5174,
+      "step": 4298
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.26325557226057e-05,
+      "loss": 1.7362,
+      "step": 4299
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.261917709749562e-05,
+      "loss": 1.6268,
+      "step": 4300
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.260579847238554e-05,
+      "loss": 1.5965,
+      "step": 4301
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.259241984727547e-05,
+      "loss": 1.5233,
+      "step": 4302
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.25790412221654e-05,
+      "loss": 1.5163,
+      "step": 4303
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.256566259705533e-05,
+      "loss": 1.7054,
+      "step": 4304
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.255228397194523e-05,
+      "loss": 1.572,
+      "step": 4305
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.253890534683515e-05,
+      "loss": 1.4627,
+      "step": 4306
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.252552672172509e-05,
+      "loss": 1.5418,
+      "step": 4307
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.251214809661501e-05,
+      "loss": 1.7373,
+      "step": 4308
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.249876947150493e-05,
+      "loss": 1.4045,
+      "step": 4309
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.248539084639486e-05,
+      "loss": 1.5467,
+      "step": 4310
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.247201222128478e-05,
+      "loss": 1.4924,
+      "step": 4311
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.245863359617472e-05,
+      "loss": 1.5209,
+      "step": 4312
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.244525497106462e-05,
+      "loss": 1.7976,
+      "step": 4313
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.243187634595454e-05,
+      "loss": 1.5347,
+      "step": 4314
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.241849772084448e-05,
+      "loss": 1.3406,
+      "step": 4315
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.24051190957344e-05,
+      "loss": 1.4341,
+      "step": 4316
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.239174047062433e-05,
+      "loss": 1.7756,
+      "step": 4317
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.237836184551425e-05,
+      "loss": 1.5478,
+      "step": 4318
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.236498322040417e-05,
+      "loss": 1.4648,
+      "step": 4319
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.23516045952941e-05,
+      "loss": 1.6238,
+      "step": 4320
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.233822597018401e-05,
+      "loss": 1.3676,
+      "step": 4321
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.232484734507393e-05,
+      "loss": 1.3549,
+      "step": 4322
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.231146871996386e-05,
+      "loss": 1.2882,
+      "step": 4323
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.229809009485378e-05,
+      "loss": 1.2017,
+      "step": 4324
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.228471146974372e-05,
+      "loss": 1.3862,
+      "step": 4325
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.227133284463364e-05,
+      "loss": 1.3911,
+      "step": 4326
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.225795421952356e-05,
+      "loss": 1.2458,
+      "step": 4327
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.224457559441348e-05,
+      "loss": 1.0016,
+      "step": 4328
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.22311969693034e-05,
+      "loss": 1.1664,
+      "step": 4329
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.221781834419332e-05,
+      "loss": 1.1087,
+      "step": 4330
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.220443971908325e-05,
+      "loss": 1.0405,
+      "step": 4331
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.219106109397317e-05,
+      "loss": 1.2887,
+      "step": 4332
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.21776824688631e-05,
+      "loss": 0.9853,
+      "step": 4333
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.216430384375302e-05,
+      "loss": 1.0577,
+      "step": 4334
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.215092521864294e-05,
+      "loss": 1.1949,
+      "step": 4335
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.213754659353286e-05,
+      "loss": 1.1725,
+      "step": 4336
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.212416796842278e-05,
+      "loss": 0.9512,
+      "step": 4337
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.211078934331272e-05,
+      "loss": 0.9586,
+      "step": 4338
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.209741071820264e-05,
+      "loss": 0.8554,
+      "step": 4339
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.208403209309256e-05,
+      "loss": 0.7415,
+      "step": 4340
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.207065346798249e-05,
+      "loss": 2.3265,
+      "step": 4341
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.205727484287241e-05,
+      "loss": 2.085,
+      "step": 4342
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.204389621776233e-05,
+      "loss": 1.9231,
+      "step": 4343
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.203051759265225e-05,
+      "loss": 1.6207,
+      "step": 4344
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.201713896754217e-05,
+      "loss": 1.7939,
+      "step": 4345
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.20037603424321e-05,
+      "loss": 1.7911,
+      "step": 4346
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.199038171732202e-05,
+      "loss": 1.6761,
+      "step": 4347
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.197700309221194e-05,
+      "loss": 1.7795,
+      "step": 4348
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.196362446710188e-05,
+      "loss": 1.753,
+      "step": 4349
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.19502458419918e-05,
+      "loss": 1.6527,
+      "step": 4350
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.193686721688172e-05,
+      "loss": 1.7374,
+      "step": 4351
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.192348859177164e-05,
+      "loss": 1.5893,
+      "step": 4352
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.191010996666156e-05,
+      "loss": 1.4695,
+      "step": 4353
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.189673134155149e-05,
+      "loss": 1.4539,
+      "step": 4354
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.188335271644141e-05,
+      "loss": 1.4358,
+      "step": 4355
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.186997409133133e-05,
+      "loss": 1.6323,
+      "step": 4356
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.185659546622126e-05,
+      "loss": 1.5659,
+      "step": 4357
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.184321684111118e-05,
+      "loss": 1.5182,
+      "step": 4358
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.18298382160011e-05,
+      "loss": 1.4951,
+      "step": 4359
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.181645959089102e-05,
+      "loss": 1.6688,
+      "step": 4360
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.180308096578094e-05,
+      "loss": 1.5519,
+      "step": 4361
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.178970234067088e-05,
+      "loss": 1.6162,
+      "step": 4362
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.17763237155608e-05,
+      "loss": 1.5107,
+      "step": 4363
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.176294509045072e-05,
+      "loss": 1.4823,
+      "step": 4364
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.174956646534065e-05,
+      "loss": 1.3774,
+      "step": 4365
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.173618784023057e-05,
+      "loss": 1.4537,
+      "step": 4366
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.172280921512049e-05,
+      "loss": 1.5061,
+      "step": 4367
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.170943059001041e-05,
+      "loss": 1.2117,
+      "step": 4368
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.169605196490033e-05,
+      "loss": 1.3174,
+      "step": 4369
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.168267333979027e-05,
+      "loss": 1.5115,
+      "step": 4370
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.166929471468019e-05,
+      "loss": 1.3106,
+      "step": 4371
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.165591608957012e-05,
+      "loss": 1.3588,
+      "step": 4372
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.164253746446004e-05,
+      "loss": 1.1552,
+      "step": 4373
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.162915883934996e-05,
+      "loss": 1.3644,
+      "step": 4374
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.161578021423988e-05,
+      "loss": 1.2809,
+      "step": 4375
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.16024015891298e-05,
+      "loss": 1.2223,
+      "step": 4376
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.158902296401972e-05,
+      "loss": 1.3599,
+      "step": 4377
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.157564433890965e-05,
+      "loss": 1.2612,
+      "step": 4378
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.156226571379957e-05,
+      "loss": 1.1727,
+      "step": 4379
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.15488870886895e-05,
+      "loss": 1.301,
+      "step": 4380
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.153550846357943e-05,
+      "loss": 1.3233,
+      "step": 4381
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.152212983846933e-05,
+      "loss": 1.3394,
+      "step": 4382
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.150875121335927e-05,
+      "loss": 1.0825,
+      "step": 4383
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.149537258824919e-05,
+      "loss": 1.1811,
+      "step": 4384
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.148199396313912e-05,
+      "loss": 1.1889,
+      "step": 4385
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.146861533802904e-05,
+      "loss": 1.1741,
+      "step": 4386
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.145523671291896e-05,
+      "loss": 1.0408,
+      "step": 4387
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.144185808780889e-05,
+      "loss": 0.8977,
+      "step": 4388
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.142847946269881e-05,
+      "loss": 0.9265,
+      "step": 4389
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.141510083758872e-05,
+      "loss": 0.9564,
+      "step": 4390
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.140172221247865e-05,
+      "loss": 2.3115,
+      "step": 4391
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.138834358736857e-05,
+      "loss": 1.9906,
+      "step": 4392
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.13749649622585e-05,
+      "loss": 1.8686,
+      "step": 4393
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.136158633714843e-05,
+      "loss": 1.8152,
+      "step": 4394
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.134820771203835e-05,
+      "loss": 1.8791,
+      "step": 4395
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.133482908692828e-05,
+      "loss": 1.822,
+      "step": 4396
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.13214504618182e-05,
+      "loss": 1.5298,
+      "step": 4397
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.13080718367081e-05,
+      "loss": 1.6982,
+      "step": 4398
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.129469321159804e-05,
+      "loss": 1.8207,
+      "step": 4399
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.128131458648796e-05,
+      "loss": 1.8822,
+      "step": 4400
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.12679359613779e-05,
+      "loss": 1.8439,
+      "step": 4401
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.125455733626781e-05,
+      "loss": 1.8174,
+      "step": 4402
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.124117871115773e-05,
+      "loss": 1.7982,
+      "step": 4403
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.122780008604767e-05,
+      "loss": 2.0311,
+      "step": 4404
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.121442146093757e-05,
+      "loss": 1.7725,
+      "step": 4405
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.120104283582751e-05,
+      "loss": 1.4551,
+      "step": 4406
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.118766421071743e-05,
+      "loss": 1.9312,
+      "step": 4407
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.117428558560735e-05,
+      "loss": 1.515,
+      "step": 4408
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.116090696049728e-05,
+      "loss": 1.6651,
+      "step": 4409
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.11475283353872e-05,
+      "loss": 1.6907,
+      "step": 4410
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.113414971027712e-05,
+      "loss": 1.3742,
+      "step": 4411
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.112077108516705e-05,
+      "loss": 1.4935,
+      "step": 4412
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.110739246005696e-05,
+      "loss": 1.4028,
+      "step": 4413
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.10940138349469e-05,
+      "loss": 1.703,
+      "step": 4414
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.108063520983681e-05,
+      "loss": 1.5034,
+      "step": 4415
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.106725658472673e-05,
+      "loss": 1.5154,
+      "step": 4416
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.105387795961667e-05,
+      "loss": 1.477,
+      "step": 4417
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.104049933450659e-05,
+      "loss": 1.4603,
+      "step": 4418
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.102712070939652e-05,
+      "loss": 1.4031,
+      "step": 4419
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.101374208428644e-05,
+      "loss": 1.4982,
+      "step": 4420
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.100036345917635e-05,
+      "loss": 1.3698,
+      "step": 4421
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.098698483406628e-05,
+      "loss": 1.5016,
+      "step": 4422
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.09736062089562e-05,
+      "loss": 1.3166,
+      "step": 4423
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.096022758384612e-05,
+      "loss": 1.4078,
+      "step": 4424
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.094684895873606e-05,
+      "loss": 1.3029,
+      "step": 4425
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.093347033362598e-05,
+      "loss": 1.2665,
+      "step": 4426
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.092009170851591e-05,
+      "loss": 1.3555,
+      "step": 4427
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.090671308340583e-05,
+      "loss": 1.3396,
+      "step": 4428
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.089333445829574e-05,
+      "loss": 1.2745,
+      "step": 4429
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.087995583318567e-05,
+      "loss": 1.0853,
+      "step": 4430
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.086657720807559e-05,
+      "loss": 1.1409,
+      "step": 4431
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.085319858296552e-05,
+      "loss": 1.2398,
+      "step": 4432
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.083981995785544e-05,
+      "loss": 1.2194,
+      "step": 4433
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.082644133274536e-05,
+      "loss": 1.0792,
+      "step": 4434
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.08130627076353e-05,
+      "loss": 1.3199,
+      "step": 4435
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.07996840825252e-05,
+      "loss": 1.0032,
+      "step": 4436
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.078630545741512e-05,
+      "loss": 1.0561,
+      "step": 4437
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.077292683230506e-05,
+      "loss": 0.7955,
+      "step": 4438
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.075954820719498e-05,
+      "loss": 0.9352,
+      "step": 4439
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.074616958208491e-05,
+      "loss": 0.9208,
+      "step": 4440
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.073279095697483e-05,
+      "loss": 2.4437,
+      "step": 4441
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.071941233186475e-05,
+      "loss": 1.9985,
+      "step": 4442
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.070603370675468e-05,
+      "loss": 2.0779,
+      "step": 4443
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.069265508164459e-05,
+      "loss": 1.8214,
+      "step": 4444
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.067927645653451e-05,
+      "loss": 1.93,
+      "step": 4445
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.066589783142444e-05,
+      "loss": 1.7193,
+      "step": 4446
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.065251920631436e-05,
+      "loss": 1.7055,
+      "step": 4447
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.06391405812043e-05,
+      "loss": 1.6467,
+      "step": 4448
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.062576195609422e-05,
+      "loss": 1.7341,
+      "step": 4449
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.061238333098414e-05,
+      "loss": 1.4428,
+      "step": 4450
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.059900470587407e-05,
+      "loss": 1.6307,
+      "step": 4451
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.058562608076398e-05,
+      "loss": 1.7578,
+      "step": 4452
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.057224745565391e-05,
+      "loss": 1.6154,
+      "step": 4453
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.055886883054383e-05,
+      "loss": 1.8444,
+      "step": 4454
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.054549020543375e-05,
+      "loss": 1.5563,
+      "step": 4455
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.053211158032368e-05,
+      "loss": 1.5615,
+      "step": 4456
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.05187329552136e-05,
+      "loss": 1.6555,
+      "step": 4457
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.050535433010352e-05,
+      "loss": 1.5371,
+      "step": 4458
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.049197570499344e-05,
+      "loss": 1.6923,
+      "step": 4459
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.047859707988336e-05,
+      "loss": 1.8252,
+      "step": 4460
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.04652184547733e-05,
+      "loss": 1.4896,
+      "step": 4461
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.045183982966322e-05,
+      "loss": 1.5415,
+      "step": 4462
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.043846120455314e-05,
+      "loss": 1.4948,
+      "step": 4463
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.042508257944307e-05,
+      "loss": 1.6075,
+      "step": 4464
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.041170395433299e-05,
+      "loss": 1.4557,
+      "step": 4465
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.039832532922292e-05,
+      "loss": 1.3123,
+      "step": 4466
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.038494670411283e-05,
+      "loss": 1.506,
+      "step": 4467
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.037156807900275e-05,
+      "loss": 1.5092,
+      "step": 4468
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.035818945389268e-05,
+      "loss": 1.5054,
+      "step": 4469
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.03448108287826e-05,
+      "loss": 1.3294,
+      "step": 4470
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.033143220367252e-05,
+      "loss": 1.3393,
+      "step": 4471
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.031805357856246e-05,
+      "loss": 1.2653,
+      "step": 4472
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.030467495345238e-05,
+      "loss": 1.1627,
+      "step": 4473
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.029129632834231e-05,
+      "loss": 1.2238,
+      "step": 4474
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.027791770323222e-05,
+      "loss": 1.4332,
+      "step": 4475
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.026453907812214e-05,
+      "loss": 1.4426,
+      "step": 4476
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.025116045301207e-05,
+      "loss": 1.3445,
+      "step": 4477
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.023778182790199e-05,
+      "loss": 1.2305,
+      "step": 4478
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.022440320279191e-05,
+      "loss": 1.1611,
+      "step": 4479
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.021102457768185e-05,
+      "loss": 1.2569,
+      "step": 4480
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.019764595257177e-05,
+      "loss": 1.2585,
+      "step": 4481
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.01842673274617e-05,
+      "loss": 1.2094,
+      "step": 4482
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.01708887023516e-05,
+      "loss": 1.1667,
+      "step": 4483
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.015751007724153e-05,
+      "loss": 1.2238,
+      "step": 4484
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.014413145213146e-05,
+      "loss": 0.985,
+      "step": 4485
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.013075282702138e-05,
+      "loss": 1.2808,
+      "step": 4486
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.011737420191131e-05,
+      "loss": 1.0381,
+      "step": 4487
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.010399557680123e-05,
+      "loss": 0.9233,
+      "step": 4488
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.009061695169115e-05,
+      "loss": 0.924,
+      "step": 4489
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.007723832658107e-05,
+      "loss": 0.7578,
+      "step": 4490
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.006385970147099e-05,
+      "loss": 2.2578,
+      "step": 4491
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.005048107636091e-05,
+      "loss": 1.9857,
+      "step": 4492
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.003710245125085e-05,
+      "loss": 1.9102,
+      "step": 4493
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.002372382614077e-05,
+      "loss": 1.7836,
+      "step": 4494
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.00103452010307e-05,
+      "loss": 1.5562,
+      "step": 4495
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.999696657592062e-05,
+      "loss": 1.5899,
+      "step": 4496
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.998358795081054e-05,
+      "loss": 1.5511,
+      "step": 4497
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.997020932570046e-05,
+      "loss": 1.5468,
+      "step": 4498
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.995683070059038e-05,
+      "loss": 1.7856,
+      "step": 4499
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.994345207548031e-05,
+      "loss": 1.8298,
+      "step": 4500
+    },
+    {
+      "epoch": 1.39,
+      "eval_bleu": 2.2108988861067766e-14,
+      "eval_loss": 1.8817064762115479,
+      "eval_runtime": 2552.8126,
+      "eval_samples_per_second": 5.782,
+      "eval_steps_per_second": 0.723,
+      "step": 4500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.993007345037023e-05,
+      "loss": 1.8291,
+      "step": 4501
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.991669482526015e-05,
+      "loss": 1.4921,
+      "step": 4502
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.990331620015009e-05,
+      "loss": 1.4788,
+      "step": 4503
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.988993757504e-05,
+      "loss": 1.511,
+      "step": 4504
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.987655894992993e-05,
+      "loss": 1.6948,
+      "step": 4505
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.986318032481985e-05,
+      "loss": 1.5454,
+      "step": 4506
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.984980169970977e-05,
+      "loss": 1.4468,
+      "step": 4507
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.98364230745997e-05,
+      "loss": 1.5013,
+      "step": 4508
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.982304444948962e-05,
+      "loss": 1.7455,
+      "step": 4509
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.980966582437954e-05,
+      "loss": 1.6199,
+      "step": 4510
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.979628719926947e-05,
+      "loss": 1.6671,
+      "step": 4511
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.97829085741594e-05,
+      "loss": 1.6328,
+      "step": 4512
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.97695299490493e-05,
+      "loss": 1.6066,
+      "step": 4513
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.975615132393923e-05,
+      "loss": 1.4747,
+      "step": 4514
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.974277269882915e-05,
+      "loss": 1.6273,
+      "step": 4515
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.972939407371909e-05,
+      "loss": 1.3899,
+      "step": 4516
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.971601544860901e-05,
+      "loss": 1.4959,
+      "step": 4517
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.970263682349893e-05,
+      "loss": 1.3845,
+      "step": 4518
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.968925819838886e-05,
+      "loss": 1.5342,
+      "step": 4519
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.967587957327878e-05,
+      "loss": 1.3194,
+      "step": 4520
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.96625009481687e-05,
+      "loss": 1.1793,
+      "step": 4521
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.964912232305862e-05,
+      "loss": 1.4526,
+      "step": 4522
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.963574369794854e-05,
+      "loss": 1.2458,
+      "step": 4523
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.962236507283847e-05,
+      "loss": 1.1992,
+      "step": 4524
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.96089864477284e-05,
+      "loss": 1.2159,
+      "step": 4525
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.959560782261831e-05,
+      "loss": 1.2888,
+      "step": 4526
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.958222919750825e-05,
+      "loss": 1.3514,
+      "step": 4527
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.956885057239817e-05,
+      "loss": 1.217,
+      "step": 4528
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.955547194728809e-05,
+      "loss": 1.4004,
+      "step": 4529
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.954209332217801e-05,
+      "loss": 1.2527,
+      "step": 4530
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.952871469706793e-05,
+      "loss": 1.0971,
+      "step": 4531
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.951533607195786e-05,
+      "loss": 1.0721,
+      "step": 4532
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.950195744684778e-05,
+      "loss": 1.0733,
+      "step": 4533
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.948857882173771e-05,
+      "loss": 0.9573,
+      "step": 4534
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.947520019662763e-05,
+      "loss": 1.0846,
+      "step": 4535
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.946182157151754e-05,
+      "loss": 1.0682,
+      "step": 4536
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.944844294640747e-05,
+      "loss": 0.9188,
+      "step": 4537
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.94350643212974e-05,
+      "loss": 1.0543,
+      "step": 4538
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.942168569618731e-05,
+      "loss": 0.9083,
+      "step": 4539
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.940830707107725e-05,
+      "loss": 0.9551,
+      "step": 4540
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.939492844596717e-05,
+      "loss": 2.1402,
+      "step": 4541
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.93815498208571e-05,
+      "loss": 2.0455,
+      "step": 4542
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.936817119574702e-05,
+      "loss": 1.9808,
+      "step": 4543
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.935479257063693e-05,
+      "loss": 1.8248,
+      "step": 4544
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.934141394552686e-05,
+      "loss": 1.6554,
+      "step": 4545
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.932803532041678e-05,
+      "loss": 1.3998,
+      "step": 4546
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.931465669530672e-05,
+      "loss": 1.6089,
+      "step": 4547
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.930127807019664e-05,
+      "loss": 1.6531,
+      "step": 4548
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.928789944508656e-05,
+      "loss": 1.6563,
+      "step": 4549
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.927452081997649e-05,
+      "loss": 1.5785,
+      "step": 4550
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.926114219486641e-05,
+      "loss": 1.6533,
+      "step": 4551
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.924776356975632e-05,
+      "loss": 1.6709,
+      "step": 4552
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.923438494464625e-05,
+      "loss": 1.3542,
+      "step": 4553
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.922100631953617e-05,
+      "loss": 1.5817,
+      "step": 4554
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.92076276944261e-05,
+      "loss": 1.6297,
+      "step": 4555
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.919424906931602e-05,
+      "loss": 1.5408,
+      "step": 4556
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.918087044420594e-05,
+      "loss": 1.6686,
+      "step": 4557
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.916749181909588e-05,
+      "loss": 1.6556,
+      "step": 4558
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.91541131939858e-05,
+      "loss": 1.6006,
+      "step": 4559
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.91407345688757e-05,
+      "loss": 1.4036,
+      "step": 4560
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.912735594376564e-05,
+      "loss": 1.5148,
+      "step": 4561
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.911397731865556e-05,
+      "loss": 1.491,
+      "step": 4562
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.910059869354549e-05,
+      "loss": 1.5074,
+      "step": 4563
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.908722006843541e-05,
+      "loss": 1.3413,
+      "step": 4564
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.907384144332533e-05,
+      "loss": 1.5485,
+      "step": 4565
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.906046281821526e-05,
+      "loss": 1.5707,
+      "step": 4566
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.904708419310517e-05,
+      "loss": 1.3756,
+      "step": 4567
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.90337055679951e-05,
+      "loss": 1.4366,
+      "step": 4568
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.902032694288502e-05,
+      "loss": 1.4123,
+      "step": 4569
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.900694831777494e-05,
+      "loss": 1.4214,
+      "step": 4570
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.899356969266488e-05,
+      "loss": 1.3113,
+      "step": 4571
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.89801910675548e-05,
+      "loss": 1.4138,
+      "step": 4572
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.896681244244472e-05,
+      "loss": 1.442,
+      "step": 4573
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.895343381733465e-05,
+      "loss": 1.2952,
+      "step": 4574
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.894005519222456e-05,
+      "loss": 1.4539,
+      "step": 4575
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.892667656711449e-05,
+      "loss": 1.2604,
+      "step": 4576
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.891329794200441e-05,
+      "loss": 1.2156,
+      "step": 4577
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.889991931689433e-05,
+      "loss": 1.1413,
+      "step": 4578
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.888654069178426e-05,
+      "loss": 1.215,
+      "step": 4579
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.887316206667418e-05,
+      "loss": 1.3712,
+      "step": 4580
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.885978344156412e-05,
+      "loss": 1.1814,
+      "step": 4581
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.884640481645404e-05,
+      "loss": 1.1121,
+      "step": 4582
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.883302619134394e-05,
+      "loss": 1.0939,
+      "step": 4583
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.881964756623388e-05,
+      "loss": 1.1124,
+      "step": 4584
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.88062689411238e-05,
+      "loss": 1.0951,
+      "step": 4585
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.879289031601372e-05,
+      "loss": 1.1034,
+      "step": 4586
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.877951169090365e-05,
+      "loss": 0.9143,
+      "step": 4587
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.876613306579357e-05,
+      "loss": 0.9492,
+      "step": 4588
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.87527544406835e-05,
+      "loss": 0.8677,
+      "step": 4589
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.873937581557341e-05,
+      "loss": 0.9419,
+      "step": 4590
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.872599719046333e-05,
+      "loss": 1.8769,
+      "step": 4591
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.871261856535326e-05,
+      "loss": 2.0096,
+      "step": 4592
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.869923994024318e-05,
+      "loss": 1.9617,
+      "step": 4593
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.86858613151331e-05,
+      "loss": 1.7755,
+      "step": 4594
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.867248269002304e-05,
+      "loss": 1.589,
+      "step": 4595
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.865910406491296e-05,
+      "loss": 1.8707,
+      "step": 4596
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.864572543980289e-05,
+      "loss": 1.5818,
+      "step": 4597
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.86323468146928e-05,
+      "loss": 1.8012,
+      "step": 4598
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.861896818958272e-05,
+      "loss": 1.727,
+      "step": 4599
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.860558956447265e-05,
+      "loss": 1.4537,
+      "step": 4600
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.859221093936257e-05,
+      "loss": 1.5817,
+      "step": 4601
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.85788323142525e-05,
+      "loss": 1.5348,
+      "step": 4602
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.856545368914243e-05,
+      "loss": 1.5101,
+      "step": 4603
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.855207506403235e-05,
+      "loss": 1.541,
+      "step": 4604
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.853869643892228e-05,
+      "loss": 1.5067,
+      "step": 4605
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.852531781381219e-05,
+      "loss": 1.5038,
+      "step": 4606
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.85119391887021e-05,
+      "loss": 1.5343,
+      "step": 4607
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.849856056359204e-05,
+      "loss": 1.5134,
+      "step": 4608
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.848518193848196e-05,
+      "loss": 1.6359,
+      "step": 4609
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.847180331337189e-05,
+      "loss": 1.6285,
+      "step": 4610
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.845842468826181e-05,
+      "loss": 1.3394,
+      "step": 4611
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.844504606315173e-05,
+      "loss": 1.7649,
+      "step": 4612
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.843166743804167e-05,
+      "loss": 1.6514,
+      "step": 4613
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.841828881293157e-05,
+      "loss": 1.6882,
+      "step": 4614
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.84049101878215e-05,
+      "loss": 1.6298,
+      "step": 4615
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.839153156271143e-05,
+      "loss": 1.3563,
+      "step": 4616
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.837815293760135e-05,
+      "loss": 1.3278,
+      "step": 4617
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.836477431249128e-05,
+      "loss": 1.3216,
+      "step": 4618
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.83513956873812e-05,
+      "loss": 1.7827,
+      "step": 4619
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.833801706227112e-05,
+      "loss": 1.5944,
+      "step": 4620
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.832463843716104e-05,
+      "loss": 1.3134,
+      "step": 4621
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.831125981205096e-05,
+      "loss": 1.3511,
+      "step": 4622
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.82978811869409e-05,
+      "loss": 1.1936,
+      "step": 4623
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.828450256183081e-05,
+      "loss": 1.1805,
+      "step": 4624
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.827112393672073e-05,
+      "loss": 1.2945,
+      "step": 4625
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.825774531161067e-05,
+      "loss": 1.3204,
+      "step": 4626
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.824436668650059e-05,
+      "loss": 1.3049,
+      "step": 4627
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.823098806139052e-05,
+      "loss": 1.2247,
+      "step": 4628
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.821760943628043e-05,
+      "loss": 1.1679,
+      "step": 4629
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.820423081117035e-05,
+      "loss": 1.106,
+      "step": 4630
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.819085218606028e-05,
+      "loss": 1.0776,
+      "step": 4631
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.81774735609502e-05,
+      "loss": 1.0456,
+      "step": 4632
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.816409493584012e-05,
+      "loss": 1.0907,
+      "step": 4633
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.815071631073005e-05,
+      "loss": 1.0345,
+      "step": 4634
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.813733768561997e-05,
+      "loss": 1.1438,
+      "step": 4635
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.812395906050991e-05,
+      "loss": 1.0132,
+      "step": 4636
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.811058043539981e-05,
+      "loss": 1.0652,
+      "step": 4637
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.809720181028973e-05,
+      "loss": 1.0448,
+      "step": 4638
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.808382318517967e-05,
+      "loss": 0.7745,
+      "step": 4639
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.807044456006959e-05,
+      "loss": 0.9772,
+      "step": 4640
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.805706593495951e-05,
+      "loss": 2.0769,
+      "step": 4641
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.804368730984944e-05,
+      "loss": 1.8925,
+      "step": 4642
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.803030868473936e-05,
+      "loss": 1.7895,
+      "step": 4643
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.801693005962928e-05,
+      "loss": 1.7579,
+      "step": 4644
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.80035514345192e-05,
+      "loss": 1.7795,
+      "step": 4645
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.799017280940912e-05,
+      "loss": 1.6619,
+      "step": 4646
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.797679418429905e-05,
+      "loss": 1.6786,
+      "step": 4647
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.796341555918897e-05,
+      "loss": 1.8346,
+      "step": 4648
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.795003693407891e-05,
+      "loss": 1.4797,
+      "step": 4649
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.793665830896883e-05,
+      "loss": 1.7547,
+      "step": 4650
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.792327968385875e-05,
+      "loss": 1.5404,
+      "step": 4651
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.790990105874867e-05,
+      "loss": 1.6847,
+      "step": 4652
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.789652243363859e-05,
+      "loss": 1.5316,
+      "step": 4653
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.788314380852851e-05,
+      "loss": 1.4622,
+      "step": 4654
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.786976518341844e-05,
+      "loss": 1.6367,
+      "step": 4655
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.785638655830836e-05,
+      "loss": 1.401,
+      "step": 4656
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.78430079331983e-05,
+      "loss": 1.5885,
+      "step": 4657
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.782962930808822e-05,
+      "loss": 1.6626,
+      "step": 4658
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.781625068297814e-05,
+      "loss": 1.7027,
+      "step": 4659
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.780287205786806e-05,
+      "loss": 1.5372,
+      "step": 4660
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.778949343275798e-05,
+      "loss": 1.5237,
+      "step": 4661
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.777611480764791e-05,
+      "loss": 1.6363,
+      "step": 4662
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.776273618253783e-05,
+      "loss": 1.4805,
+      "step": 4663
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.774935755742775e-05,
+      "loss": 1.4567,
+      "step": 4664
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.773597893231768e-05,
+      "loss": 1.4529,
+      "step": 4665
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.77226003072076e-05,
+      "loss": 1.3311,
+      "step": 4666
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.770922168209751e-05,
+      "loss": 1.3959,
+      "step": 4667
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.769584305698744e-05,
+      "loss": 1.3731,
+      "step": 4668
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.768246443187736e-05,
+      "loss": 1.5866,
+      "step": 4669
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.76690858067673e-05,
+      "loss": 1.3088,
+      "step": 4670
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.765570718165722e-05,
+      "loss": 1.2507,
+      "step": 4671
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.764232855654714e-05,
+      "loss": 1.4477,
+      "step": 4672
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.762894993143707e-05,
+      "loss": 1.3502,
+      "step": 4673
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.761557130632699e-05,
+      "loss": 1.3915,
+      "step": 4674
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.76021926812169e-05,
+      "loss": 1.1663,
+      "step": 4675
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.758881405610683e-05,
+      "loss": 1.2087,
+      "step": 4676
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.757543543099675e-05,
+      "loss": 1.2523,
+      "step": 4677
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.756205680588668e-05,
+      "loss": 1.0668,
+      "step": 4678
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.75486781807766e-05,
+      "loss": 1.0945,
+      "step": 4679
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.753529955566652e-05,
+      "loss": 1.2961,
+      "step": 4680
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.752192093055646e-05,
+      "loss": 1.0996,
+      "step": 4681
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.750854230544638e-05,
+      "loss": 1.2516,
+      "step": 4682
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.74951636803363e-05,
+      "loss": 1.1068,
+      "step": 4683
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.748178505522622e-05,
+      "loss": 1.0622,
+      "step": 4684
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.746840643011614e-05,
+      "loss": 0.8968,
+      "step": 4685
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.745502780500607e-05,
+      "loss": 0.9663,
+      "step": 4686
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.744164917989599e-05,
+      "loss": 0.9303,
+      "step": 4687
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.742827055478591e-05,
+      "loss": 0.9856,
+      "step": 4688
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.741489192967584e-05,
+      "loss": 0.8368,
+      "step": 4689
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.740151330456576e-05,
+      "loss": 0.7814,
+      "step": 4690
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.738813467945568e-05,
+      "loss": 2.1924,
+      "step": 4691
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.73747560543456e-05,
+      "loss": 1.7965,
+      "step": 4692
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.736137742923552e-05,
+      "loss": 1.8912,
+      "step": 4693
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.734799880412546e-05,
+      "loss": 1.7824,
+      "step": 4694
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.733462017901538e-05,
+      "loss": 1.574,
+      "step": 4695
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.732124155390531e-05,
+      "loss": 1.9138,
+      "step": 4696
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.730786292879523e-05,
+      "loss": 1.7371,
+      "step": 4697
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.729448430368514e-05,
+      "loss": 1.6367,
+      "step": 4698
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.728110567857507e-05,
+      "loss": 1.5872,
+      "step": 4699
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.726772705346499e-05,
+      "loss": 1.506,
+      "step": 4700
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.725434842835491e-05,
+      "loss": 1.6094,
+      "step": 4701
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.724096980324484e-05,
+      "loss": 1.5982,
+      "step": 4702
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.722759117813476e-05,
+      "loss": 1.5578,
+      "step": 4703
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.72142125530247e-05,
+      "loss": 1.652,
+      "step": 4704
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.720083392791462e-05,
+      "loss": 1.5249,
+      "step": 4705
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.718745530280452e-05,
+      "loss": 1.518,
+      "step": 4706
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.717407667769446e-05,
+      "loss": 1.4963,
+      "step": 4707
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.716069805258438e-05,
+      "loss": 1.5358,
+      "step": 4708
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.71473194274743e-05,
+      "loss": 1.4092,
+      "step": 4709
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.713394080236423e-05,
+      "loss": 1.4944,
+      "step": 4710
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.712056217725415e-05,
+      "loss": 1.7304,
+      "step": 4711
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.710718355214409e-05,
+      "loss": 1.5141,
+      "step": 4712
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.7093804927034e-05,
+      "loss": 1.5134,
+      "step": 4713
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.708042630192391e-05,
+      "loss": 1.6096,
+      "step": 4714
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.706704767681385e-05,
+      "loss": 1.3834,
+      "step": 4715
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.705366905170377e-05,
+      "loss": 1.5654,
+      "step": 4716
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.70402904265937e-05,
+      "loss": 1.3401,
+      "step": 4717
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.702691180148362e-05,
+      "loss": 1.4417,
+      "step": 4718
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.701353317637354e-05,
+      "loss": 1.3242,
+      "step": 4719
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.700015455126347e-05,
+      "loss": 1.3982,
+      "step": 4720
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.698677592615338e-05,
+      "loss": 1.2695,
+      "step": 4721
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.69733973010433e-05,
+      "loss": 1.3637,
+      "step": 4722
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.696001867593323e-05,
+      "loss": 1.563,
+      "step": 4723
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.694664005082315e-05,
+      "loss": 1.2659,
+      "step": 4724
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.693326142571309e-05,
+      "loss": 1.2107,
+      "step": 4725
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.6919882800603e-05,
+      "loss": 1.2051,
+      "step": 4726
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.690650417549293e-05,
+      "loss": 1.196,
+      "step": 4727
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.689312555038286e-05,
+      "loss": 1.3136,
+      "step": 4728
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.687974692527277e-05,
+      "loss": 1.2431,
+      "step": 4729
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.68663683001627e-05,
+      "loss": 0.9248,
+      "step": 4730
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.685298967505262e-05,
+      "loss": 1.2333,
+      "step": 4731
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.683961104994254e-05,
+      "loss": 1.2264,
+      "step": 4732
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.682623242483247e-05,
+      "loss": 1.1823,
+      "step": 4733
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.681285379972239e-05,
+      "loss": 1.2401,
+      "step": 4734
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.679947517461231e-05,
+      "loss": 0.8921,
+      "step": 4735
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.678609654950225e-05,
+      "loss": 0.9814,
+      "step": 4736
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.677271792439215e-05,
+      "loss": 0.9165,
+      "step": 4737
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.675933929928209e-05,
+      "loss": 0.8534,
+      "step": 4738
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.6745960674172e-05,
+      "loss": 0.8291,
+      "step": 4739
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.673258204906193e-05,
+      "loss": 0.9896,
+      "step": 4740
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.671920342395186e-05,
+      "loss": 2.1427,
+      "step": 4741
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.670582479884178e-05,
+      "loss": 2.0132,
+      "step": 4742
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.669244617373171e-05,
+      "loss": 1.8143,
+      "step": 4743
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.667906754862163e-05,
+      "loss": 1.6068,
+      "step": 4744
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.666568892351154e-05,
+      "loss": 1.855,
+      "step": 4745
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.665231029840147e-05,
+      "loss": 1.572,
+      "step": 4746
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.66389316732914e-05,
+      "loss": 1.655,
+      "step": 4747
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.662555304818131e-05,
+      "loss": 1.5481,
+      "step": 4748
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.661217442307125e-05,
+      "loss": 1.4613,
+      "step": 4749
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.659879579796117e-05,
+      "loss": 1.5591,
+      "step": 4750
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.65854171728511e-05,
+      "loss": 1.6457,
+      "step": 4751
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.657203854774101e-05,
+      "loss": 1.6314,
+      "step": 4752
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.655865992263093e-05,
+      "loss": 1.474,
+      "step": 4753
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.654528129752086e-05,
+      "loss": 1.3628,
+      "step": 4754
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.653190267241078e-05,
+      "loss": 1.6703,
+      "step": 4755
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.65185240473007e-05,
+      "loss": 1.6677,
+      "step": 4756
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.650514542219063e-05,
+      "loss": 1.5427,
+      "step": 4757
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.649176679708055e-05,
+      "loss": 1.2052,
+      "step": 4758
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.647838817197049e-05,
+      "loss": 1.4739,
+      "step": 4759
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.64650095468604e-05,
+      "loss": 1.6927,
+      "step": 4760
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.645163092175031e-05,
+      "loss": 1.7457,
+      "step": 4761
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.643825229664025e-05,
+      "loss": 1.4634,
+      "step": 4762
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.642487367153017e-05,
+      "loss": 1.4495,
+      "step": 4763
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.64114950464201e-05,
+      "loss": 1.432,
+      "step": 4764
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.639811642131002e-05,
+      "loss": 1.471,
+      "step": 4765
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.638473779619994e-05,
+      "loss": 1.3594,
+      "step": 4766
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.637135917108988e-05,
+      "loss": 1.4562,
+      "step": 4767
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.635798054597978e-05,
+      "loss": 1.381,
+      "step": 4768
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.63446019208697e-05,
+      "loss": 1.2479,
+      "step": 4769
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.633122329575964e-05,
+      "loss": 1.3754,
+      "step": 4770
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.631784467064956e-05,
+      "loss": 1.2419,
+      "step": 4771
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.630446604553949e-05,
+      "loss": 1.288,
+      "step": 4772
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.629108742042941e-05,
+      "loss": 1.2315,
+      "step": 4773
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.627770879531933e-05,
+      "loss": 1.3502,
+      "step": 4774
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.626433017020925e-05,
+      "loss": 1.3569,
+      "step": 4775
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.625095154509917e-05,
+      "loss": 1.2259,
+      "step": 4776
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.62375729199891e-05,
+      "loss": 1.1899,
+      "step": 4777
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.622419429487902e-05,
+      "loss": 1.2454,
+      "step": 4778
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.621081566976894e-05,
+      "loss": 1.2679,
+      "step": 4779
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.619743704465888e-05,
+      "loss": 1.1301,
+      "step": 4780
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.61840584195488e-05,
+      "loss": 1.0582,
+      "step": 4781
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.617067979443872e-05,
+      "loss": 1.2074,
+      "step": 4782
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.615730116932864e-05,
+      "loss": 1.1246,
+      "step": 4783
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.614392254421856e-05,
+      "loss": 1.223,
+      "step": 4784
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.613054391910849e-05,
+      "loss": 0.8835,
+      "step": 4785
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.611716529399841e-05,
+      "loss": 1.1411,
+      "step": 4786
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.610378666888833e-05,
+      "loss": 1.0892,
+      "step": 4787
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.609040804377826e-05,
+      "loss": 1.033,
+      "step": 4788
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.607702941866818e-05,
+      "loss": 0.902,
+      "step": 4789
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.60636507935581e-05,
+      "loss": 0.7173,
+      "step": 4790
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.605027216844802e-05,
+      "loss": 1.9904,
+      "step": 4791
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.603689354333794e-05,
+      "loss": 2.3129,
+      "step": 4792
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.602351491822788e-05,
+      "loss": 1.7786,
+      "step": 4793
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.60101362931178e-05,
+      "loss": 1.8583,
+      "step": 4794
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.599675766800772e-05,
+      "loss": 1.82,
+      "step": 4795
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.598337904289765e-05,
+      "loss": 1.5552,
+      "step": 4796
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.597000041778757e-05,
+      "loss": 1.6037,
+      "step": 4797
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.59566217926775e-05,
+      "loss": 1.8813,
+      "step": 4798
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.594324316756741e-05,
+      "loss": 1.5308,
+      "step": 4799
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.592986454245733e-05,
+      "loss": 1.5417,
+      "step": 4800
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.591648591734726e-05,
+      "loss": 1.571,
+      "step": 4801
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.590310729223718e-05,
+      "loss": 1.5247,
+      "step": 4802
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.58897286671271e-05,
+      "loss": 1.5331,
+      "step": 4803
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.587635004201704e-05,
+      "loss": 1.6629,
+      "step": 4804
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.586297141690696e-05,
+      "loss": 1.6926,
+      "step": 4805
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.584959279179688e-05,
+      "loss": 1.6286,
+      "step": 4806
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.58362141666868e-05,
+      "loss": 1.6019,
+      "step": 4807
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.582283554157672e-05,
+      "loss": 1.7294,
+      "step": 4808
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.580945691646665e-05,
+      "loss": 1.5603,
+      "step": 4809
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.579607829135657e-05,
+      "loss": 1.3469,
+      "step": 4810
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.57826996662465e-05,
+      "loss": 1.6821,
+      "step": 4811
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.576932104113642e-05,
+      "loss": 1.5687,
+      "step": 4812
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.575594241602634e-05,
+      "loss": 1.4757,
+      "step": 4813
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.574256379091626e-05,
+      "loss": 1.4487,
+      "step": 4814
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.572918516580618e-05,
+      "loss": 1.4646,
+      "step": 4815
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.57158065406961e-05,
+      "loss": 1.5422,
+      "step": 4816
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.570242791558604e-05,
+      "loss": 1.3196,
+      "step": 4817
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.568904929047596e-05,
+      "loss": 1.4404,
+      "step": 4818
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.567567066536589e-05,
+      "loss": 1.5014,
+      "step": 4819
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.566229204025581e-05,
+      "loss": 1.3495,
+      "step": 4820
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.564891341514573e-05,
+      "loss": 1.3525,
+      "step": 4821
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.563553479003565e-05,
+      "loss": 1.3861,
+      "step": 4822
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.562215616492557e-05,
+      "loss": 1.3533,
+      "step": 4823
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.56087775398155e-05,
+      "loss": 1.2134,
+      "step": 4824
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.559539891470542e-05,
+      "loss": 1.1725,
+      "step": 4825
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.558202028959534e-05,
+      "loss": 1.3335,
+      "step": 4826
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.556864166448528e-05,
+      "loss": 1.2196,
+      "step": 4827
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.55552630393752e-05,
+      "loss": 1.27,
+      "step": 4828
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.55418844142651e-05,
+      "loss": 1.1415,
+      "step": 4829
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.552850578915504e-05,
+      "loss": 1.2679,
+      "step": 4830
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.551512716404496e-05,
+      "loss": 1.2023,
+      "step": 4831
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.550174853893489e-05,
+      "loss": 1.0472,
+      "step": 4832
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.548836991382481e-05,
+      "loss": 1.0577,
+      "step": 4833
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.547499128871473e-05,
+      "loss": 1.0066,
+      "step": 4834
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.546161266360467e-05,
+      "loss": 1.0353,
+      "step": 4835
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.544823403849459e-05,
+      "loss": 0.9089,
+      "step": 4836
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.543485541338449e-05,
+      "loss": 0.9714,
+      "step": 4837
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.542147678827443e-05,
+      "loss": 0.9175,
+      "step": 4838
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.540809816316435e-05,
+      "loss": 0.8031,
+      "step": 4839
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.539471953805428e-05,
+      "loss": 0.9029,
+      "step": 4840
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.53813409129442e-05,
+      "loss": 2.1645,
+      "step": 4841
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.536796228783412e-05,
+      "loss": 1.864,
+      "step": 4842
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.535458366272405e-05,
+      "loss": 1.9474,
+      "step": 4843
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.534120503761397e-05,
+      "loss": 1.7077,
+      "step": 4844
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.532782641250389e-05,
+      "loss": 1.6226,
+      "step": 4845
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.531444778739381e-05,
+      "loss": 1.5898,
+      "step": 4846
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.530106916228373e-05,
+      "loss": 1.5232,
+      "step": 4847
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.528769053717367e-05,
+      "loss": 1.6503,
+      "step": 4848
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.527431191206359e-05,
+      "loss": 1.7408,
+      "step": 4849
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.52609332869535e-05,
+      "loss": 1.6614,
+      "step": 4850
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.524755466184344e-05,
+      "loss": 1.5223,
+      "step": 4851
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.523417603673336e-05,
+      "loss": 1.4567,
+      "step": 4852
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.522079741162328e-05,
+      "loss": 1.4666,
+      "step": 4853
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.52074187865132e-05,
+      "loss": 1.4848,
+      "step": 4854
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.519404016140312e-05,
+      "loss": 1.755,
+      "step": 4855
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.518066153629305e-05,
+      "loss": 1.7871,
+      "step": 4856
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.516728291118297e-05,
+      "loss": 1.4012,
+      "step": 4857
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.515390428607291e-05,
+      "loss": 1.1948,
+      "step": 4858
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.514052566096283e-05,
+      "loss": 1.5647,
+      "step": 4859
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.512714703585273e-05,
+      "loss": 1.4496,
+      "step": 4860
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.511376841074267e-05,
+      "loss": 1.531,
+      "step": 4861
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.510038978563259e-05,
+      "loss": 1.3851,
+      "step": 4862
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.508701116052251e-05,
+      "loss": 1.6411,
+      "step": 4863
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.507363253541244e-05,
+      "loss": 1.5395,
+      "step": 4864
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.506025391030236e-05,
+      "loss": 1.3943,
+      "step": 4865
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.50468752851923e-05,
+      "loss": 1.4434,
+      "step": 4866
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.503349666008221e-05,
+      "loss": 1.3877,
+      "step": 4867
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.502011803497212e-05,
+      "loss": 1.2371,
+      "step": 4868
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.500673940986205e-05,
+      "loss": 1.3479,
+      "step": 4869
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.499336078475197e-05,
+      "loss": 1.3774,
+      "step": 4870
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.49799821596419e-05,
+      "loss": 1.2015,
+      "step": 4871
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.496660353453183e-05,
+      "loss": 1.2774,
+      "step": 4872
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.495322490942175e-05,
+      "loss": 1.2566,
+      "step": 4873
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.493984628431168e-05,
+      "loss": 1.2547,
+      "step": 4874
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.49264676592016e-05,
+      "loss": 1.4861,
+      "step": 4875
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.491308903409151e-05,
+      "loss": 1.3377,
+      "step": 4876
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.489971040898144e-05,
+      "loss": 1.2505,
+      "step": 4877
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.488633178387136e-05,
+      "loss": 1.1414,
+      "step": 4878
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.48729531587613e-05,
+      "loss": 1.1166,
+      "step": 4879
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.485957453365121e-05,
+      "loss": 1.0685,
+      "step": 4880
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.484619590854113e-05,
+      "loss": 1.1768,
+      "step": 4881
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.483281728343107e-05,
+      "loss": 1.0092,
+      "step": 4882
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.481943865832097e-05,
+      "loss": 0.9274,
+      "step": 4883
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.48060600332109e-05,
+      "loss": 0.8301,
+      "step": 4884
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.479268140810083e-05,
+      "loss": 0.8413,
+      "step": 4885
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.477930278299075e-05,
+      "loss": 0.928,
+      "step": 4886
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.476592415788068e-05,
+      "loss": 0.8149,
+      "step": 4887
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.47525455327706e-05,
+      "loss": 0.7333,
+      "step": 4888
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.473916690766052e-05,
+      "loss": 0.9468,
+      "step": 4889
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.472578828255046e-05,
+      "loss": 0.9495,
+      "step": 4890
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.471240965744036e-05,
+      "loss": 2.1692,
+      "step": 4891
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.46990310323303e-05,
+      "loss": 1.9684,
+      "step": 4892
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.468565240722022e-05,
+      "loss": 1.788,
+      "step": 4893
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.467227378211014e-05,
+      "loss": 1.8715,
+      "step": 4894
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.465889515700007e-05,
+      "loss": 1.7169,
+      "step": 4895
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.464551653188999e-05,
+      "loss": 1.7002,
+      "step": 4896
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.463213790677991e-05,
+      "loss": 1.6222,
+      "step": 4897
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.461875928166984e-05,
+      "loss": 1.7376,
+      "step": 4898
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.460538065655975e-05,
+      "loss": 1.7216,
+      "step": 4899
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.459200203144968e-05,
+      "loss": 1.5237,
+      "step": 4900
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.45786234063396e-05,
+      "loss": 1.641,
+      "step": 4901
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.456524478122952e-05,
+      "loss": 1.7669,
+      "step": 4902
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.455186615611946e-05,
+      "loss": 1.5182,
+      "step": 4903
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.453848753100938e-05,
+      "loss": 1.6543,
+      "step": 4904
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.45251089058993e-05,
+      "loss": 1.5356,
+      "step": 4905
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.451173028078923e-05,
+      "loss": 1.5121,
+      "step": 4906
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.449835165567914e-05,
+      "loss": 1.5481,
+      "step": 4907
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.448497303056907e-05,
+      "loss": 1.4789,
+      "step": 4908
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.447159440545899e-05,
+      "loss": 1.4694,
+      "step": 4909
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.445821578034891e-05,
+      "loss": 1.4045,
+      "step": 4910
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.444483715523884e-05,
+      "loss": 1.4346,
+      "step": 4911
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.443145853012876e-05,
+      "loss": 1.5455,
+      "step": 4912
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.44180799050187e-05,
+      "loss": 1.4813,
+      "step": 4913
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.44047012799086e-05,
+      "loss": 1.3904,
+      "step": 4914
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.439132265479852e-05,
+      "loss": 1.4074,
+      "step": 4915
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.437794402968846e-05,
+      "loss": 1.5203,
+      "step": 4916
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.436456540457838e-05,
+      "loss": 1.405,
+      "step": 4917
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.43511867794683e-05,
+      "loss": 1.3342,
+      "step": 4918
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.433780815435823e-05,
+      "loss": 1.3329,
+      "step": 4919
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.432442952924815e-05,
+      "loss": 1.4575,
+      "step": 4920
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.431105090413808e-05,
+      "loss": 1.2537,
+      "step": 4921
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.429767227902799e-05,
+      "loss": 1.2203,
+      "step": 4922
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.428429365391791e-05,
+      "loss": 1.2578,
+      "step": 4923
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.427091502880784e-05,
+      "loss": 0.9441,
+      "step": 4924
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.425753640369776e-05,
+      "loss": 1.4951,
+      "step": 4925
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.42441577785877e-05,
+      "loss": 1.4511,
+      "step": 4926
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.423077915347762e-05,
+      "loss": 1.3423,
+      "step": 4927
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.421740052836754e-05,
+      "loss": 1.1211,
+      "step": 4928
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.420402190325747e-05,
+      "loss": 1.2817,
+      "step": 4929
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.419064327814738e-05,
+      "loss": 1.1435,
+      "step": 4930
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.41772646530373e-05,
+      "loss": 1.0633,
+      "step": 4931
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.416388602792723e-05,
+      "loss": 1.0803,
+      "step": 4932
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.415050740281715e-05,
+      "loss": 1.0856,
+      "step": 4933
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.413712877770708e-05,
+      "loss": 0.8894,
+      "step": 4934
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.4123750152597e-05,
+      "loss": 1.0657,
+      "step": 4935
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.411037152748692e-05,
+      "loss": 0.942,
+      "step": 4936
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.409699290237684e-05,
+      "loss": 0.8134,
+      "step": 4937
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.408361427726676e-05,
+      "loss": 0.8599,
+      "step": 4938
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.40702356521567e-05,
+      "loss": 0.7015,
+      "step": 4939
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.405685702704662e-05,
+      "loss": 0.9382,
+      "step": 4940
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.404347840193654e-05,
+      "loss": 1.9341,
+      "step": 4941
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.403009977682647e-05,
+      "loss": 1.8288,
+      "step": 4942
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.401672115171639e-05,
+      "loss": 1.556,
+      "step": 4943
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.400334252660631e-05,
+      "loss": 1.6348,
+      "step": 4944
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.398996390149623e-05,
+      "loss": 1.731,
+      "step": 4945
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.397658527638615e-05,
+      "loss": 1.6185,
+      "step": 4946
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.396320665127609e-05,
+      "loss": 1.3923,
+      "step": 4947
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.3949828026166e-05,
+      "loss": 1.5971,
+      "step": 4948
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.393644940105593e-05,
+      "loss": 1.5835,
+      "step": 4949
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.392307077594586e-05,
+      "loss": 1.4704,
+      "step": 4950
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.390969215083578e-05,
+      "loss": 1.6373,
+      "step": 4951
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.38963135257257e-05,
+      "loss": 1.5622,
+      "step": 4952
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.388293490061562e-05,
+      "loss": 1.531,
+      "step": 4953
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.386955627550554e-05,
+      "loss": 1.4928,
+      "step": 4954
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.385617765039547e-05,
+      "loss": 1.4451,
+      "step": 4955
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.384279902528539e-05,
+      "loss": 1.6912,
+      "step": 4956
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.382942040017531e-05,
+      "loss": 1.5256,
+      "step": 4957
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.381604177506525e-05,
+      "loss": 1.5378,
+      "step": 4958
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.380266314995517e-05,
+      "loss": 1.637,
+      "step": 4959
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.37892845248451e-05,
+      "loss": 1.6286,
+      "step": 4960
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.3775905899735e-05,
+      "loss": 1.296,
+      "step": 4961
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.376252727462493e-05,
+      "loss": 1.6385,
+      "step": 4962
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.374914864951486e-05,
+      "loss": 1.5399,
+      "step": 4963
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.373577002440478e-05,
+      "loss": 1.351,
+      "step": 4964
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.37223913992947e-05,
+      "loss": 1.1955,
+      "step": 4965
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.370901277418463e-05,
+      "loss": 1.422,
+      "step": 4966
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.369563414907455e-05,
+      "loss": 1.3284,
+      "step": 4967
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.368225552396447e-05,
+      "loss": 1.4364,
+      "step": 4968
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.366887689885439e-05,
+      "loss": 1.3497,
+      "step": 4969
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.365549827374431e-05,
+      "loss": 1.6789,
+      "step": 4970
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.364211964863425e-05,
+      "loss": 1.1267,
+      "step": 4971
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.362874102352417e-05,
+      "loss": 1.4462,
+      "step": 4972
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.36153623984141e-05,
+      "loss": 1.2144,
+      "step": 4973
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.360198377330402e-05,
+      "loss": 1.3141,
+      "step": 4974
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.358860514819394e-05,
+      "loss": 1.3447,
+      "step": 4975
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.357522652308386e-05,
+      "loss": 1.1843,
+      "step": 4976
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.356184789797378e-05,
+      "loss": 1.2614,
+      "step": 4977
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.35484692728637e-05,
+      "loss": 1.2353,
+      "step": 4978
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.353509064775363e-05,
+      "loss": 1.1734,
+      "step": 4979
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.352171202264355e-05,
+      "loss": 1.0906,
+      "step": 4980
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.350833339753349e-05,
+      "loss": 1.164,
+      "step": 4981
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.349495477242341e-05,
+      "loss": 0.9469,
+      "step": 4982
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.348157614731333e-05,
+      "loss": 1.065,
+      "step": 4983
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.346819752220325e-05,
+      "loss": 0.8774,
+      "step": 4984
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.345481889709317e-05,
+      "loss": 1.0154,
+      "step": 4985
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.344144027198309e-05,
+      "loss": 0.847,
+      "step": 4986
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.342806164687302e-05,
+      "loss": 1.0347,
+      "step": 4987
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.341468302176294e-05,
+      "loss": 0.8735,
+      "step": 4988
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.340130439665287e-05,
+      "loss": 0.9285,
+      "step": 4989
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.33879257715428e-05,
+      "loss": 0.8829,
+      "step": 4990
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.33745471464327e-05,
+      "loss": 2.0452,
+      "step": 4991
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.336116852132263e-05,
+      "loss": 1.9473,
+      "step": 4992
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.334778989621255e-05,
+      "loss": 1.7906,
+      "step": 4993
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.333441127110249e-05,
+      "loss": 1.6987,
+      "step": 4994
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.332103264599241e-05,
+      "loss": 1.6764,
+      "step": 4995
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.330765402088233e-05,
+      "loss": 1.732,
+      "step": 4996
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.329427539577226e-05,
+      "loss": 1.8032,
+      "step": 4997
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.328089677066218e-05,
+      "loss": 1.781,
+      "step": 4998
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.326751814555209e-05,
+      "loss": 1.5821,
+      "step": 4999
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.325413952044202e-05,
+      "loss": 1.5955,
+      "step": 5000
+    },
+    {
+      "epoch": 1.54,
+      "eval_bleu": 2.3356841069051653e-16,
+      "eval_loss": 1.817099690437317,
+      "eval_runtime": 2568.5571,
+      "eval_samples_per_second": 5.746,
+      "eval_steps_per_second": 0.718,
+      "step": 5000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.324076089533194e-05,
+      "loss": 1.5781,
+      "step": 5001
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.322738227022188e-05,
+      "loss": 1.8052,
+      "step": 5002
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.32140036451118e-05,
+      "loss": 1.7081,
+      "step": 5003
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.320062502000172e-05,
+      "loss": 1.5154,
+      "step": 5004
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.318724639489165e-05,
+      "loss": 1.5801,
+      "step": 5005
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.317386776978157e-05,
+      "loss": 1.3162,
+      "step": 5006
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.316048914467149e-05,
+      "loss": 1.4229,
+      "step": 5007
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.314711051956141e-05,
+      "loss": 1.4708,
+      "step": 5008
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.313373189445133e-05,
+      "loss": 1.5269,
+      "step": 5009
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.312035326934126e-05,
+      "loss": 1.502,
+      "step": 5010
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.310697464423118e-05,
+      "loss": 1.6309,
+      "step": 5011
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.30935960191211e-05,
+      "loss": 1.6402,
+      "step": 5012
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.308021739401104e-05,
+      "loss": 1.3892,
+      "step": 5013
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.306683876890094e-05,
+      "loss": 1.5594,
+      "step": 5014
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.305346014379088e-05,
+      "loss": 1.4291,
+      "step": 5015
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.30400815186808e-05,
+      "loss": 1.3721,
+      "step": 5016
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.302670289357072e-05,
+      "loss": 1.4082,
+      "step": 5017
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.301332426846065e-05,
+      "loss": 1.4923,
+      "step": 5018
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.299994564335057e-05,
+      "loss": 1.4851,
+      "step": 5019
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.29865670182405e-05,
+      "loss": 1.1311,
+      "step": 5020
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.297318839313042e-05,
+      "loss": 1.236,
+      "step": 5021
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.295980976802033e-05,
+      "loss": 1.251,
+      "step": 5022
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.294643114291026e-05,
+      "loss": 1.3848,
+      "step": 5023
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.293305251780018e-05,
+      "loss": 1.1513,
+      "step": 5024
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.29196738926901e-05,
+      "loss": 1.292,
+      "step": 5025
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.290629526758004e-05,
+      "loss": 1.1423,
+      "step": 5026
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.289291664246996e-05,
+      "loss": 1.1535,
+      "step": 5027
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.287953801735989e-05,
+      "loss": 1.323,
+      "step": 5028
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.286615939224981e-05,
+      "loss": 1.1745,
+      "step": 5029
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.285278076713972e-05,
+      "loss": 1.045,
+      "step": 5030
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.283940214202965e-05,
+      "loss": 1.2318,
+      "step": 5031
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.282602351691957e-05,
+      "loss": 1.2626,
+      "step": 5032
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.281264489180949e-05,
+      "loss": 1.1117,
+      "step": 5033
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.279926626669942e-05,
+      "loss": 0.9668,
+      "step": 5034
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.278588764158934e-05,
+      "loss": 0.8903,
+      "step": 5035
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.277250901647928e-05,
+      "loss": 0.9147,
+      "step": 5036
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.27591303913692e-05,
+      "loss": 0.9231,
+      "step": 5037
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.27457517662591e-05,
+      "loss": 0.9547,
+      "step": 5038
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.273237314114904e-05,
+      "loss": 0.8629,
+      "step": 5039
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.271899451603896e-05,
+      "loss": 0.8151,
+      "step": 5040
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.270561589092889e-05,
+      "loss": 2.1027,
+      "step": 5041
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.269223726581881e-05,
+      "loss": 1.8255,
+      "step": 5042
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.267885864070873e-05,
+      "loss": 1.8818,
+      "step": 5043
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.266548001559866e-05,
+      "loss": 1.6993,
+      "step": 5044
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.265210139048857e-05,
+      "loss": 1.534,
+      "step": 5045
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.263872276537849e-05,
+      "loss": 1.63,
+      "step": 5046
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.262534414026842e-05,
+      "loss": 1.6606,
+      "step": 5047
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.261196551515834e-05,
+      "loss": 1.5557,
+      "step": 5048
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.259858689004828e-05,
+      "loss": 1.7196,
+      "step": 5049
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.25852082649382e-05,
+      "loss": 1.4864,
+      "step": 5050
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.257182963982812e-05,
+      "loss": 1.5357,
+      "step": 5051
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.255845101471805e-05,
+      "loss": 1.6205,
+      "step": 5052
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.254507238960796e-05,
+      "loss": 1.6588,
+      "step": 5053
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.253169376449789e-05,
+      "loss": 1.4589,
+      "step": 5054
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.251831513938781e-05,
+      "loss": 1.5451,
+      "step": 5055
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.250493651427773e-05,
+      "loss": 1.5708,
+      "step": 5056
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.249155788916766e-05,
+      "loss": 1.5031,
+      "step": 5057
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.247817926405758e-05,
+      "loss": 1.6337,
+      "step": 5058
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.24648006389475e-05,
+      "loss": 1.3616,
+      "step": 5059
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.245142201383744e-05,
+      "loss": 1.5696,
+      "step": 5060
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.243804338872734e-05,
+      "loss": 1.4985,
+      "step": 5061
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.242466476361728e-05,
+      "loss": 1.554,
+      "step": 5062
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.24112861385072e-05,
+      "loss": 1.4279,
+      "step": 5063
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.239790751339712e-05,
+      "loss": 1.3824,
+      "step": 5064
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.238452888828705e-05,
+      "loss": 1.2575,
+      "step": 5065
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.237115026317697e-05,
+      "loss": 1.3506,
+      "step": 5066
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.235777163806689e-05,
+      "loss": 1.141,
+      "step": 5067
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.234439301295681e-05,
+      "loss": 1.3749,
+      "step": 5068
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.233101438784673e-05,
+      "loss": 1.0558,
+      "step": 5069
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.231763576273667e-05,
+      "loss": 1.2072,
+      "step": 5070
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.230425713762659e-05,
+      "loss": 1.239,
+      "step": 5071
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.22908785125165e-05,
+      "loss": 1.3135,
+      "step": 5072
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.227749988740644e-05,
+      "loss": 1.2292,
+      "step": 5073
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.226412126229636e-05,
+      "loss": 1.0321,
+      "step": 5074
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.225074263718629e-05,
+      "loss": 1.1312,
+      "step": 5075
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.22373640120762e-05,
+      "loss": 1.3439,
+      "step": 5076
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.222398538696612e-05,
+      "loss": 1.0581,
+      "step": 5077
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.221060676185605e-05,
+      "loss": 1.191,
+      "step": 5078
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.219722813674597e-05,
+      "loss": 1.0999,
+      "step": 5079
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.218384951163589e-05,
+      "loss": 1.1654,
+      "step": 5080
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.217047088652583e-05,
+      "loss": 1.0683,
+      "step": 5081
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.215709226141575e-05,
+      "loss": 1.0007,
+      "step": 5082
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.214371363630568e-05,
+      "loss": 1.1088,
+      "step": 5083
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.213033501119559e-05,
+      "loss": 1.2788,
+      "step": 5084
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.21169563860855e-05,
+      "loss": 1.0818,
+      "step": 5085
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.210357776097544e-05,
+      "loss": 0.9252,
+      "step": 5086
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.209019913586536e-05,
+      "loss": 1.0022,
+      "step": 5087
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.20768205107553e-05,
+      "loss": 0.7139,
+      "step": 5088
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.206344188564521e-05,
+      "loss": 0.7697,
+      "step": 5089
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.205006326053513e-05,
+      "loss": 0.9254,
+      "step": 5090
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.203668463542507e-05,
+      "loss": 1.877,
+      "step": 5091
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.202330601031497e-05,
+      "loss": 1.873,
+      "step": 5092
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.20099273852049e-05,
+      "loss": 1.8146,
+      "step": 5093
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.199654876009483e-05,
+      "loss": 1.7099,
+      "step": 5094
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.198317013498475e-05,
+      "loss": 1.793,
+      "step": 5095
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.196979150987468e-05,
+      "loss": 1.5166,
+      "step": 5096
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.19564128847646e-05,
+      "loss": 1.7547,
+      "step": 5097
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.194303425965452e-05,
+      "loss": 1.7615,
+      "step": 5098
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.192965563454444e-05,
+      "loss": 1.6822,
+      "step": 5099
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.191627700943436e-05,
+      "loss": 1.5281,
+      "step": 5100
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.190289838432428e-05,
+      "loss": 1.56,
+      "step": 5101
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.188951975921421e-05,
+      "loss": 1.552,
+      "step": 5102
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.187614113410413e-05,
+      "loss": 1.6855,
+      "step": 5103
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.186276250899407e-05,
+      "loss": 1.6292,
+      "step": 5104
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.184938388388399e-05,
+      "loss": 1.5546,
+      "step": 5105
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.183600525877391e-05,
+      "loss": 1.5555,
+      "step": 5106
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.182262663366383e-05,
+      "loss": 1.5197,
+      "step": 5107
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.180924800855375e-05,
+      "loss": 1.4922,
+      "step": 5108
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.179586938344368e-05,
+      "loss": 1.427,
+      "step": 5109
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.17824907583336e-05,
+      "loss": 1.6584,
+      "step": 5110
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.176911213322352e-05,
+      "loss": 1.4981,
+      "step": 5111
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.175573350811345e-05,
+      "loss": 1.6362,
+      "step": 5112
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.174235488300337e-05,
+      "loss": 1.626,
+      "step": 5113
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.17289762578933e-05,
+      "loss": 1.4566,
+      "step": 5114
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.171559763278321e-05,
+      "loss": 1.2407,
+      "step": 5115
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.170221900767313e-05,
+      "loss": 1.4167,
+      "step": 5116
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.168884038256307e-05,
+      "loss": 1.3898,
+      "step": 5117
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.167546175745299e-05,
+      "loss": 1.3818,
+      "step": 5118
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.166208313234291e-05,
+      "loss": 1.1145,
+      "step": 5119
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.164870450723284e-05,
+      "loss": 1.2591,
+      "step": 5120
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.163532588212276e-05,
+      "loss": 1.2001,
+      "step": 5121
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.162194725701268e-05,
+      "loss": 1.0165,
+      "step": 5122
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.16085686319026e-05,
+      "loss": 1.4704,
+      "step": 5123
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.159519000679254e-05,
+      "loss": 1.3676,
+      "step": 5124
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.158181138168246e-05,
+      "loss": 1.3686,
+      "step": 5125
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.156843275657238e-05,
+      "loss": 1.1521,
+      "step": 5126
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.15550541314623e-05,
+      "loss": 1.1619,
+      "step": 5127
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.154167550635223e-05,
+      "loss": 1.0596,
+      "step": 5128
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.152829688124215e-05,
+      "loss": 0.9858,
+      "step": 5129
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.151491825613207e-05,
+      "loss": 1.0515,
+      "step": 5130
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.150153963102199e-05,
+      "loss": 1.2491,
+      "step": 5131
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.148816100591192e-05,
+      "loss": 1.1769,
+      "step": 5132
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.147478238080184e-05,
+      "loss": 0.9635,
+      "step": 5133
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.146140375569176e-05,
+      "loss": 1.1079,
+      "step": 5134
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.144802513058168e-05,
+      "loss": 1.058,
+      "step": 5135
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.143464650547162e-05,
+      "loss": 1.1598,
+      "step": 5136
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.142126788036154e-05,
+      "loss": 0.785,
+      "step": 5137
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.140788925525146e-05,
+      "loss": 0.9058,
+      "step": 5138
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.139451063014138e-05,
+      "loss": 0.7126,
+      "step": 5139
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.138113200503131e-05,
+      "loss": 0.7287,
+      "step": 5140
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.136775337992123e-05,
+      "loss": 1.9319,
+      "step": 5141
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.135437475481115e-05,
+      "loss": 1.6959,
+      "step": 5142
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.134099612970107e-05,
+      "loss": 1.7657,
+      "step": 5143
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.1327617504591e-05,
+      "loss": 1.5907,
+      "step": 5144
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.131423887948092e-05,
+      "loss": 1.7099,
+      "step": 5145
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.130086025437084e-05,
+      "loss": 1.5796,
+      "step": 5146
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.128748162926076e-05,
+      "loss": 1.4291,
+      "step": 5147
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.12741030041507e-05,
+      "loss": 1.6542,
+      "step": 5148
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.126072437904062e-05,
+      "loss": 1.5915,
+      "step": 5149
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.124734575393054e-05,
+      "loss": 1.3223,
+      "step": 5150
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.123396712882046e-05,
+      "loss": 1.7025,
+      "step": 5151
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.122058850371039e-05,
+      "loss": 1.5886,
+      "step": 5152
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.120720987860031e-05,
+      "loss": 1.5421,
+      "step": 5153
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.119383125349023e-05,
+      "loss": 1.5135,
+      "step": 5154
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.118045262838015e-05,
+      "loss": 1.4172,
+      "step": 5155
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.116707400327008e-05,
+      "loss": 1.4849,
+      "step": 5156
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.115369537816e-05,
+      "loss": 1.3728,
+      "step": 5157
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.114031675304992e-05,
+      "loss": 1.3467,
+      "step": 5158
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.112693812793984e-05,
+      "loss": 1.4697,
+      "step": 5159
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.111355950282978e-05,
+      "loss": 1.3673,
+      "step": 5160
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.11001808777197e-05,
+      "loss": 1.3591,
+      "step": 5161
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.108680225260962e-05,
+      "loss": 1.4326,
+      "step": 5162
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.107342362749954e-05,
+      "loss": 1.4759,
+      "step": 5163
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.106004500238947e-05,
+      "loss": 1.4592,
+      "step": 5164
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.104666637727939e-05,
+      "loss": 1.4591,
+      "step": 5165
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.103328775216932e-05,
+      "loss": 1.3963,
+      "step": 5166
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.101990912705923e-05,
+      "loss": 1.3594,
+      "step": 5167
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.100653050194916e-05,
+      "loss": 1.3958,
+      "step": 5168
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.0993151876839085e-05,
+      "loss": 1.3857,
+      "step": 5169
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.097977325172901e-05,
+      "loss": 1.3583,
+      "step": 5170
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.096639462661893e-05,
+      "loss": 1.2399,
+      "step": 5171
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.095301600150885e-05,
+      "loss": 1.405,
+      "step": 5172
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.093963737639878e-05,
+      "loss": 1.2126,
+      "step": 5173
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.09262587512887e-05,
+      "loss": 1.2381,
+      "step": 5174
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.0912880126178625e-05,
+      "loss": 1.2042,
+      "step": 5175
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.0899501501068545e-05,
+      "loss": 1.0932,
+      "step": 5176
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.088612287595847e-05,
+      "loss": 1.3304,
+      "step": 5177
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.087274425084839e-05,
+      "loss": 0.9512,
+      "step": 5178
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.085936562573832e-05,
+      "loss": 1.238,
+      "step": 5179
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.084598700062824e-05,
+      "loss": 1.1258,
+      "step": 5180
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.0832608375518165e-05,
+      "loss": 1.1891,
+      "step": 5181
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.0819229750408085e-05,
+      "loss": 0.8875,
+      "step": 5182
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.080585112529801e-05,
+      "loss": 0.9107,
+      "step": 5183
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.079247250018794e-05,
+      "loss": 1.1134,
+      "step": 5184
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.077909387507786e-05,
+      "loss": 0.9293,
+      "step": 5185
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.076571524996778e-05,
+      "loss": 0.9269,
+      "step": 5186
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.0752336624857706e-05,
+      "loss": 0.8314,
+      "step": 5187
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.073895799974763e-05,
+      "loss": 0.6991,
+      "step": 5188
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.072557937463755e-05,
+      "loss": 0.8969,
+      "step": 5189
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.071220074952747e-05,
+      "loss": 0.7452,
+      "step": 5190
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.06988221244174e-05,
+      "loss": 2.1689,
+      "step": 5191
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.0685443499307326e-05,
+      "loss": 1.8607,
+      "step": 5192
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.0672064874197246e-05,
+      "loss": 1.7496,
+      "step": 5193
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.0658686249087166e-05,
+      "loss": 1.7288,
+      "step": 5194
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.064530762397709e-05,
+      "loss": 1.6206,
+      "step": 5195
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.063192899886702e-05,
+      "loss": 1.8621,
+      "step": 5196
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.061855037375693e-05,
+      "loss": 1.719,
+      "step": 5197
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.060517174864686e-05,
+      "loss": 1.5708,
+      "step": 5198
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.0591793123536786e-05,
+      "loss": 1.5845,
+      "step": 5199
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.057841449842671e-05,
+      "loss": 1.3863,
+      "step": 5200
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.056503587331663e-05,
+      "loss": 1.5522,
+      "step": 5201
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.055165724820655e-05,
+      "loss": 1.5933,
+      "step": 5202
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.053827862309648e-05,
+      "loss": 1.4736,
+      "step": 5203
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.052489999798641e-05,
+      "loss": 1.5217,
+      "step": 5204
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.051152137287633e-05,
+      "loss": 1.6491,
+      "step": 5205
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.049814274776625e-05,
+      "loss": 1.6151,
+      "step": 5206
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.0484764122656173e-05,
+      "loss": 1.6877,
+      "step": 5207
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.04713854975461e-05,
+      "loss": 1.5614,
+      "step": 5208
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.045800687243602e-05,
+      "loss": 1.5944,
+      "step": 5209
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.044462824732594e-05,
+      "loss": 1.5898,
+      "step": 5210
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.043124962221587e-05,
+      "loss": 1.4612,
+      "step": 5211
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.0417870997105794e-05,
+      "loss": 1.4667,
+      "step": 5212
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.0404492371995714e-05,
+      "loss": 1.2805,
+      "step": 5213
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.0391113746885634e-05,
+      "loss": 1.4612,
+      "step": 5214
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.037773512177556e-05,
+      "loss": 1.3456,
+      "step": 5215
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.036435649666549e-05,
+      "loss": 1.2658,
+      "step": 5216
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.035097787155541e-05,
+      "loss": 1.3781,
+      "step": 5217
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.0337599246445334e-05,
+      "loss": 1.1882,
+      "step": 5218
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.0324220621335254e-05,
+      "loss": 1.1967,
+      "step": 5219
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.031084199622518e-05,
+      "loss": 1.4803,
+      "step": 5220
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.02974633711151e-05,
+      "loss": 1.1105,
+      "step": 5221
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.028408474600503e-05,
+      "loss": 1.3094,
+      "step": 5222
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.027070612089495e-05,
+      "loss": 1.0885,
+      "step": 5223
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.025732749578487e-05,
+      "loss": 1.2551,
+      "step": 5224
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.0243948870674794e-05,
+      "loss": 1.1937,
+      "step": 5225
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.023057024556472e-05,
+      "loss": 1.1965,
+      "step": 5226
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.021719162045464e-05,
+      "loss": 1.1668,
+      "step": 5227
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.020381299534456e-05,
+      "loss": 1.1826,
+      "step": 5228
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.019043437023449e-05,
+      "loss": 1.175,
+      "step": 5229
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.0177055745124415e-05,
+      "loss": 1.0306,
+      "step": 5230
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.0163677120014335e-05,
+      "loss": 1.2402,
+      "step": 5231
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.0150298494904255e-05,
+      "loss": 1.0945,
+      "step": 5232
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 6.013691986979418e-05,
+      "loss": 0.8997,
+      "step": 5233
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 6.012354124468411e-05,
+      "loss": 1.043,
+      "step": 5234
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 6.0110162619574035e-05,
+      "loss": 1.0594,
+      "step": 5235
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 6.009678399446395e-05,
+      "loss": 0.8631,
+      "step": 5236
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 6.0083405369353875e-05,
+      "loss": 0.8269,
+      "step": 5237
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 6.00700267442438e-05,
+      "loss": 0.7579,
+      "step": 5238
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 6.005664811913373e-05,
+      "loss": 0.8413,
+      "step": 5239
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 6.004326949402364e-05,
+      "loss": 0.6793,
+      "step": 5240
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 6.002989086891357e-05,
+      "loss": 1.8268,
+      "step": 5241
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 6.0016512243803496e-05,
+      "loss": 1.6956,
+      "step": 5242
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 6.000313361869342e-05,
+      "loss": 1.7834,
+      "step": 5243
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.9989754993583336e-05,
+      "loss": 1.7211,
+      "step": 5244
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.997637636847326e-05,
+      "loss": 1.8183,
+      "step": 5245
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.996299774336319e-05,
+      "loss": 1.6333,
+      "step": 5246
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.9949619118253116e-05,
+      "loss": 1.6846,
+      "step": 5247
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.993624049314303e-05,
+      "loss": 1.5196,
+      "step": 5248
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.9922861868032956e-05,
+      "loss": 1.4242,
+      "step": 5249
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.990948324292288e-05,
+      "loss": 1.4,
+      "step": 5250
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.98961046178128e-05,
+      "loss": 1.5289,
+      "step": 5251
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.988272599270273e-05,
+      "loss": 1.701,
+      "step": 5252
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.986934736759265e-05,
+      "loss": 1.5343,
+      "step": 5253
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.9855968742482576e-05,
+      "loss": 1.5174,
+      "step": 5254
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.9842590117372496e-05,
+      "loss": 1.2899,
+      "step": 5255
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.982921149226242e-05,
+      "loss": 1.5159,
+      "step": 5256
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.981583286715234e-05,
+      "loss": 1.5469,
+      "step": 5257
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.980245424204227e-05,
+      "loss": 1.4448,
+      "step": 5258
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.978907561693219e-05,
+      "loss": 1.3606,
+      "step": 5259
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.9775696991822117e-05,
+      "loss": 1.4439,
+      "step": 5260
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.9762318366712037e-05,
+      "loss": 1.5248,
+      "step": 5261
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.974893974160196e-05,
+      "loss": 1.3755,
+      "step": 5262
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.973556111649188e-05,
+      "loss": 1.4265,
+      "step": 5263
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.972218249138181e-05,
+      "loss": 1.324,
+      "step": 5264
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.970880386627173e-05,
+      "loss": 1.3924,
+      "step": 5265
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.969542524116166e-05,
+      "loss": 1.1917,
+      "step": 5266
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.968204661605158e-05,
+      "loss": 1.4914,
+      "step": 5267
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.9668667990941504e-05,
+      "loss": 1.3807,
+      "step": 5268
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.965528936583143e-05,
+      "loss": 1.2473,
+      "step": 5269
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.964191074072135e-05,
+      "loss": 1.3465,
+      "step": 5270
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.962853211561127e-05,
+      "loss": 1.424,
+      "step": 5271
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.96151534905012e-05,
+      "loss": 1.3406,
+      "step": 5272
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.9601774865391124e-05,
+      "loss": 1.2711,
+      "step": 5273
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.9588396240281044e-05,
+      "loss": 1.2738,
+      "step": 5274
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.9575017615170964e-05,
+      "loss": 1.2866,
+      "step": 5275
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.956163899006089e-05,
+      "loss": 1.218,
+      "step": 5276
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.954826036495082e-05,
+      "loss": 1.1956,
+      "step": 5277
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.953488173984073e-05,
+      "loss": 1.1334,
+      "step": 5278
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.952150311473066e-05,
+      "loss": 0.9466,
+      "step": 5279
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.9508124489620584e-05,
+      "loss": 1.0502,
+      "step": 5280
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.949474586451051e-05,
+      "loss": 1.0705,
+      "step": 5281
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.948136723940043e-05,
+      "loss": 1.0237,
+      "step": 5282
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.946798861429035e-05,
+      "loss": 0.9351,
+      "step": 5283
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.945460998918028e-05,
+      "loss": 0.903,
+      "step": 5284
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.9441231364070205e-05,
+      "loss": 0.9829,
+      "step": 5285
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.9427852738960125e-05,
+      "loss": 0.7484,
+      "step": 5286
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.9414474113850045e-05,
+      "loss": 0.7591,
+      "step": 5287
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.940109548873997e-05,
+      "loss": 0.9018,
+      "step": 5288
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.93877168636299e-05,
+      "loss": 0.9222,
+      "step": 5289
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.937433823851982e-05,
+      "loss": 0.6649,
+      "step": 5290
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.936095961340974e-05,
+      "loss": 2.0678,
+      "step": 5291
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.9347580988299665e-05,
+      "loss": 1.8606,
+      "step": 5292
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.933420236318959e-05,
+      "loss": 1.6844,
+      "step": 5293
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.932082373807951e-05,
+      "loss": 1.5502,
+      "step": 5294
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.930744511296943e-05,
+      "loss": 1.7198,
+      "step": 5295
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.929406648785936e-05,
+      "loss": 1.6466,
+      "step": 5296
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.9280687862749285e-05,
+      "loss": 1.4614,
+      "step": 5297
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.9267309237639205e-05,
+      "loss": 1.7481,
+      "step": 5298
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.925393061252913e-05,
+      "loss": 1.5959,
+      "step": 5299
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.924055198741905e-05,
+      "loss": 1.708,
+      "step": 5300
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.922717336230898e-05,
+      "loss": 1.5311,
+      "step": 5301
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.92137947371989e-05,
+      "loss": 1.5283,
+      "step": 5302
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.9200416112088826e-05,
+      "loss": 1.3733,
+      "step": 5303
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.9187037486978746e-05,
+      "loss": 1.5675,
+      "step": 5304
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.9173658861868666e-05,
+      "loss": 1.3596,
+      "step": 5305
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.916028023675859e-05,
+      "loss": 1.4333,
+      "step": 5306
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.914690161164852e-05,
+      "loss": 1.6578,
+      "step": 5307
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.913352298653844e-05,
+      "loss": 1.2442,
+      "step": 5308
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.912014436142836e-05,
+      "loss": 1.5287,
+      "step": 5309
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.9106765736318286e-05,
+      "loss": 1.425,
+      "step": 5310
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.909338711120821e-05,
+      "loss": 1.4425,
+      "step": 5311
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.908000848609813e-05,
+      "loss": 1.5012,
+      "step": 5312
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.906662986098805e-05,
+      "loss": 1.3773,
+      "step": 5313
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.905325123587798e-05,
+      "loss": 1.377,
+      "step": 5314
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.9039872610767906e-05,
+      "loss": 1.505,
+      "step": 5315
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.902649398565783e-05,
+      "loss": 1.4177,
+      "step": 5316
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.9013115360547746e-05,
+      "loss": 1.2485,
+      "step": 5317
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.899973673543767e-05,
+      "loss": 1.3458,
+      "step": 5318
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.89863581103276e-05,
+      "loss": 1.244,
+      "step": 5319
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.897297948521753e-05,
+      "loss": 1.1923,
+      "step": 5320
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.895960086010744e-05,
+      "loss": 1.2238,
+      "step": 5321
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.894622223499737e-05,
+      "loss": 1.0914,
+      "step": 5322
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.8932843609887294e-05,
+      "loss": 1.1061,
+      "step": 5323
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.891946498477722e-05,
+      "loss": 1.2438,
+      "step": 5324
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.8906086359667134e-05,
+      "loss": 1.1572,
+      "step": 5325
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.889270773455706e-05,
+      "loss": 1.1149,
+      "step": 5326
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.887932910944699e-05,
+      "loss": 1.1799,
+      "step": 5327
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.8865950484336914e-05,
+      "loss": 1.2351,
+      "step": 5328
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.885257185922683e-05,
+      "loss": 0.9103,
+      "step": 5329
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.8839193234116754e-05,
+      "loss": 1.2744,
+      "step": 5330
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.882581460900668e-05,
+      "loss": 0.9082,
+      "step": 5331
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.88124359838966e-05,
+      "loss": 1.2326,
+      "step": 5332
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.879905735878653e-05,
+      "loss": 0.9628,
+      "step": 5333
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.878567873367645e-05,
+      "loss": 1.0279,
+      "step": 5334
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.8772300108566374e-05,
+      "loss": 0.799,
+      "step": 5335
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.8758921483456294e-05,
+      "loss": 1.0002,
+      "step": 5336
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.874554285834622e-05,
+      "loss": 0.8228,
+      "step": 5337
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.873216423323614e-05,
+      "loss": 0.9171,
+      "step": 5338
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.871878560812607e-05,
+      "loss": 0.7057,
+      "step": 5339
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.870540698301599e-05,
+      "loss": 1.1429,
+      "step": 5340
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.8692028357905915e-05,
+      "loss": 2.0074,
+      "step": 5341
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.8678649732795835e-05,
+      "loss": 2.0585,
+      "step": 5342
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.866527110768576e-05,
+      "loss": 1.6961,
+      "step": 5343
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.865189248257568e-05,
+      "loss": 1.7482,
+      "step": 5344
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.863851385746561e-05,
+      "loss": 1.5811,
+      "step": 5345
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.862513523235553e-05,
+      "loss": 1.6977,
+      "step": 5346
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.8611756607245455e-05,
+      "loss": 1.662,
+      "step": 5347
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.8598377982135375e-05,
+      "loss": 1.462,
+      "step": 5348
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.85849993570253e-05,
+      "loss": 1.681,
+      "step": 5349
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.857162073191523e-05,
+      "loss": 1.4321,
+      "step": 5350
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.855824210680515e-05,
+      "loss": 1.4078,
+      "step": 5351
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.854486348169507e-05,
+      "loss": 1.5003,
+      "step": 5352
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.8531484856584995e-05,
+      "loss": 1.4194,
+      "step": 5353
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.851810623147492e-05,
+      "loss": 1.3992,
+      "step": 5354
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.850472760636484e-05,
+      "loss": 1.5062,
+      "step": 5355
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.849134898125476e-05,
+      "loss": 1.6143,
+      "step": 5356
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.847797035614469e-05,
+      "loss": 1.2613,
+      "step": 5357
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.8464591731034616e-05,
+      "loss": 1.4366,
+      "step": 5358
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.845121310592453e-05,
+      "loss": 1.3731,
+      "step": 5359
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.8437834480814456e-05,
+      "loss": 1.321,
+      "step": 5360
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.842445585570438e-05,
+      "loss": 1.4662,
+      "step": 5361
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.841107723059431e-05,
+      "loss": 1.5408,
+      "step": 5362
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.839769860548422e-05,
+      "loss": 1.3825,
+      "step": 5363
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.838431998037415e-05,
+      "loss": 1.442,
+      "step": 5364
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.8370941355264076e-05,
+      "loss": 1.5575,
+      "step": 5365
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.8357562730154e-05,
+      "loss": 1.4168,
+      "step": 5366
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.834418410504392e-05,
+      "loss": 1.2748,
+      "step": 5367
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.833080547993384e-05,
+      "loss": 1.5158,
+      "step": 5368
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.831742685482377e-05,
+      "loss": 1.3485,
+      "step": 5369
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.8304048229713696e-05,
+      "loss": 1.6129,
+      "step": 5370
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.8290669604603616e-05,
+      "loss": 1.3062,
+      "step": 5371
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.8277290979493536e-05,
+      "loss": 1.4637,
+      "step": 5372
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.826391235438346e-05,
+      "loss": 1.1894,
+      "step": 5373
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.825053372927339e-05,
+      "loss": 1.1767,
+      "step": 5374
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.823715510416331e-05,
+      "loss": 1.1877,
+      "step": 5375
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.822377647905323e-05,
+      "loss": 1.3144,
+      "step": 5376
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.8210397853943157e-05,
+      "loss": 1.2551,
+      "step": 5377
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.819701922883308e-05,
+      "loss": 0.9537,
+      "step": 5378
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.8183640603723e-05,
+      "loss": 1.0504,
+      "step": 5379
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.817026197861293e-05,
+      "loss": 0.9888,
+      "step": 5380
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.815688335350285e-05,
+      "loss": 0.9347,
+      "step": 5381
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.814350472839278e-05,
+      "loss": 1.1052,
+      "step": 5382
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.81301261032827e-05,
+      "loss": 0.9103,
+      "step": 5383
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.8116747478172624e-05,
+      "loss": 0.9661,
+      "step": 5384
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.8103368853062544e-05,
+      "loss": 0.9357,
+      "step": 5385
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.8089990227952464e-05,
+      "loss": 0.8912,
+      "step": 5386
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.807661160284239e-05,
+      "loss": 0.9478,
+      "step": 5387
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.806323297773232e-05,
+      "loss": 0.8079,
+      "step": 5388
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.804985435262224e-05,
+      "loss": 0.8618,
+      "step": 5389
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.803647572751216e-05,
+      "loss": 0.7471,
+      "step": 5390
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.8023097102402084e-05,
+      "loss": 2.0274,
+      "step": 5391
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.800971847729201e-05,
+      "loss": 2.0139,
+      "step": 5392
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.799633985218193e-05,
+      "loss": 1.9106,
+      "step": 5393
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.798296122707185e-05,
+      "loss": 1.7066,
+      "step": 5394
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.796958260196178e-05,
+      "loss": 1.6034,
+      "step": 5395
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.7956203976851704e-05,
+      "loss": 1.6188,
+      "step": 5396
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.794282535174163e-05,
+      "loss": 1.7452,
+      "step": 5397
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.7929446726631544e-05,
+      "loss": 1.7492,
+      "step": 5398
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.791606810152147e-05,
+      "loss": 1.7774,
+      "step": 5399
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.79026894764114e-05,
+      "loss": 1.5714,
+      "step": 5400
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.7889310851301325e-05,
+      "loss": 1.4788,
+      "step": 5401
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.787593222619124e-05,
+      "loss": 1.3891,
+      "step": 5402
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.7862553601081165e-05,
+      "loss": 1.4631,
+      "step": 5403
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.784917497597109e-05,
+      "loss": 1.5663,
+      "step": 5404
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.783579635086102e-05,
+      "loss": 1.5983,
+      "step": 5405
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.782241772575093e-05,
+      "loss": 1.3839,
+      "step": 5406
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.780903910064086e-05,
+      "loss": 1.4438,
+      "step": 5407
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.7795660475530785e-05,
+      "loss": 1.5907,
+      "step": 5408
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.778228185042071e-05,
+      "loss": 1.3118,
+      "step": 5409
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.7768903225310625e-05,
+      "loss": 1.4611,
+      "step": 5410
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.775552460020055e-05,
+      "loss": 1.3172,
+      "step": 5411
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.774214597509048e-05,
+      "loss": 1.293,
+      "step": 5412
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.77287673499804e-05,
+      "loss": 1.3422,
+      "step": 5413
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.7715388724870325e-05,
+      "loss": 1.5424,
+      "step": 5414
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.7702010099760245e-05,
+      "loss": 1.2987,
+      "step": 5415
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.768863147465017e-05,
+      "loss": 1.4346,
+      "step": 5416
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.767525284954009e-05,
+      "loss": 1.183,
+      "step": 5417
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.766187422443002e-05,
+      "loss": 1.1135,
+      "step": 5418
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.764849559931994e-05,
+      "loss": 1.346,
+      "step": 5419
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.7635116974209866e-05,
+      "loss": 1.1144,
+      "step": 5420
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.7621738349099786e-05,
+      "loss": 1.339,
+      "step": 5421
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.760835972398971e-05,
+      "loss": 1.179,
+      "step": 5422
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.759498109887963e-05,
+      "loss": 1.3087,
+      "step": 5423
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.758160247376956e-05,
+      "loss": 1.0818,
+      "step": 5424
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.756822384865948e-05,
+      "loss": 1.0732,
+      "step": 5425
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.7554845223549406e-05,
+      "loss": 1.276,
+      "step": 5426
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.7541466598439326e-05,
+      "loss": 1.0568,
+      "step": 5427
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.752808797332925e-05,
+      "loss": 1.1899,
+      "step": 5428
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.751470934821917e-05,
+      "loss": 1.0984,
+      "step": 5429
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.75013307231091e-05,
+      "loss": 1.1863,
+      "step": 5430
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.7487952097999026e-05,
+      "loss": 1.1239,
+      "step": 5431
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.7474573472888946e-05,
+      "loss": 1.1371,
+      "step": 5432
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.7461194847778866e-05,
+      "loss": 1.0515,
+      "step": 5433
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.744781622266879e-05,
+      "loss": 1.0718,
+      "step": 5434
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.743443759755872e-05,
+      "loss": 1.073,
+      "step": 5435
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.742105897244863e-05,
+      "loss": 0.8966,
+      "step": 5436
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.740768034733856e-05,
+      "loss": 0.7239,
+      "step": 5437
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.739430172222849e-05,
+      "loss": 0.799,
+      "step": 5438
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.7380923097118414e-05,
+      "loss": 0.7844,
+      "step": 5439
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.736754447200833e-05,
+      "loss": 0.7824,
+      "step": 5440
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.7354165846898254e-05,
+      "loss": 2.1556,
+      "step": 5441
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.734078722178818e-05,
+      "loss": 1.7545,
+      "step": 5442
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.732740859667811e-05,
+      "loss": 1.6455,
+      "step": 5443
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.731402997156802e-05,
+      "loss": 1.5554,
+      "step": 5444
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.730065134645795e-05,
+      "loss": 1.7329,
+      "step": 5445
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.7287272721347874e-05,
+      "loss": 1.6795,
+      "step": 5446
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.72738940962378e-05,
+      "loss": 1.5634,
+      "step": 5447
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.726051547112772e-05,
+      "loss": 1.6822,
+      "step": 5448
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.724713684601764e-05,
+      "loss": 1.7735,
+      "step": 5449
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.723375822090757e-05,
+      "loss": 1.4982,
+      "step": 5450
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.7220379595797494e-05,
+      "loss": 1.6417,
+      "step": 5451
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.7207000970687414e-05,
+      "loss": 1.4576,
+      "step": 5452
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.7193622345577334e-05,
+      "loss": 1.5326,
+      "step": 5453
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.718024372046726e-05,
+      "loss": 1.5411,
+      "step": 5454
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.716686509535719e-05,
+      "loss": 1.4755,
+      "step": 5455
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.715348647024711e-05,
+      "loss": 1.6754,
+      "step": 5456
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.714010784513703e-05,
+      "loss": 1.4479,
+      "step": 5457
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.7126729220026955e-05,
+      "loss": 1.553,
+      "step": 5458
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.711335059491688e-05,
+      "loss": 1.5092,
+      "step": 5459
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.70999719698068e-05,
+      "loss": 1.3748,
+      "step": 5460
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.708659334469672e-05,
+      "loss": 1.3412,
+      "step": 5461
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.707321471958665e-05,
+      "loss": 1.3274,
+      "step": 5462
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.705983609447657e-05,
+      "loss": 1.4116,
+      "step": 5463
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.7046457469366495e-05,
+      "loss": 1.4051,
+      "step": 5464
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.703307884425642e-05,
+      "loss": 1.2879,
+      "step": 5465
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.701970021914634e-05,
+      "loss": 1.3443,
+      "step": 5466
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.700632159403626e-05,
+      "loss": 1.3445,
+      "step": 5467
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.699294296892619e-05,
+      "loss": 1.2566,
+      "step": 5468
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.6979564343816115e-05,
+      "loss": 1.2266,
+      "step": 5469
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.6966185718706035e-05,
+      "loss": 1.2803,
+      "step": 5470
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.6952807093595955e-05,
+      "loss": 1.2406,
+      "step": 5471
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.693942846848588e-05,
+      "loss": 1.2215,
+      "step": 5472
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.692604984337581e-05,
+      "loss": 1.2526,
+      "step": 5473
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.691267121826573e-05,
+      "loss": 1.0866,
+      "step": 5474
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.689929259315565e-05,
+      "loss": 1.2689,
+      "step": 5475
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.6885913968045576e-05,
+      "loss": 1.2706,
+      "step": 5476
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.68725353429355e-05,
+      "loss": 1.2274,
+      "step": 5477
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.685915671782543e-05,
+      "loss": 1.0218,
+      "step": 5478
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.684577809271534e-05,
+      "loss": 1.0318,
+      "step": 5479
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.683239946760527e-05,
+      "loss": 1.0488,
+      "step": 5480
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.6819020842495196e-05,
+      "loss": 1.1364,
+      "step": 5481
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.680564221738512e-05,
+      "loss": 1.0562,
+      "step": 5482
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.6792263592275036e-05,
+      "loss": 1.1411,
+      "step": 5483
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.677888496716496e-05,
+      "loss": 1.0584,
+      "step": 5484
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.676550634205489e-05,
+      "loss": 0.7719,
+      "step": 5485
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.6752127716944816e-05,
+      "loss": 0.9468,
+      "step": 5486
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.673874909183473e-05,
+      "loss": 0.9273,
+      "step": 5487
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.6725370466724656e-05,
+      "loss": 1.1064,
+      "step": 5488
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.671199184161458e-05,
+      "loss": 0.793,
+      "step": 5489
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.66986132165045e-05,
+      "loss": 0.6555,
+      "step": 5490
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.668523459139442e-05,
+      "loss": 1.9761,
+      "step": 5491
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.667185596628435e-05,
+      "loss": 1.877,
+      "step": 5492
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.6658477341174277e-05,
+      "loss": 1.6734,
+      "step": 5493
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.6645098716064197e-05,
+      "loss": 1.604,
+      "step": 5494
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.663172009095412e-05,
+      "loss": 1.5899,
+      "step": 5495
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.6618341465844043e-05,
+      "loss": 1.7287,
+      "step": 5496
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.660496284073397e-05,
+      "loss": 1.4926,
+      "step": 5497
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.659158421562389e-05,
+      "loss": 1.731,
+      "step": 5498
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.657820559051382e-05,
+      "loss": 1.661,
+      "step": 5499
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.656482696540374e-05,
+      "loss": 1.6288,
+      "step": 5500
+    },
+    {
+      "epoch": 1.7,
+      "eval_bleu": 2.1764352694111017e-14,
+      "eval_loss": 1.7821362018585205,
+      "eval_runtime": 2548.5487,
+      "eval_samples_per_second": 5.792,
+      "eval_steps_per_second": 0.724,
+      "step": 5500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.6551448340293664e-05,
+      "loss": 1.5075,
+      "step": 5501
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.6538069715183584e-05,
+      "loss": 1.6619,
+      "step": 5502
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.652469109007351e-05,
+      "loss": 1.6139,
+      "step": 5503
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.651131246496343e-05,
+      "loss": 1.4467,
+      "step": 5504
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.649793383985336e-05,
+      "loss": 1.5041,
+      "step": 5505
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.648455521474328e-05,
+      "loss": 1.4722,
+      "step": 5506
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.6471176589633204e-05,
+      "loss": 1.4569,
+      "step": 5507
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.6457797964523124e-05,
+      "loss": 1.4035,
+      "step": 5508
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.644441933941305e-05,
+      "loss": 1.5094,
+      "step": 5509
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.643104071430297e-05,
+      "loss": 1.6371,
+      "step": 5510
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.64176620891929e-05,
+      "loss": 1.3571,
+      "step": 5511
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.6404283464082824e-05,
+      "loss": 1.4726,
+      "step": 5512
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.6390904838972744e-05,
+      "loss": 1.4375,
+      "step": 5513
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.6377526213862664e-05,
+      "loss": 1.3648,
+      "step": 5514
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.636414758875259e-05,
+      "loss": 1.3878,
+      "step": 5515
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.635076896364252e-05,
+      "loss": 1.344,
+      "step": 5516
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.633739033853243e-05,
+      "loss": 1.3615,
+      "step": 5517
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.632401171342236e-05,
+      "loss": 1.5252,
+      "step": 5518
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.6310633088312285e-05,
+      "loss": 1.1552,
+      "step": 5519
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.629725446320221e-05,
+      "loss": 1.3946,
+      "step": 5520
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.6283875838092125e-05,
+      "loss": 1.1686,
+      "step": 5521
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.627049721298205e-05,
+      "loss": 1.2706,
+      "step": 5522
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.625711858787198e-05,
+      "loss": 1.2915,
+      "step": 5523
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.6243739962761905e-05,
+      "loss": 1.0241,
+      "step": 5524
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.623036133765182e-05,
+      "loss": 1.352,
+      "step": 5525
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.6216982712541745e-05,
+      "loss": 1.2772,
+      "step": 5526
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.620360408743167e-05,
+      "loss": 1.1888,
+      "step": 5527
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.61902254623216e-05,
+      "loss": 1.1269,
+      "step": 5528
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.617684683721152e-05,
+      "loss": 1.0112,
+      "step": 5529
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.616346821210144e-05,
+      "loss": 1.0491,
+      "step": 5530
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.6150089586991365e-05,
+      "loss": 1.1024,
+      "step": 5531
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.613671096188129e-05,
+      "loss": 1.0875,
+      "step": 5532
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.612333233677121e-05,
+      "loss": 0.8966,
+      "step": 5533
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.610995371166113e-05,
+      "loss": 1.0129,
+      "step": 5534
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.609657508655106e-05,
+      "loss": 0.8351,
+      "step": 5535
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.6083196461440986e-05,
+      "loss": 0.697,
+      "step": 5536
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.6069817836330906e-05,
+      "loss": 0.9642,
+      "step": 5537
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.6056439211220826e-05,
+      "loss": 0.8432,
+      "step": 5538
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.604306058611075e-05,
+      "loss": 0.6962,
+      "step": 5539
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.602968196100068e-05,
+      "loss": 0.6628,
+      "step": 5540
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.60163033358906e-05,
+      "loss": 2.1256,
+      "step": 5541
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.600292471078052e-05,
+      "loss": 1.7606,
+      "step": 5542
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.5989546085670446e-05,
+      "loss": 1.6693,
+      "step": 5543
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.5976167460560366e-05,
+      "loss": 1.4854,
+      "step": 5544
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.596278883545029e-05,
+      "loss": 1.7871,
+      "step": 5545
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.594941021034022e-05,
+      "loss": 1.6002,
+      "step": 5546
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.593603158523014e-05,
+      "loss": 1.6155,
+      "step": 5547
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.592265296012006e-05,
+      "loss": 1.5315,
+      "step": 5548
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.5909274335009986e-05,
+      "loss": 1.5368,
+      "step": 5549
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.589589570989991e-05,
+      "loss": 1.5598,
+      "step": 5550
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.588251708478983e-05,
+      "loss": 1.4445,
+      "step": 5551
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.586913845967975e-05,
+      "loss": 1.5475,
+      "step": 5552
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.585575983456968e-05,
+      "loss": 1.5669,
+      "step": 5553
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.584238120945961e-05,
+      "loss": 1.495,
+      "step": 5554
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.582900258434953e-05,
+      "loss": 1.4731,
+      "step": 5555
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.581562395923945e-05,
+      "loss": 1.464,
+      "step": 5556
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.5802245334129374e-05,
+      "loss": 1.2415,
+      "step": 5557
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.57888667090193e-05,
+      "loss": 1.5802,
+      "step": 5558
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.577548808390922e-05,
+      "loss": 1.5865,
+      "step": 5559
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.576210945879914e-05,
+      "loss": 1.3483,
+      "step": 5560
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.574873083368907e-05,
+      "loss": 1.2445,
+      "step": 5561
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.5735352208578994e-05,
+      "loss": 1.2508,
+      "step": 5562
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.572197358346892e-05,
+      "loss": 1.4123,
+      "step": 5563
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.5708594958358834e-05,
+      "loss": 1.3796,
+      "step": 5564
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.569521633324876e-05,
+      "loss": 1.522,
+      "step": 5565
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.568183770813869e-05,
+      "loss": 1.2797,
+      "step": 5566
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.5668459083028614e-05,
+      "loss": 1.1727,
+      "step": 5567
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.565508045791853e-05,
+      "loss": 1.5577,
+      "step": 5568
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.5641701832808454e-05,
+      "loss": 1.3192,
+      "step": 5569
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.562832320769838e-05,
+      "loss": 1.2416,
+      "step": 5570
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.56149445825883e-05,
+      "loss": 1.133,
+      "step": 5571
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.560156595747822e-05,
+      "loss": 1.4075,
+      "step": 5572
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.558818733236815e-05,
+      "loss": 1.3142,
+      "step": 5573
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.5574808707258075e-05,
+      "loss": 1.2896,
+      "step": 5574
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.5561430082147995e-05,
+      "loss": 1.1244,
+      "step": 5575
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.554805145703792e-05,
+      "loss": 1.2093,
+      "step": 5576
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.553467283192784e-05,
+      "loss": 1.0644,
+      "step": 5577
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.552129420681777e-05,
+      "loss": 1.2555,
+      "step": 5578
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.550791558170769e-05,
+      "loss": 1.1647,
+      "step": 5579
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.5494536956597615e-05,
+      "loss": 1.1835,
+      "step": 5580
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.5481158331487535e-05,
+      "loss": 1.0399,
+      "step": 5581
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.546777970637746e-05,
+      "loss": 1.0481,
+      "step": 5582
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.545440108126738e-05,
+      "loss": 0.8603,
+      "step": 5583
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.544102245615731e-05,
+      "loss": 0.9616,
+      "step": 5584
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.542764383104723e-05,
+      "loss": 0.909,
+      "step": 5585
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.5414265205937155e-05,
+      "loss": 0.8525,
+      "step": 5586
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.5400886580827075e-05,
+      "loss": 0.8205,
+      "step": 5587
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.5387507955717e-05,
+      "loss": 0.8107,
+      "step": 5588
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.537412933060692e-05,
+      "loss": 0.664,
+      "step": 5589
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.536075070549685e-05,
+      "loss": 1.0147,
+      "step": 5590
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.534737208038677e-05,
+      "loss": 1.8556,
+      "step": 5591
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.5333993455276696e-05,
+      "loss": 1.7389,
+      "step": 5592
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.532061483016662e-05,
+      "loss": 1.705,
+      "step": 5593
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.530723620505654e-05,
+      "loss": 1.7593,
+      "step": 5594
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.529385757994646e-05,
+      "loss": 1.4713,
+      "step": 5595
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.528047895483639e-05,
+      "loss": 1.3351,
+      "step": 5596
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.5267100329726316e-05,
+      "loss": 1.7082,
+      "step": 5597
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.525372170461623e-05,
+      "loss": 1.3338,
+      "step": 5598
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.5240343079506156e-05,
+      "loss": 1.4993,
+      "step": 5599
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.522696445439608e-05,
+      "loss": 1.466,
+      "step": 5600
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.521358582928601e-05,
+      "loss": 1.487,
+      "step": 5601
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.520020720417592e-05,
+      "loss": 1.5083,
+      "step": 5602
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.518682857906585e-05,
+      "loss": 1.3215,
+      "step": 5603
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.5173449953955776e-05,
+      "loss": 1.6621,
+      "step": 5604
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.51600713288457e-05,
+      "loss": 1.428,
+      "step": 5605
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.5146692703735616e-05,
+      "loss": 1.3968,
+      "step": 5606
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.513331407862554e-05,
+      "loss": 1.5669,
+      "step": 5607
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.511993545351547e-05,
+      "loss": 1.4754,
+      "step": 5608
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.51065568284054e-05,
+      "loss": 1.2938,
+      "step": 5609
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.509317820329532e-05,
+      "loss": 1.4822,
+      "step": 5610
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.507979957818524e-05,
+      "loss": 1.3004,
+      "step": 5611
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.5066420953075163e-05,
+      "loss": 1.4493,
+      "step": 5612
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.505304232796509e-05,
+      "loss": 1.3965,
+      "step": 5613
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.503966370285501e-05,
+      "loss": 1.1953,
+      "step": 5614
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.502628507774493e-05,
+      "loss": 1.313,
+      "step": 5615
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.501290645263486e-05,
+      "loss": 1.4429,
+      "step": 5616
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.4999527827524784e-05,
+      "loss": 1.4578,
+      "step": 5617
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.4986149202414704e-05,
+      "loss": 1.3542,
+      "step": 5618
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.4972770577304624e-05,
+      "loss": 1.3295,
+      "step": 5619
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.495939195219455e-05,
+      "loss": 1.2126,
+      "step": 5620
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.494601332708448e-05,
+      "loss": 1.1131,
+      "step": 5621
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.49326347019744e-05,
+      "loss": 1.2554,
+      "step": 5622
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.491925607686432e-05,
+      "loss": 1.2369,
+      "step": 5623
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.4905877451754244e-05,
+      "loss": 1.0679,
+      "step": 5624
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.4892498826644164e-05,
+      "loss": 1.2051,
+      "step": 5625
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.487912020153409e-05,
+      "loss": 1.171,
+      "step": 5626
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.486574157642402e-05,
+      "loss": 1.1002,
+      "step": 5627
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.485236295131394e-05,
+      "loss": 0.992,
+      "step": 5628
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.483898432620386e-05,
+      "loss": 1.2291,
+      "step": 5629
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.4825605701093784e-05,
+      "loss": 1.0327,
+      "step": 5630
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.481222707598371e-05,
+      "loss": 1.015,
+      "step": 5631
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.479884845087363e-05,
+      "loss": 0.9163,
+      "step": 5632
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.478546982576355e-05,
+      "loss": 0.9894,
+      "step": 5633
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.477209120065348e-05,
+      "loss": 0.8898,
+      "step": 5634
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.4758712575543405e-05,
+      "loss": 0.9611,
+      "step": 5635
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.4745333950433325e-05,
+      "loss": 0.8524,
+      "step": 5636
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.4731955325323245e-05,
+      "loss": 0.7621,
+      "step": 5637
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.471857670021317e-05,
+      "loss": 0.8055,
+      "step": 5638
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.47051980751031e-05,
+      "loss": 0.7998,
+      "step": 5639
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.469181944999302e-05,
+      "loss": 1.2637,
+      "step": 5640
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.467844082488294e-05,
+      "loss": 1.9323,
+      "step": 5641
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.4665062199772865e-05,
+      "loss": 1.9295,
+      "step": 5642
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.465168357466279e-05,
+      "loss": 1.7132,
+      "step": 5643
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.463830494955272e-05,
+      "loss": 1.6969,
+      "step": 5644
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.462492632444263e-05,
+      "loss": 1.661,
+      "step": 5645
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.461154769933256e-05,
+      "loss": 1.6118,
+      "step": 5646
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.4598169074222485e-05,
+      "loss": 1.5933,
+      "step": 5647
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.458479044911241e-05,
+      "loss": 1.5609,
+      "step": 5648
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.4571411824002325e-05,
+      "loss": 1.4674,
+      "step": 5649
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.455803319889225e-05,
+      "loss": 1.354,
+      "step": 5650
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.454465457378218e-05,
+      "loss": 1.5372,
+      "step": 5651
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.45312759486721e-05,
+      "loss": 1.6472,
+      "step": 5652
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.451789732356202e-05,
+      "loss": 1.2461,
+      "step": 5653
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.4504518698451946e-05,
+      "loss": 1.1307,
+      "step": 5654
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.449114007334187e-05,
+      "loss": 1.4484,
+      "step": 5655
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.447776144823179e-05,
+      "loss": 1.478,
+      "step": 5656
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.446438282312171e-05,
+      "loss": 1.5962,
+      "step": 5657
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.445100419801164e-05,
+      "loss": 1.4034,
+      "step": 5658
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.4437625572901566e-05,
+      "loss": 1.2675,
+      "step": 5659
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.4424246947791486e-05,
+      "loss": 1.3169,
+      "step": 5660
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.441086832268141e-05,
+      "loss": 1.3467,
+      "step": 5661
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.439748969757133e-05,
+      "loss": 1.3481,
+      "step": 5662
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.438411107246126e-05,
+      "loss": 1.5742,
+      "step": 5663
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.437073244735118e-05,
+      "loss": 1.196,
+      "step": 5664
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.4357353822241106e-05,
+      "loss": 1.1922,
+      "step": 5665
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.4343975197131026e-05,
+      "loss": 1.4994,
+      "step": 5666
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.433059657202095e-05,
+      "loss": 1.4731,
+      "step": 5667
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.431721794691087e-05,
+      "loss": 1.2168,
+      "step": 5668
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.43038393218008e-05,
+      "loss": 1.4595,
+      "step": 5669
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.429046069669072e-05,
+      "loss": 1.211,
+      "step": 5670
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.427708207158065e-05,
+      "loss": 1.22,
+      "step": 5671
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.426370344647057e-05,
+      "loss": 1.1101,
+      "step": 5672
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.4250324821360494e-05,
+      "loss": 1.109,
+      "step": 5673
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.423694619625042e-05,
+      "loss": 0.9561,
+      "step": 5674
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.4223567571140334e-05,
+      "loss": 1.2449,
+      "step": 5675
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.421018894603026e-05,
+      "loss": 1.2775,
+      "step": 5676
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.419681032092019e-05,
+      "loss": 1.2124,
+      "step": 5677
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.4183431695810114e-05,
+      "loss": 1.0675,
+      "step": 5678
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.417005307070003e-05,
+      "loss": 1.1056,
+      "step": 5679
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.4156674445589954e-05,
+      "loss": 1.1584,
+      "step": 5680
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.414329582047988e-05,
+      "loss": 1.043,
+      "step": 5681
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.412991719536981e-05,
+      "loss": 0.8995,
+      "step": 5682
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.411653857025972e-05,
+      "loss": 0.9237,
+      "step": 5683
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.410315994514965e-05,
+      "loss": 1.1542,
+      "step": 5684
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.4089781320039574e-05,
+      "loss": 0.974,
+      "step": 5685
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.40764026949295e-05,
+      "loss": 0.9291,
+      "step": 5686
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.4063024069819414e-05,
+      "loss": 0.7515,
+      "step": 5687
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.404964544470934e-05,
+      "loss": 0.6632,
+      "step": 5688
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.403626681959927e-05,
+      "loss": 0.7795,
+      "step": 5689
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.4022888194489195e-05,
+      "loss": 0.8816,
+      "step": 5690
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.4009509569379115e-05,
+      "loss": 1.9432,
+      "step": 5691
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.3996130944269035e-05,
+      "loss": 1.7379,
+      "step": 5692
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.398275231915896e-05,
+      "loss": 1.6065,
+      "step": 5693
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.396937369404889e-05,
+      "loss": 1.8658,
+      "step": 5694
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.395599506893881e-05,
+      "loss": 1.345,
+      "step": 5695
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.394261644382873e-05,
+      "loss": 1.6214,
+      "step": 5696
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.3929237818718655e-05,
+      "loss": 1.4498,
+      "step": 5697
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.391585919360858e-05,
+      "loss": 1.5001,
+      "step": 5698
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.39024805684985e-05,
+      "loss": 1.7471,
+      "step": 5699
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.388910194338842e-05,
+      "loss": 1.3923,
+      "step": 5700
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.387572331827835e-05,
+      "loss": 1.5527,
+      "step": 5701
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.386234469316827e-05,
+      "loss": 1.5076,
+      "step": 5702
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.3848966068058195e-05,
+      "loss": 1.4015,
+      "step": 5703
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.3835587442948115e-05,
+      "loss": 1.2806,
+      "step": 5704
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.382220881783804e-05,
+      "loss": 1.4623,
+      "step": 5705
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.380883019272796e-05,
+      "loss": 1.5435,
+      "step": 5706
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.379545156761789e-05,
+      "loss": 1.3933,
+      "step": 5707
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.3782072942507816e-05,
+      "loss": 1.3575,
+      "step": 5708
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.3768694317397736e-05,
+      "loss": 1.5487,
+      "step": 5709
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.3755315692287656e-05,
+      "loss": 1.325,
+      "step": 5710
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.374193706717758e-05,
+      "loss": 1.3686,
+      "step": 5711
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.372855844206751e-05,
+      "loss": 1.2979,
+      "step": 5712
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.371517981695743e-05,
+      "loss": 1.3087,
+      "step": 5713
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.370180119184735e-05,
+      "loss": 1.5401,
+      "step": 5714
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.3688422566737276e-05,
+      "loss": 1.3897,
+      "step": 5715
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.36750439416272e-05,
+      "loss": 1.4023,
+      "step": 5716
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.366166531651712e-05,
+      "loss": 1.2725,
+      "step": 5717
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.364828669140704e-05,
+      "loss": 1.3083,
+      "step": 5718
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.363490806629697e-05,
+      "loss": 1.0955,
+      "step": 5719
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.3621529441186896e-05,
+      "loss": 1.3388,
+      "step": 5720
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.3608150816076816e-05,
+      "loss": 1.281,
+      "step": 5721
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.3594772190966736e-05,
+      "loss": 1.2717,
+      "step": 5722
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.358139356585666e-05,
+      "loss": 1.3258,
+      "step": 5723
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.356801494074659e-05,
+      "loss": 1.0747,
+      "step": 5724
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.355463631563652e-05,
+      "loss": 1.3034,
+      "step": 5725
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.354125769052643e-05,
+      "loss": 1.1843,
+      "step": 5726
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.352787906541636e-05,
+      "loss": 1.0237,
+      "step": 5727
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.3514500440306283e-05,
+      "loss": 0.9553,
+      "step": 5728
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.3501121815196203e-05,
+      "loss": 1.0161,
+      "step": 5729
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.3487743190086123e-05,
+      "loss": 0.9173,
+      "step": 5730
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.347436456497605e-05,
+      "loss": 1.0957,
+      "step": 5731
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.346098593986598e-05,
+      "loss": 1.0003,
+      "step": 5732
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.34476073147559e-05,
+      "loss": 0.8774,
+      "step": 5733
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.343422868964582e-05,
+      "loss": 0.9547,
+      "step": 5734
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.3420850064535744e-05,
+      "loss": 0.9763,
+      "step": 5735
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.340747143942567e-05,
+      "loss": 1.0164,
+      "step": 5736
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.339409281431559e-05,
+      "loss": 0.9117,
+      "step": 5737
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.338071418920551e-05,
+      "loss": 0.7688,
+      "step": 5738
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.336733556409544e-05,
+      "loss": 0.7898,
+      "step": 5739
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.3353956938985364e-05,
+      "loss": 0.8172,
+      "step": 5740
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.3340578313875284e-05,
+      "loss": 2.1981,
+      "step": 5741
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.332719968876521e-05,
+      "loss": 1.9617,
+      "step": 5742
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.331382106365513e-05,
+      "loss": 1.7009,
+      "step": 5743
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.330044243854506e-05,
+      "loss": 1.6344,
+      "step": 5744
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.328706381343498e-05,
+      "loss": 1.5732,
+      "step": 5745
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.3273685188324904e-05,
+      "loss": 1.5374,
+      "step": 5746
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.3260306563214824e-05,
+      "loss": 1.5675,
+      "step": 5747
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.324692793810475e-05,
+      "loss": 1.4747,
+      "step": 5748
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.323354931299467e-05,
+      "loss": 1.6895,
+      "step": 5749
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.32201706878846e-05,
+      "loss": 1.4597,
+      "step": 5750
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.320679206277452e-05,
+      "loss": 1.395,
+      "step": 5751
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.3193413437664445e-05,
+      "loss": 1.5035,
+      "step": 5752
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.3180034812554365e-05,
+      "loss": 1.3074,
+      "step": 5753
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.316665618744429e-05,
+      "loss": 1.3021,
+      "step": 5754
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.315327756233421e-05,
+      "loss": 1.6345,
+      "step": 5755
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.313989893722413e-05,
+      "loss": 1.3901,
+      "step": 5756
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.312652031211406e-05,
+      "loss": 1.473,
+      "step": 5757
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.3113141687003985e-05,
+      "loss": 1.5336,
+      "step": 5758
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.309976306189391e-05,
+      "loss": 1.1758,
+      "step": 5759
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.3086384436783825e-05,
+      "loss": 1.3955,
+      "step": 5760
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.307300581167375e-05,
+      "loss": 1.4177,
+      "step": 5761
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.305962718656368e-05,
+      "loss": 1.4314,
+      "step": 5762
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.3046248561453606e-05,
+      "loss": 1.4785,
+      "step": 5763
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.303286993634352e-05,
+      "loss": 1.3437,
+      "step": 5764
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.3019491311233446e-05,
+      "loss": 0.9918,
+      "step": 5765
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.300611268612337e-05,
+      "loss": 1.2555,
+      "step": 5766
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.29927340610133e-05,
+      "loss": 1.395,
+      "step": 5767
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.297935543590321e-05,
+      "loss": 1.1645,
+      "step": 5768
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.296597681079314e-05,
+      "loss": 1.2279,
+      "step": 5769
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.2952598185683066e-05,
+      "loss": 1.2543,
+      "step": 5770
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.293921956057299e-05,
+      "loss": 1.2996,
+      "step": 5771
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.292584093546291e-05,
+      "loss": 1.3608,
+      "step": 5772
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.291246231035283e-05,
+      "loss": 1.0086,
+      "step": 5773
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.289908368524276e-05,
+      "loss": 1.1377,
+      "step": 5774
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.2885705060132686e-05,
+      "loss": 1.1258,
+      "step": 5775
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.2872326435022606e-05,
+      "loss": 0.9545,
+      "step": 5776
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.2858947809912526e-05,
+      "loss": 1.1496,
+      "step": 5777
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.284556918480245e-05,
+      "loss": 1.0382,
+      "step": 5778
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.283219055969238e-05,
+      "loss": 1.0107,
+      "step": 5779
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.28188119345823e-05,
+      "loss": 1.0155,
+      "step": 5780
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.280543330947222e-05,
+      "loss": 1.0423,
+      "step": 5781
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.2792054684362147e-05,
+      "loss": 0.9451,
+      "step": 5782
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.2778676059252067e-05,
+      "loss": 0.9759,
+      "step": 5783
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.276529743414199e-05,
+      "loss": 0.8552,
+      "step": 5784
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.275191880903191e-05,
+      "loss": 0.8112,
+      "step": 5785
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.273854018392184e-05,
+      "loss": 0.9698,
+      "step": 5786
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.272516155881176e-05,
+      "loss": 0.868,
+      "step": 5787
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.271178293370169e-05,
+      "loss": 0.7083,
+      "step": 5788
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.2698404308591614e-05,
+      "loss": 0.7282,
+      "step": 5789
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.2685025683481534e-05,
+      "loss": 0.9484,
+      "step": 5790
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.2671647058371454e-05,
+      "loss": 2.0476,
+      "step": 5791
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.265826843326138e-05,
+      "loss": 1.7743,
+      "step": 5792
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.264488980815131e-05,
+      "loss": 1.7148,
+      "step": 5793
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.263151118304123e-05,
+      "loss": 1.5047,
+      "step": 5794
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.261813255793115e-05,
+      "loss": 1.7614,
+      "step": 5795
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.2604753932821074e-05,
+      "loss": 1.4847,
+      "step": 5796
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.2591375307711e-05,
+      "loss": 1.6313,
+      "step": 5797
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.257799668260092e-05,
+      "loss": 1.4758,
+      "step": 5798
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.256461805749084e-05,
+      "loss": 1.3163,
+      "step": 5799
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.255123943238077e-05,
+      "loss": 1.546,
+      "step": 5800
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.2537860807270694e-05,
+      "loss": 1.4108,
+      "step": 5801
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.2524482182160614e-05,
+      "loss": 1.603,
+      "step": 5802
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.2511103557050534e-05,
+      "loss": 1.3908,
+      "step": 5803
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.249772493194046e-05,
+      "loss": 1.3828,
+      "step": 5804
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.248434630683039e-05,
+      "loss": 1.4317,
+      "step": 5805
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.2470967681720315e-05,
+      "loss": 1.5954,
+      "step": 5806
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.245758905661023e-05,
+      "loss": 1.5643,
+      "step": 5807
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.2444210431500155e-05,
+      "loss": 1.306,
+      "step": 5808
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.243083180639008e-05,
+      "loss": 1.4941,
+      "step": 5809
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.241745318128e-05,
+      "loss": 1.3043,
+      "step": 5810
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.240407455616992e-05,
+      "loss": 1.4378,
+      "step": 5811
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.239069593105985e-05,
+      "loss": 1.3967,
+      "step": 5812
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.2377317305949775e-05,
+      "loss": 1.3848,
+      "step": 5813
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.2363938680839695e-05,
+      "loss": 1.4415,
+      "step": 5814
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.2350560055729615e-05,
+      "loss": 1.3107,
+      "step": 5815
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.233718143061954e-05,
+      "loss": 1.3497,
+      "step": 5816
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.232380280550947e-05,
+      "loss": 1.5253,
+      "step": 5817
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.231042418039939e-05,
+      "loss": 1.1233,
+      "step": 5818
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.229704555528931e-05,
+      "loss": 1.0856,
+      "step": 5819
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.2283666930179235e-05,
+      "loss": 1.0824,
+      "step": 5820
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.227028830506916e-05,
+      "loss": 1.27,
+      "step": 5821
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.225690967995908e-05,
+      "loss": 1.326,
+      "step": 5822
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.224353105484901e-05,
+      "loss": 1.1779,
+      "step": 5823
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.223015242973893e-05,
+      "loss": 1.1179,
+      "step": 5824
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.2216773804628856e-05,
+      "loss": 1.1624,
+      "step": 5825
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.2203395179518776e-05,
+      "loss": 1.1506,
+      "step": 5826
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.21900165544087e-05,
+      "loss": 1.0394,
+      "step": 5827
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.217663792929862e-05,
+      "loss": 1.1099,
+      "step": 5828
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.216325930418855e-05,
+      "loss": 1.1271,
+      "step": 5829
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.214988067907847e-05,
+      "loss": 1.1405,
+      "step": 5830
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.2136502053968396e-05,
+      "loss": 0.9436,
+      "step": 5831
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.2123123428858316e-05,
+      "loss": 1.0279,
+      "step": 5832
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.210974480374824e-05,
+      "loss": 0.8551,
+      "step": 5833
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.209636617863816e-05,
+      "loss": 1.0319,
+      "step": 5834
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.208298755352809e-05,
+      "loss": 0.7494,
+      "step": 5835
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.206960892841801e-05,
+      "loss": 0.8772,
+      "step": 5836
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.205623030330793e-05,
+      "loss": 0.5954,
+      "step": 5837
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.2042851678197856e-05,
+      "loss": 0.9454,
+      "step": 5838
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.202947305308778e-05,
+      "loss": 0.6793,
+      "step": 5839
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.201609442797771e-05,
+      "loss": 0.8764,
+      "step": 5840
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.200271580286762e-05,
+      "loss": 1.9406,
+      "step": 5841
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.198933717775755e-05,
+      "loss": 1.7936,
+      "step": 5842
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.197595855264748e-05,
+      "loss": 1.6322,
+      "step": 5843
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.1962579927537403e-05,
+      "loss": 1.6019,
+      "step": 5844
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.194920130242732e-05,
+      "loss": 1.7055,
+      "step": 5845
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.1935822677317244e-05,
+      "loss": 1.5929,
+      "step": 5846
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.192244405220717e-05,
+      "loss": 1.6849,
+      "step": 5847
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.19090654270971e-05,
+      "loss": 1.6656,
+      "step": 5848
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.189568680198701e-05,
+      "loss": 1.5209,
+      "step": 5849
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.188230817687694e-05,
+      "loss": 1.5562,
+      "step": 5850
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.1868929551766864e-05,
+      "loss": 1.6654,
+      "step": 5851
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.185555092665679e-05,
+      "loss": 1.4474,
+      "step": 5852
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.1842172301546704e-05,
+      "loss": 1.6367,
+      "step": 5853
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.182879367643663e-05,
+      "loss": 1.5212,
+      "step": 5854
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.181541505132656e-05,
+      "loss": 1.1735,
+      "step": 5855
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.1802036426216484e-05,
+      "loss": 1.4116,
+      "step": 5856
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.1788657801106404e-05,
+      "loss": 1.3182,
+      "step": 5857
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.1775279175996324e-05,
+      "loss": 1.4559,
+      "step": 5858
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.176190055088625e-05,
+      "loss": 1.4448,
+      "step": 5859
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.174852192577618e-05,
+      "loss": 1.1747,
+      "step": 5860
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.17351433006661e-05,
+      "loss": 1.4671,
+      "step": 5861
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.172176467555602e-05,
+      "loss": 1.219,
+      "step": 5862
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.1708386050445945e-05,
+      "loss": 1.3953,
+      "step": 5863
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.1695007425335865e-05,
+      "loss": 1.6538,
+      "step": 5864
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.168162880022579e-05,
+      "loss": 1.3572,
+      "step": 5865
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.166825017511571e-05,
+      "loss": 1.3014,
+      "step": 5866
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.165487155000564e-05,
+      "loss": 1.2569,
+      "step": 5867
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.164149292489556e-05,
+      "loss": 1.255,
+      "step": 5868
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.1628114299785485e-05,
+      "loss": 1.2515,
+      "step": 5869
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.161473567467541e-05,
+      "loss": 1.2022,
+      "step": 5870
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.160135704956533e-05,
+      "loss": 1.1958,
+      "step": 5871
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.158797842445525e-05,
+      "loss": 1.4179,
+      "step": 5872
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.157459979934518e-05,
+      "loss": 1.2215,
+      "step": 5873
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.1561221174235105e-05,
+      "loss": 1.1362,
+      "step": 5874
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.1547842549125025e-05,
+      "loss": 1.073,
+      "step": 5875
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.1534463924014945e-05,
+      "loss": 1.0713,
+      "step": 5876
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.152108529890487e-05,
+      "loss": 1.0196,
+      "step": 5877
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.15077066737948e-05,
+      "loss": 0.949,
+      "step": 5878
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.149432804868472e-05,
+      "loss": 0.8822,
+      "step": 5879
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.148094942357464e-05,
+      "loss": 0.9809,
+      "step": 5880
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.1467570798464566e-05,
+      "loss": 1.0464,
+      "step": 5881
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.145419217335449e-05,
+      "loss": 1.0638,
+      "step": 5882
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.144081354824441e-05,
+      "loss": 1.0929,
+      "step": 5883
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.142743492313433e-05,
+      "loss": 0.9633,
+      "step": 5884
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.141405629802426e-05,
+      "loss": 0.8508,
+      "step": 5885
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.1400677672914186e-05,
+      "loss": 0.8474,
+      "step": 5886
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.1387299047804106e-05,
+      "loss": 0.7205,
+      "step": 5887
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.1373920422694026e-05,
+      "loss": 0.8091,
+      "step": 5888
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.136054179758395e-05,
+      "loss": 0.9966,
+      "step": 5889
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.134716317247388e-05,
+      "loss": 0.9377,
+      "step": 5890
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.13337845473638e-05,
+      "loss": 2.0769,
+      "step": 5891
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.132040592225372e-05,
+      "loss": 1.7827,
+      "step": 5892
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.1307027297143646e-05,
+      "loss": 1.6424,
+      "step": 5893
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.129364867203357e-05,
+      "loss": 1.662,
+      "step": 5894
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.128027004692349e-05,
+      "loss": 1.6647,
+      "step": 5895
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.126689142181341e-05,
+      "loss": 1.6458,
+      "step": 5896
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.125351279670334e-05,
+      "loss": 1.4985,
+      "step": 5897
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.1240134171593267e-05,
+      "loss": 1.3784,
+      "step": 5898
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.1226755546483187e-05,
+      "loss": 1.4248,
+      "step": 5899
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.1213376921373107e-05,
+      "loss": 1.6445,
+      "step": 5900
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.119999829626303e-05,
+      "loss": 1.4288,
+      "step": 5901
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.118661967115296e-05,
+      "loss": 1.4277,
+      "step": 5902
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.117324104604288e-05,
+      "loss": 1.706,
+      "step": 5903
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.115986242093281e-05,
+      "loss": 1.541,
+      "step": 5904
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.114648379582273e-05,
+      "loss": 1.4608,
+      "step": 5905
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.1133105170712654e-05,
+      "loss": 1.5307,
+      "step": 5906
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.1119726545602574e-05,
+      "loss": 1.6025,
+      "step": 5907
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.11063479204925e-05,
+      "loss": 1.2505,
+      "step": 5908
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.109296929538242e-05,
+      "loss": 1.4288,
+      "step": 5909
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.107959067027235e-05,
+      "loss": 1.2503,
+      "step": 5910
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.106621204516227e-05,
+      "loss": 1.4493,
+      "step": 5911
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.1052833420052194e-05,
+      "loss": 1.2732,
+      "step": 5912
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.1039454794942114e-05,
+      "loss": 1.4982,
+      "step": 5913
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.1026076169832034e-05,
+      "loss": 1.3851,
+      "step": 5914
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.101269754472196e-05,
+      "loss": 1.3289,
+      "step": 5915
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.099931891961189e-05,
+      "loss": 1.321,
+      "step": 5916
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.098594029450181e-05,
+      "loss": 1.2793,
+      "step": 5917
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.097256166939173e-05,
+      "loss": 1.2676,
+      "step": 5918
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.0959183044281654e-05,
+      "loss": 1.3074,
+      "step": 5919
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.094580441917158e-05,
+      "loss": 1.1069,
+      "step": 5920
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.093242579406151e-05,
+      "loss": 1.2219,
+      "step": 5921
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.091904716895142e-05,
+      "loss": 1.1254,
+      "step": 5922
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.090566854384135e-05,
+      "loss": 1.5344,
+      "step": 5923
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.0892289918731275e-05,
+      "loss": 1.311,
+      "step": 5924
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.08789112936212e-05,
+      "loss": 1.1513,
+      "step": 5925
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.0865532668511115e-05,
+      "loss": 1.1181,
+      "step": 5926
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.085215404340104e-05,
+      "loss": 1.0796,
+      "step": 5927
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.083877541829097e-05,
+      "loss": 1.1571,
+      "step": 5928
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.0825396793180895e-05,
+      "loss": 1.0052,
+      "step": 5929
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.081201816807081e-05,
+      "loss": 0.9428,
+      "step": 5930
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.0798639542960735e-05,
+      "loss": 1.0477,
+      "step": 5931
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.078526091785066e-05,
+      "loss": 0.9421,
+      "step": 5932
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.077188229274059e-05,
+      "loss": 0.6903,
+      "step": 5933
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.07585036676305e-05,
+      "loss": 0.9645,
+      "step": 5934
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.074512504252043e-05,
+      "loss": 0.9362,
+      "step": 5935
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.0731746417410355e-05,
+      "loss": 0.6264,
+      "step": 5936
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.071836779230028e-05,
+      "loss": 0.7526,
+      "step": 5937
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.07049891671902e-05,
+      "loss": 0.7217,
+      "step": 5938
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.069161054208012e-05,
+      "loss": 0.708,
+      "step": 5939
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.067823191697005e-05,
+      "loss": 0.9261,
+      "step": 5940
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.066485329185997e-05,
+      "loss": 1.9677,
+      "step": 5941
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.0651474666749896e-05,
+      "loss": 1.7551,
+      "step": 5942
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.0638096041639816e-05,
+      "loss": 1.702,
+      "step": 5943
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.062471741652974e-05,
+      "loss": 1.5967,
+      "step": 5944
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.061133879141966e-05,
+      "loss": 1.5968,
+      "step": 5945
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.059796016630959e-05,
+      "loss": 1.7479,
+      "step": 5946
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.058458154119951e-05,
+      "loss": 1.6872,
+      "step": 5947
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.0571202916089436e-05,
+      "loss": 1.7235,
+      "step": 5948
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.0557824290979356e-05,
+      "loss": 1.596,
+      "step": 5949
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.054444566586928e-05,
+      "loss": 1.3843,
+      "step": 5950
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.05310670407592e-05,
+      "loss": 1.5065,
+      "step": 5951
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.051768841564913e-05,
+      "loss": 1.3015,
+      "step": 5952
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.050430979053905e-05,
+      "loss": 1.3209,
+      "step": 5953
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.0490931165428976e-05,
+      "loss": 1.5701,
+      "step": 5954
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.04775525403189e-05,
+      "loss": 1.5913,
+      "step": 5955
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.046417391520882e-05,
+      "loss": 1.5356,
+      "step": 5956
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.045079529009874e-05,
+      "loss": 1.5915,
+      "step": 5957
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.043741666498867e-05,
+      "loss": 1.3631,
+      "step": 5958
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.04240380398786e-05,
+      "loss": 1.4723,
+      "step": 5959
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.041065941476852e-05,
+      "loss": 1.403,
+      "step": 5960
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.039728078965844e-05,
+      "loss": 1.3812,
+      "step": 5961
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.0383902164548364e-05,
+      "loss": 1.3443,
+      "step": 5962
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.037052353943829e-05,
+      "loss": 1.5439,
+      "step": 5963
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.035714491432821e-05,
+      "loss": 1.2645,
+      "step": 5964
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.034376628921813e-05,
+      "loss": 1.4349,
+      "step": 5965
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.033038766410806e-05,
+      "loss": 1.3441,
+      "step": 5966
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.0317009038997984e-05,
+      "loss": 1.307,
+      "step": 5967
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.03036304138879e-05,
+      "loss": 1.3716,
+      "step": 5968
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.0290251788777824e-05,
+      "loss": 1.3036,
+      "step": 5969
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.027687316366775e-05,
+      "loss": 1.3442,
+      "step": 5970
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.026349453855768e-05,
+      "loss": 1.2208,
+      "step": 5971
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.02501159134476e-05,
+      "loss": 1.1521,
+      "step": 5972
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.023673728833752e-05,
+      "loss": 0.948,
+      "step": 5973
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.0223358663227444e-05,
+      "loss": 1.0607,
+      "step": 5974
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.020998003811737e-05,
+      "loss": 1.3662,
+      "step": 5975
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.019660141300729e-05,
+      "loss": 1.0046,
+      "step": 5976
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.018322278789721e-05,
+      "loss": 1.2145,
+      "step": 5977
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.016984416278714e-05,
+      "loss": 1.0113,
+      "step": 5978
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.0156465537677065e-05,
+      "loss": 1.0024,
+      "step": 5979
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.0143086912566985e-05,
+      "loss": 1.0787,
+      "step": 5980
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.0129708287456905e-05,
+      "loss": 1.0227,
+      "step": 5981
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.011632966234683e-05,
+      "loss": 0.9127,
+      "step": 5982
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.010295103723676e-05,
+      "loss": 0.9515,
+      "step": 5983
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.008957241212668e-05,
+      "loss": 0.8833,
+      "step": 5984
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.0076193787016605e-05,
+      "loss": 0.8435,
+      "step": 5985
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.0062815161906525e-05,
+      "loss": 0.6976,
+      "step": 5986
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.004943653679645e-05,
+      "loss": 0.9407,
+      "step": 5987
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.003605791168637e-05,
+      "loss": 0.8476,
+      "step": 5988
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.00226792865763e-05,
+      "loss": 0.7399,
+      "step": 5989
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.000930066146622e-05,
+      "loss": 0.9244,
+      "step": 5990
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.9995922036356145e-05,
+      "loss": 1.9939,
+      "step": 5991
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.9982543411246065e-05,
+      "loss": 1.6247,
+      "step": 5992
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.996916478613599e-05,
+      "loss": 1.6967,
+      "step": 5993
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.995578616102591e-05,
+      "loss": 1.6432,
+      "step": 5994
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.994240753591583e-05,
+      "loss": 1.4766,
+      "step": 5995
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.992902891080576e-05,
+      "loss": 1.5849,
+      "step": 5996
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.9915650285695686e-05,
+      "loss": 1.5309,
+      "step": 5997
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.9902271660585606e-05,
+      "loss": 1.5013,
+      "step": 5998
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.9888893035475526e-05,
+      "loss": 1.4676,
+      "step": 5999
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.987551441036545e-05,
+      "loss": 1.4107,
+      "step": 6000
+    },
+    {
+      "epoch": 1.85,
+      "eval_bleu": 1.935996854081167e-15,
+      "eval_loss": 1.716983437538147,
+      "eval_runtime": 2547.8175,
+      "eval_samples_per_second": 5.793,
+      "eval_steps_per_second": 0.724,
+      "step": 6000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.986213578525538e-05,
+      "loss": 1.7726,
+      "step": 6001
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.9848757160145306e-05,
+      "loss": 1.4524,
+      "step": 6002
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.983537853503522e-05,
+      "loss": 1.4586,
+      "step": 6003
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.9821999909925146e-05,
+      "loss": 1.5466,
+      "step": 6004
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.980862128481507e-05,
+      "loss": 1.5258,
+      "step": 6005
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.9795242659705e-05,
+      "loss": 1.4703,
+      "step": 6006
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.978186403459491e-05,
+      "loss": 1.4103,
+      "step": 6007
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.976848540948484e-05,
+      "loss": 1.3665,
+      "step": 6008
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.9755106784374766e-05,
+      "loss": 1.3957,
+      "step": 6009
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.974172815926469e-05,
+      "loss": 1.3962,
+      "step": 6010
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.9728349534154606e-05,
+      "loss": 1.4448,
+      "step": 6011
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.971497090904453e-05,
+      "loss": 1.5973,
+      "step": 6012
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.970159228393446e-05,
+      "loss": 1.5247,
+      "step": 6013
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.9688213658824387e-05,
+      "loss": 1.3027,
+      "step": 6014
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.96748350337143e-05,
+      "loss": 1.5197,
+      "step": 6015
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.9661456408604227e-05,
+      "loss": 1.2958,
+      "step": 6016
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.964807778349415e-05,
+      "loss": 1.4188,
+      "step": 6017
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.963469915838408e-05,
+      "loss": 1.377,
+      "step": 6018
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.9621320533274e-05,
+      "loss": 1.3407,
+      "step": 6019
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.960794190816392e-05,
+      "loss": 1.2508,
+      "step": 6020
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.959456328305385e-05,
+      "loss": 1.2046,
+      "step": 6021
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.958118465794377e-05,
+      "loss": 1.3353,
+      "step": 6022
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.9567806032833694e-05,
+      "loss": 1.0874,
+      "step": 6023
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.9554427407723614e-05,
+      "loss": 1.0109,
+      "step": 6024
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.954104878261354e-05,
+      "loss": 1.2184,
+      "step": 6025
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.952767015750346e-05,
+      "loss": 1.1324,
+      "step": 6026
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.951429153239339e-05,
+      "loss": 1.1764,
+      "step": 6027
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.950091290728331e-05,
+      "loss": 1.1147,
+      "step": 6028
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.9487534282173234e-05,
+      "loss": 1.1629,
+      "step": 6029
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.9474155657063154e-05,
+      "loss": 1.2252,
+      "step": 6030
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.946077703195308e-05,
+      "loss": 0.9605,
+      "step": 6031
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.9447398406843e-05,
+      "loss": 0.989,
+      "step": 6032
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.943401978173293e-05,
+      "loss": 0.8173,
+      "step": 6033
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.942064115662285e-05,
+      "loss": 0.8842,
+      "step": 6034
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.9407262531512774e-05,
+      "loss": 0.9206,
+      "step": 6035
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.93938839064027e-05,
+      "loss": 0.7354,
+      "step": 6036
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.938050528129262e-05,
+      "loss": 0.6927,
+      "step": 6037
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.936712665618254e-05,
+      "loss": 0.7877,
+      "step": 6038
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.935374803107247e-05,
+      "loss": 0.6619,
+      "step": 6039
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.9340369405962395e-05,
+      "loss": 0.7793,
+      "step": 6040
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.9326990780852315e-05,
+      "loss": 1.9358,
+      "step": 6041
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.9313612155742235e-05,
+      "loss": 1.8227,
+      "step": 6042
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.930023353063216e-05,
+      "loss": 1.7016,
+      "step": 6043
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.928685490552209e-05,
+      "loss": 1.6113,
+      "step": 6044
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.9273476280412e-05,
+      "loss": 1.4135,
+      "step": 6045
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.926009765530193e-05,
+      "loss": 1.3864,
+      "step": 6046
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.9246719030191855e-05,
+      "loss": 1.6495,
+      "step": 6047
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.923334040508178e-05,
+      "loss": 1.432,
+      "step": 6048
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.9219961779971695e-05,
+      "loss": 1.5778,
+      "step": 6049
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.920658315486162e-05,
+      "loss": 1.5705,
+      "step": 6050
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.919320452975155e-05,
+      "loss": 1.3281,
+      "step": 6051
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.9179825904641475e-05,
+      "loss": 1.3682,
+      "step": 6052
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.9166447279531395e-05,
+      "loss": 1.2158,
+      "step": 6053
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.9153068654421315e-05,
+      "loss": 1.2683,
+      "step": 6054
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.913969002931124e-05,
+      "loss": 1.4774,
+      "step": 6055
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.912631140420117e-05,
+      "loss": 1.3569,
+      "step": 6056
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.911293277909109e-05,
+      "loss": 1.3719,
+      "step": 6057
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.909955415398101e-05,
+      "loss": 1.4078,
+      "step": 6058
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.9086175528870936e-05,
+      "loss": 1.4245,
+      "step": 6059
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.907279690376086e-05,
+      "loss": 1.3225,
+      "step": 6060
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.905941827865078e-05,
+      "loss": 1.1899,
+      "step": 6061
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.90460396535407e-05,
+      "loss": 1.2628,
+      "step": 6062
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.903266102843063e-05,
+      "loss": 1.3925,
+      "step": 6063
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.9019282403320556e-05,
+      "loss": 1.374,
+      "step": 6064
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.9005903778210476e-05,
+      "loss": 1.3251,
+      "step": 6065
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.8992525153100396e-05,
+      "loss": 1.385,
+      "step": 6066
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.897914652799032e-05,
+      "loss": 1.2928,
+      "step": 6067
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.896576790288025e-05,
+      "loss": 1.2088,
+      "step": 6068
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.895238927777017e-05,
+      "loss": 1.1052,
+      "step": 6069
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.8939010652660096e-05,
+      "loss": 1.2247,
+      "step": 6070
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.8925632027550016e-05,
+      "loss": 1.3285,
+      "step": 6071
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.8912253402439936e-05,
+      "loss": 1.1247,
+      "step": 6072
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.889887477732986e-05,
+      "loss": 1.4053,
+      "step": 6073
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.888549615221979e-05,
+      "loss": 1.279,
+      "step": 6074
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.887211752710971e-05,
+      "loss": 1.1138,
+      "step": 6075
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.885873890199963e-05,
+      "loss": 0.9967,
+      "step": 6076
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.884536027688956e-05,
+      "loss": 1.009,
+      "step": 6077
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.8831981651779484e-05,
+      "loss": 1.156,
+      "step": 6078
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.8818603026669404e-05,
+      "loss": 1.0447,
+      "step": 6079
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.8805224401559324e-05,
+      "loss": 0.8393,
+      "step": 6080
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.879184577644925e-05,
+      "loss": 0.8257,
+      "step": 6081
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.877846715133918e-05,
+      "loss": 1.0765,
+      "step": 6082
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.8765088526229104e-05,
+      "loss": 0.9429,
+      "step": 6083
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.875170990111902e-05,
+      "loss": 0.9114,
+      "step": 6084
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.8738331276008944e-05,
+      "loss": 1.0389,
+      "step": 6085
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.872495265089887e-05,
+      "loss": 0.9902,
+      "step": 6086
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.87115740257888e-05,
+      "loss": 0.9178,
+      "step": 6087
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.869819540067871e-05,
+      "loss": 0.8328,
+      "step": 6088
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.868481677556864e-05,
+      "loss": 0.7183,
+      "step": 6089
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.8671438150458564e-05,
+      "loss": 0.6411,
+      "step": 6090
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.865805952534849e-05,
+      "loss": 1.9909,
+      "step": 6091
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.8644680900238404e-05,
+      "loss": 1.5448,
+      "step": 6092
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.863130227512833e-05,
+      "loss": 1.5366,
+      "step": 6093
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.861792365001826e-05,
+      "loss": 1.5865,
+      "step": 6094
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.8604545024908185e-05,
+      "loss": 1.5435,
+      "step": 6095
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.85911663997981e-05,
+      "loss": 1.5021,
+      "step": 6096
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.8577787774688025e-05,
+      "loss": 1.524,
+      "step": 6097
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.856440914957795e-05,
+      "loss": 1.3071,
+      "step": 6098
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.855103052446787e-05,
+      "loss": 1.3986,
+      "step": 6099
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.85376518993578e-05,
+      "loss": 1.4322,
+      "step": 6100
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.852427327424772e-05,
+      "loss": 1.4911,
+      "step": 6101
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.8510894649137645e-05,
+      "loss": 1.3151,
+      "step": 6102
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.8497516024027565e-05,
+      "loss": 1.5433,
+      "step": 6103
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.848413739891749e-05,
+      "loss": 1.3241,
+      "step": 6104
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.847075877380741e-05,
+      "loss": 1.2056,
+      "step": 6105
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.845738014869734e-05,
+      "loss": 1.4854,
+      "step": 6106
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.844400152358726e-05,
+      "loss": 1.4723,
+      "step": 6107
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.8430622898477185e-05,
+      "loss": 1.4138,
+      "step": 6108
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.8417244273367105e-05,
+      "loss": 1.5422,
+      "step": 6109
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.840386564825703e-05,
+      "loss": 1.3759,
+      "step": 6110
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.839048702314695e-05,
+      "loss": 1.2483,
+      "step": 6111
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.837710839803688e-05,
+      "loss": 1.307,
+      "step": 6112
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.83637297729268e-05,
+      "loss": 1.4217,
+      "step": 6113
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.8350351147816726e-05,
+      "loss": 1.3691,
+      "step": 6114
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.8336972522706646e-05,
+      "loss": 1.5017,
+      "step": 6115
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.832359389759657e-05,
+      "loss": 1.1964,
+      "step": 6116
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.83102152724865e-05,
+      "loss": 1.1612,
+      "step": 6117
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.829683664737642e-05,
+      "loss": 1.1349,
+      "step": 6118
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.828345802226634e-05,
+      "loss": 1.2208,
+      "step": 6119
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.8270079397156266e-05,
+      "loss": 1.2853,
+      "step": 6120
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.825670077204619e-05,
+      "loss": 1.3531,
+      "step": 6121
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.824332214693611e-05,
+      "loss": 1.2318,
+      "step": 6122
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.822994352182603e-05,
+      "loss": 1.1042,
+      "step": 6123
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.821656489671596e-05,
+      "loss": 1.0656,
+      "step": 6124
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.8203186271605886e-05,
+      "loss": 1.0534,
+      "step": 6125
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.81898076464958e-05,
+      "loss": 1.138,
+      "step": 6126
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.8176429021385726e-05,
+      "loss": 1.0433,
+      "step": 6127
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.816305039627565e-05,
+      "loss": 1.1637,
+      "step": 6128
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.814967177116558e-05,
+      "loss": 0.9831,
+      "step": 6129
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.813629314605549e-05,
+      "loss": 0.9667,
+      "step": 6130
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.812291452094542e-05,
+      "loss": 1.0232,
+      "step": 6131
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.810953589583535e-05,
+      "loss": 0.8484,
+      "step": 6132
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.8096157270725273e-05,
+      "loss": 0.9503,
+      "step": 6133
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.8082778645615193e-05,
+      "loss": 1.032,
+      "step": 6134
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.8069400020505113e-05,
+      "loss": 0.9425,
+      "step": 6135
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.805602139539504e-05,
+      "loss": 0.835,
+      "step": 6136
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.804264277028497e-05,
+      "loss": 0.8382,
+      "step": 6137
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.802926414517489e-05,
+      "loss": 0.5981,
+      "step": 6138
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.801588552006481e-05,
+      "loss": 0.7811,
+      "step": 6139
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.8002506894954734e-05,
+      "loss": 0.7372,
+      "step": 6140
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.798912826984466e-05,
+      "loss": 1.7346,
+      "step": 6141
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.797574964473458e-05,
+      "loss": 1.7001,
+      "step": 6142
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.79623710196245e-05,
+      "loss": 1.5811,
+      "step": 6143
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.794899239451443e-05,
+      "loss": 1.5133,
+      "step": 6144
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.7935613769404354e-05,
+      "loss": 1.4302,
+      "step": 6145
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.7922235144294274e-05,
+      "loss": 1.6316,
+      "step": 6146
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.7908856519184194e-05,
+      "loss": 1.3824,
+      "step": 6147
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.789547789407412e-05,
+      "loss": 1.413,
+      "step": 6148
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.788209926896405e-05,
+      "loss": 1.4361,
+      "step": 6149
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.786872064385397e-05,
+      "loss": 1.5091,
+      "step": 6150
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.7855342018743894e-05,
+      "loss": 1.3234,
+      "step": 6151
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.7841963393633814e-05,
+      "loss": 1.4433,
+      "step": 6152
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.7828584768523734e-05,
+      "loss": 1.2817,
+      "step": 6153
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.781520614341366e-05,
+      "loss": 1.2719,
+      "step": 6154
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.780182751830359e-05,
+      "loss": 1.564,
+      "step": 6155
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.778844889319351e-05,
+      "loss": 1.5261,
+      "step": 6156
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.777507026808343e-05,
+      "loss": 1.5171,
+      "step": 6157
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.7761691642973355e-05,
+      "loss": 1.3209,
+      "step": 6158
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.774831301786328e-05,
+      "loss": 1.3548,
+      "step": 6159
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.77349343927532e-05,
+      "loss": 1.297,
+      "step": 6160
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.772155576764312e-05,
+      "loss": 1.2829,
+      "step": 6161
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.770817714253305e-05,
+      "loss": 1.2418,
+      "step": 6162
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.7694798517422975e-05,
+      "loss": 1.3476,
+      "step": 6163
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.7681419892312895e-05,
+      "loss": 1.2899,
+      "step": 6164
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.7668041267202815e-05,
+      "loss": 1.3306,
+      "step": 6165
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.765466264209274e-05,
+      "loss": 1.2292,
+      "step": 6166
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.764128401698267e-05,
+      "loss": 1.455,
+      "step": 6167
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.7627905391872595e-05,
+      "loss": 1.3008,
+      "step": 6168
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.761452676676251e-05,
+      "loss": 1.1992,
+      "step": 6169
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.7601148141652435e-05,
+      "loss": 1.308,
+      "step": 6170
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.758776951654236e-05,
+      "loss": 1.1713,
+      "step": 6171
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.757439089143229e-05,
+      "loss": 1.2048,
+      "step": 6172
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.75610122663222e-05,
+      "loss": 1.0907,
+      "step": 6173
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.754763364121213e-05,
+      "loss": 0.9655,
+      "step": 6174
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.7534255016102056e-05,
+      "loss": 1.1085,
+      "step": 6175
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.752087639099198e-05,
+      "loss": 1.056,
+      "step": 6176
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.7507497765881896e-05,
+      "loss": 1.2114,
+      "step": 6177
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.749411914077182e-05,
+      "loss": 1.0383,
+      "step": 6178
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.748074051566175e-05,
+      "loss": 1.0166,
+      "step": 6179
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.746736189055167e-05,
+      "loss": 0.9617,
+      "step": 6180
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.7453983265441596e-05,
+      "loss": 0.8874,
+      "step": 6181
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.7440604640331516e-05,
+      "loss": 0.9454,
+      "step": 6182
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.742722601522144e-05,
+      "loss": 1.0858,
+      "step": 6183
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.741384739011136e-05,
+      "loss": 0.8556,
+      "step": 6184
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.740046876500129e-05,
+      "loss": 0.8872,
+      "step": 6185
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.738709013989121e-05,
+      "loss": 0.8941,
+      "step": 6186
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.7373711514781136e-05,
+      "loss": 0.7777,
+      "step": 6187
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.7360332889671056e-05,
+      "loss": 0.6873,
+      "step": 6188
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.734695426456098e-05,
+      "loss": 0.7985,
+      "step": 6189
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.73335756394509e-05,
+      "loss": 0.5804,
+      "step": 6190
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.732019701434083e-05,
+      "loss": 1.866,
+      "step": 6191
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.730681838923075e-05,
+      "loss": 1.8526,
+      "step": 6192
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.729343976412068e-05,
+      "loss": 1.6118,
+      "step": 6193
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.72800611390106e-05,
+      "loss": 1.6608,
+      "step": 6194
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.7266682513900524e-05,
+      "loss": 1.8418,
+      "step": 6195
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.7253303888790444e-05,
+      "loss": 1.2385,
+      "step": 6196
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.723992526368037e-05,
+      "loss": 1.2432,
+      "step": 6197
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.72265466385703e-05,
+      "loss": 1.5902,
+      "step": 6198
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.721316801346022e-05,
+      "loss": 1.5739,
+      "step": 6199
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.719978938835014e-05,
+      "loss": 1.4488,
+      "step": 6200
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.7186410763240064e-05,
+      "loss": 1.5067,
+      "step": 6201
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.717303213812999e-05,
+      "loss": 1.5141,
+      "step": 6202
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.715965351301991e-05,
+      "loss": 1.5191,
+      "step": 6203
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.714627488790983e-05,
+      "loss": 1.3739,
+      "step": 6204
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.713289626279976e-05,
+      "loss": 1.3114,
+      "step": 6205
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.7119517637689684e-05,
+      "loss": 1.3111,
+      "step": 6206
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.71061390125796e-05,
+      "loss": 1.2392,
+      "step": 6207
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.7092760387469524e-05,
+      "loss": 1.283,
+      "step": 6208
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.707938176235945e-05,
+      "loss": 1.3832,
+      "step": 6209
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.706600313724938e-05,
+      "loss": 1.3643,
+      "step": 6210
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.705262451213929e-05,
+      "loss": 1.2914,
+      "step": 6211
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.703924588702922e-05,
+      "loss": 1.2295,
+      "step": 6212
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.7025867261919145e-05,
+      "loss": 1.336,
+      "step": 6213
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.701248863680907e-05,
+      "loss": 1.433,
+      "step": 6214
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.699911001169899e-05,
+      "loss": 1.3144,
+      "step": 6215
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.698573138658891e-05,
+      "loss": 1.2821,
+      "step": 6216
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.697235276147884e-05,
+      "loss": 1.2461,
+      "step": 6217
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.6958974136368765e-05,
+      "loss": 1.1671,
+      "step": 6218
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.6945595511258685e-05,
+      "loss": 1.1369,
+      "step": 6219
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.6932216886148605e-05,
+      "loss": 1.0881,
+      "step": 6220
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.691883826103853e-05,
+      "loss": 1.1682,
+      "step": 6221
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.690545963592846e-05,
+      "loss": 1.1851,
+      "step": 6222
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.689208101081838e-05,
+      "loss": 1.1611,
+      "step": 6223
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.68787023857083e-05,
+      "loss": 1.0381,
+      "step": 6224
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.6865323760598225e-05,
+      "loss": 1.0005,
+      "step": 6225
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.685194513548815e-05,
+      "loss": 1.0987,
+      "step": 6226
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.683856651037807e-05,
+      "loss": 1.0177,
+      "step": 6227
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.682518788526799e-05,
+      "loss": 0.9734,
+      "step": 6228
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.681180926015792e-05,
+      "loss": 0.982,
+      "step": 6229
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.6798430635047846e-05,
+      "loss": 1.0455,
+      "step": 6230
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.6785052009937766e-05,
+      "loss": 1.0259,
+      "step": 6231
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.677167338482769e-05,
+      "loss": 0.8396,
+      "step": 6232
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.675829475971761e-05,
+      "loss": 0.9758,
+      "step": 6233
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.674491613460753e-05,
+      "loss": 0.9343,
+      "step": 6234
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.673153750949746e-05,
+      "loss": 0.908,
+      "step": 6235
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.6718158884387386e-05,
+      "loss": 0.9154,
+      "step": 6236
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.6704780259277306e-05,
+      "loss": 0.8535,
+      "step": 6237
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.6691401634167226e-05,
+      "loss": 0.9074,
+      "step": 6238
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.667802300905715e-05,
+      "loss": 0.7304,
+      "step": 6239
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.666464438394708e-05,
+      "loss": 0.8329,
+      "step": 6240
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.6651265758837e-05,
+      "loss": 1.9061,
+      "step": 6241
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.663788713372692e-05,
+      "loss": 1.9911,
+      "step": 6242
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.6624508508616846e-05,
+      "loss": 1.4715,
+      "step": 6243
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.661112988350677e-05,
+      "loss": 1.58,
+      "step": 6244
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.659775125839669e-05,
+      "loss": 1.469,
+      "step": 6245
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.658437263328661e-05,
+      "loss": 1.4796,
+      "step": 6246
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.657099400817654e-05,
+      "loss": 1.466,
+      "step": 6247
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.655761538306647e-05,
+      "loss": 1.528,
+      "step": 6248
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.6544236757956393e-05,
+      "loss": 1.2688,
+      "step": 6249
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.653085813284631e-05,
+      "loss": 1.4012,
+      "step": 6250
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.6517479507736233e-05,
+      "loss": 1.2546,
+      "step": 6251
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.650410088262616e-05,
+      "loss": 1.7011,
+      "step": 6252
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.649072225751609e-05,
+      "loss": 1.531,
+      "step": 6253
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.6477343632406e-05,
+      "loss": 1.5034,
+      "step": 6254
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.646396500729593e-05,
+      "loss": 1.2619,
+      "step": 6255
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.6450586382185854e-05,
+      "loss": 1.3094,
+      "step": 6256
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.643720775707578e-05,
+      "loss": 1.2199,
+      "step": 6257
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.6423829131965694e-05,
+      "loss": 1.3951,
+      "step": 6258
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.641045050685562e-05,
+      "loss": 1.5422,
+      "step": 6259
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.639707188174555e-05,
+      "loss": 1.4639,
+      "step": 6260
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.638369325663547e-05,
+      "loss": 1.3529,
+      "step": 6261
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.637031463152539e-05,
+      "loss": 1.1796,
+      "step": 6262
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.6356936006415314e-05,
+      "loss": 1.3656,
+      "step": 6263
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.634355738130524e-05,
+      "loss": 1.311,
+      "step": 6264
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.633017875619516e-05,
+      "loss": 1.346,
+      "step": 6265
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.631680013108509e-05,
+      "loss": 1.1831,
+      "step": 6266
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.630342150597501e-05,
+      "loss": 1.2679,
+      "step": 6267
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.6290042880864934e-05,
+      "loss": 1.3882,
+      "step": 6268
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.6276664255754854e-05,
+      "loss": 1.1819,
+      "step": 6269
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.626328563064478e-05,
+      "loss": 1.1471,
+      "step": 6270
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.62499070055347e-05,
+      "loss": 1.2237,
+      "step": 6271
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.623652838042463e-05,
+      "loss": 1.1187,
+      "step": 6272
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.622314975531455e-05,
+      "loss": 1.2407,
+      "step": 6273
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.6209771130204475e-05,
+      "loss": 1.0216,
+      "step": 6274
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.6196392505094395e-05,
+      "loss": 1.2553,
+      "step": 6275
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.618301387998432e-05,
+      "loss": 1.0278,
+      "step": 6276
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.616963525487424e-05,
+      "loss": 1.108,
+      "step": 6277
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.615625662976417e-05,
+      "loss": 1.0149,
+      "step": 6278
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.6142878004654095e-05,
+      "loss": 0.9424,
+      "step": 6279
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.6129499379544015e-05,
+      "loss": 0.9842,
+      "step": 6280
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.6116120754433935e-05,
+      "loss": 0.9937,
+      "step": 6281
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.610274212932386e-05,
+      "loss": 0.9778,
+      "step": 6282
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.608936350421379e-05,
+      "loss": 0.9006,
+      "step": 6283
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.60759848791037e-05,
+      "loss": 0.8441,
+      "step": 6284
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.606260625399363e-05,
+      "loss": 0.959,
+      "step": 6285
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.6049227628883556e-05,
+      "loss": 0.8665,
+      "step": 6286
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.603584900377348e-05,
+      "loss": 0.7309,
+      "step": 6287
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.6022470378663396e-05,
+      "loss": 0.7101,
+      "step": 6288
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.600909175355332e-05,
+      "loss": 0.7085,
+      "step": 6289
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.599571312844325e-05,
+      "loss": 0.8401,
+      "step": 6290
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.5982334503333176e-05,
+      "loss": 1.8597,
+      "step": 6291
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.596895587822309e-05,
+      "loss": 1.7869,
+      "step": 6292
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.5955577253113016e-05,
+      "loss": 1.5603,
+      "step": 6293
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.594219862800294e-05,
+      "loss": 1.5701,
+      "step": 6294
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.592882000289287e-05,
+      "loss": 1.6178,
+      "step": 6295
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.591544137778279e-05,
+      "loss": 1.4568,
+      "step": 6296
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.590206275267271e-05,
+      "loss": 1.677,
+      "step": 6297
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.5888684127562636e-05,
+      "loss": 1.393,
+      "step": 6298
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.587530550245256e-05,
+      "loss": 1.61,
+      "step": 6299
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.586192687734248e-05,
+      "loss": 1.3302,
+      "step": 6300
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.58485482522324e-05,
+      "loss": 1.4648,
+      "step": 6301
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.583516962712233e-05,
+      "loss": 1.4494,
+      "step": 6302
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.5821791002012257e-05,
+      "loss": 1.4497,
+      "step": 6303
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.5808412376902177e-05,
+      "loss": 1.3457,
+      "step": 6304
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.5795033751792097e-05,
+      "loss": 1.5417,
+      "step": 6305
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.578165512668202e-05,
+      "loss": 1.3154,
+      "step": 6306
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.576827650157195e-05,
+      "loss": 1.3056,
+      "step": 6307
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.575489787646187e-05,
+      "loss": 1.2042,
+      "step": 6308
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.574151925135179e-05,
+      "loss": 1.2417,
+      "step": 6309
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.572814062624172e-05,
+      "loss": 1.2909,
+      "step": 6310
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.571476200113164e-05,
+      "loss": 1.2798,
+      "step": 6311
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.5701383376021564e-05,
+      "loss": 1.208,
+      "step": 6312
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.568800475091149e-05,
+      "loss": 1.3878,
+      "step": 6313
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.567462612580141e-05,
+      "loss": 1.4746,
+      "step": 6314
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.566124750069133e-05,
+      "loss": 1.5041,
+      "step": 6315
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.564786887558126e-05,
+      "loss": 1.402,
+      "step": 6316
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.5634490250471184e-05,
+      "loss": 1.3617,
+      "step": 6317
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.5621111625361104e-05,
+      "loss": 1.437,
+      "step": 6318
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.5607733000251024e-05,
+      "loss": 1.1665,
+      "step": 6319
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.559435437514095e-05,
+      "loss": 1.0323,
+      "step": 6320
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.558097575003088e-05,
+      "loss": 1.2734,
+      "step": 6321
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.55675971249208e-05,
+      "loss": 1.0352,
+      "step": 6322
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.555421849981072e-05,
+      "loss": 1.2313,
+      "step": 6323
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.5540839874700644e-05,
+      "loss": 1.0664,
+      "step": 6324
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.552746124959057e-05,
+      "loss": 1.1546,
+      "step": 6325
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.551408262448049e-05,
+      "loss": 1.0156,
+      "step": 6326
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.550070399937041e-05,
+      "loss": 0.9776,
+      "step": 6327
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.548732537426034e-05,
+      "loss": 0.8683,
+      "step": 6328
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.5473946749150265e-05,
+      "loss": 0.9054,
+      "step": 6329
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.546056812404019e-05,
+      "loss": 1.1298,
+      "step": 6330
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.5447189498930105e-05,
+      "loss": 1.0232,
+      "step": 6331
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.543381087382003e-05,
+      "loss": 0.7576,
+      "step": 6332
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.542043224870996e-05,
+      "loss": 0.9965,
+      "step": 6333
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.5407053623599885e-05,
+      "loss": 0.8363,
+      "step": 6334
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.53936749984898e-05,
+      "loss": 0.8611,
+      "step": 6335
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.5380296373379725e-05,
+      "loss": 0.856,
+      "step": 6336
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.536691774826965e-05,
+      "loss": 0.6587,
+      "step": 6337
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.535353912315957e-05,
+      "loss": 0.7726,
+      "step": 6338
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.534016049804949e-05,
+      "loss": 0.7588,
+      "step": 6339
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.532678187293942e-05,
+      "loss": 0.7359,
+      "step": 6340
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.5313403247829345e-05,
+      "loss": 1.7999,
+      "step": 6341
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.5300024622719265e-05,
+      "loss": 1.6479,
+      "step": 6342
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.5286645997609185e-05,
+      "loss": 1.7483,
+      "step": 6343
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.527326737249911e-05,
+      "loss": 1.6508,
+      "step": 6344
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.525988874738904e-05,
+      "loss": 1.3934,
+      "step": 6345
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.524651012227896e-05,
+      "loss": 1.4154,
+      "step": 6346
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.5233131497168886e-05,
+      "loss": 1.536,
+      "step": 6347
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.5219752872058806e-05,
+      "loss": 1.4093,
+      "step": 6348
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.520637424694873e-05,
+      "loss": 1.4934,
+      "step": 6349
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.519299562183865e-05,
+      "loss": 1.4134,
+      "step": 6350
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.517961699672858e-05,
+      "loss": 1.2365,
+      "step": 6351
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.51662383716185e-05,
+      "loss": 1.5762,
+      "step": 6352
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.5152859746508426e-05,
+      "loss": 1.4648,
+      "step": 6353
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.5139481121398346e-05,
+      "loss": 1.4762,
+      "step": 6354
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.512610249628827e-05,
+      "loss": 1.2037,
+      "step": 6355
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.511272387117819e-05,
+      "loss": 1.4006,
+      "step": 6356
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.509934524606812e-05,
+      "loss": 1.2577,
+      "step": 6357
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.508596662095804e-05,
+      "loss": 1.3766,
+      "step": 6358
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.5072587995847966e-05,
+      "loss": 1.5581,
+      "step": 6359
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.5059209370737886e-05,
+      "loss": 1.3277,
+      "step": 6360
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.504583074562781e-05,
+      "loss": 1.2751,
+      "step": 6361
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.503245212051773e-05,
+      "loss": 1.3104,
+      "step": 6362
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.501907349540766e-05,
+      "loss": 1.2445,
+      "step": 6363
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.500569487029759e-05,
+      "loss": 1.1138,
+      "step": 6364
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.49923162451875e-05,
+      "loss": 1.4709,
+      "step": 6365
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.497893762007743e-05,
+      "loss": 1.4458,
+      "step": 6366
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.4965558994967353e-05,
+      "loss": 1.1891,
+      "step": 6367
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.495218036985728e-05,
+      "loss": 1.2706,
+      "step": 6368
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.4938801744747194e-05,
+      "loss": 1.0326,
+      "step": 6369
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.492542311963712e-05,
+      "loss": 1.1102,
+      "step": 6370
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.491204449452705e-05,
+      "loss": 1.199,
+      "step": 6371
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.4898665869416974e-05,
+      "loss": 1.3185,
+      "step": 6372
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.488528724430689e-05,
+      "loss": 1.1687,
+      "step": 6373
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.4871908619196814e-05,
+      "loss": 1.3289,
+      "step": 6374
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.485852999408674e-05,
+      "loss": 1.1518,
+      "step": 6375
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.484515136897667e-05,
+      "loss": 1.1921,
+      "step": 6376
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.483177274386659e-05,
+      "loss": 1.0264,
+      "step": 6377
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.481839411875651e-05,
+      "loss": 0.9249,
+      "step": 6378
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.4805015493646434e-05,
+      "loss": 0.9538,
+      "step": 6379
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.479163686853636e-05,
+      "loss": 0.8264,
+      "step": 6380
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.477825824342628e-05,
+      "loss": 0.7994,
+      "step": 6381
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.47648796183162e-05,
+      "loss": 0.9796,
+      "step": 6382
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.475150099320613e-05,
+      "loss": 0.8253,
+      "step": 6383
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.4738122368096055e-05,
+      "loss": 0.764,
+      "step": 6384
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.4724743742985975e-05,
+      "loss": 0.875,
+      "step": 6385
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.4711365117875895e-05,
+      "loss": 0.7574,
+      "step": 6386
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.469798649276582e-05,
+      "loss": 0.7069,
+      "step": 6387
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.468460786765575e-05,
+      "loss": 0.7616,
+      "step": 6388
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.467122924254567e-05,
+      "loss": 0.6523,
+      "step": 6389
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.465785061743559e-05,
+      "loss": 0.8889,
+      "step": 6390
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.4644471992325515e-05,
+      "loss": 2.0642,
+      "step": 6391
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.4631093367215435e-05,
+      "loss": 1.7803,
+      "step": 6392
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.461771474210536e-05,
+      "loss": 1.47,
+      "step": 6393
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.460433611699529e-05,
+      "loss": 1.4064,
+      "step": 6394
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.459095749188521e-05,
+      "loss": 1.3875,
+      "step": 6395
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.457757886677513e-05,
+      "loss": 1.4352,
+      "step": 6396
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.4564200241665055e-05,
+      "loss": 1.3636,
+      "step": 6397
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.455082161655498e-05,
+      "loss": 1.496,
+      "step": 6398
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.45374429914449e-05,
+      "loss": 1.3023,
+      "step": 6399
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.452406436633482e-05,
+      "loss": 1.3605,
+      "step": 6400
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.451068574122475e-05,
+      "loss": 1.4606,
+      "step": 6401
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.4497307116114676e-05,
+      "loss": 1.5986,
+      "step": 6402
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.4483928491004596e-05,
+      "loss": 1.4434,
+      "step": 6403
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.4470549865894516e-05,
+      "loss": 1.4673,
+      "step": 6404
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.445717124078444e-05,
+      "loss": 1.4415,
+      "step": 6405
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.444379261567437e-05,
+      "loss": 1.3582,
+      "step": 6406
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.443041399056429e-05,
+      "loss": 1.364,
+      "step": 6407
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.441703536545421e-05,
+      "loss": 1.4402,
+      "step": 6408
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.4403656740344136e-05,
+      "loss": 1.2907,
+      "step": 6409
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.439027811523406e-05,
+      "loss": 1.384,
+      "step": 6410
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.437689949012399e-05,
+      "loss": 1.2254,
+      "step": 6411
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.43635208650139e-05,
+      "loss": 1.223,
+      "step": 6412
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.435014223990383e-05,
+      "loss": 1.3357,
+      "step": 6413
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.4336763614793756e-05,
+      "loss": 1.2269,
+      "step": 6414
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.432338498968368e-05,
+      "loss": 1.2101,
+      "step": 6415
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.4310006364573596e-05,
+      "loss": 1.0891,
+      "step": 6416
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.429662773946352e-05,
+      "loss": 1.3259,
+      "step": 6417
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.428324911435345e-05,
+      "loss": 1.1699,
+      "step": 6418
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.426987048924337e-05,
+      "loss": 1.149,
+      "step": 6419
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.425649186413329e-05,
+      "loss": 1.223,
+      "step": 6420
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.4243113239023217e-05,
+      "loss": 0.9843,
+      "step": 6421
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.422973461391314e-05,
+      "loss": 1.101,
+      "step": 6422
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.421635598880306e-05,
+      "loss": 1.1258,
+      "step": 6423
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.420297736369298e-05,
+      "loss": 0.9469,
+      "step": 6424
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.418959873858291e-05,
+      "loss": 1.1661,
+      "step": 6425
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.417622011347284e-05,
+      "loss": 1.0833,
+      "step": 6426
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.416284148836276e-05,
+      "loss": 0.8523,
+      "step": 6427
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.4149462863252684e-05,
+      "loss": 1.1017,
+      "step": 6428
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.4136084238142604e-05,
+      "loss": 0.9325,
+      "step": 6429
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.412270561303253e-05,
+      "loss": 1.0167,
+      "step": 6430
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.410932698792245e-05,
+      "loss": 1.0678,
+      "step": 6431
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.409594836281238e-05,
+      "loss": 0.8582,
+      "step": 6432
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.40825697377023e-05,
+      "loss": 1.1287,
+      "step": 6433
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.4069191112592224e-05,
+      "loss": 0.8807,
+      "step": 6434
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.4055812487482144e-05,
+      "loss": 0.7123,
+      "step": 6435
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.404243386237207e-05,
+      "loss": 0.911,
+      "step": 6436
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.402905523726199e-05,
+      "loss": 0.7817,
+      "step": 6437
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.401567661215192e-05,
+      "loss": 0.7122,
+      "step": 6438
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.400229798704184e-05,
+      "loss": 0.8263,
+      "step": 6439
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.3988919361931764e-05,
+      "loss": 0.764,
+      "step": 6440
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.3975540736821684e-05,
+      "loss": 1.7527,
+      "step": 6441
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.396216211171161e-05,
+      "loss": 1.9342,
+      "step": 6442
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.394878348660153e-05,
+      "loss": 1.3405,
+      "step": 6443
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.393540486149146e-05,
+      "loss": 1.5921,
+      "step": 6444
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.3922026236381385e-05,
+      "loss": 1.4789,
+      "step": 6445
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.39086476112713e-05,
+      "loss": 1.2643,
+      "step": 6446
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.3895268986161225e-05,
+      "loss": 1.3825,
+      "step": 6447
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.388189036105115e-05,
+      "loss": 1.4703,
+      "step": 6448
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.386851173594108e-05,
+      "loss": 1.38,
+      "step": 6449
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.385513311083099e-05,
+      "loss": 1.3066,
+      "step": 6450
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.384175448572092e-05,
+      "loss": 1.4782,
+      "step": 6451
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.3828375860610845e-05,
+      "loss": 1.4303,
+      "step": 6452
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.381499723550077e-05,
+      "loss": 1.3347,
+      "step": 6453
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.3801618610390685e-05,
+      "loss": 1.4349,
+      "step": 6454
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.378823998528061e-05,
+      "loss": 1.2505,
+      "step": 6455
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.377486136017054e-05,
+      "loss": 1.4019,
+      "step": 6456
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.3761482735060465e-05,
+      "loss": 1.2703,
+      "step": 6457
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.374810410995038e-05,
+      "loss": 1.2958,
+      "step": 6458
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.3734725484840305e-05,
+      "loss": 1.18,
+      "step": 6459
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.372134685973023e-05,
+      "loss": 1.2873,
+      "step": 6460
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.370796823462016e-05,
+      "loss": 1.1696,
+      "step": 6461
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.369458960951008e-05,
+      "loss": 1.2631,
+      "step": 6462
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.36812109844e-05,
+      "loss": 1.0691,
+      "step": 6463
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.3667832359289926e-05,
+      "loss": 1.3986,
+      "step": 6464
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.365445373417985e-05,
+      "loss": 1.204,
+      "step": 6465
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.364107510906977e-05,
+      "loss": 1.3121,
+      "step": 6466
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.362769648395969e-05,
+      "loss": 1.147,
+      "step": 6467
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.361431785884962e-05,
+      "loss": 1.179,
+      "step": 6468
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.3600939233739546e-05,
+      "loss": 1.0772,
+      "step": 6469
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.3587560608629466e-05,
+      "loss": 1.017,
+      "step": 6470
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.3574181983519386e-05,
+      "loss": 1.0312,
+      "step": 6471
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.356080335840931e-05,
+      "loss": 0.8666,
+      "step": 6472
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.354742473329923e-05,
+      "loss": 1.0902,
+      "step": 6473
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.353404610818916e-05,
+      "loss": 0.8364,
+      "step": 6474
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.3520667483079086e-05,
+      "loss": 0.9688,
+      "step": 6475
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.3507288857969006e-05,
+      "loss": 0.7215,
+      "step": 6476
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.3493910232858926e-05,
+      "loss": 0.8373,
+      "step": 6477
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.348053160774885e-05,
+      "loss": 0.6872,
+      "step": 6478
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.346715298263878e-05,
+      "loss": 0.6432,
+      "step": 6479
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.34537743575287e-05,
+      "loss": 0.7174,
+      "step": 6480
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.344039573241862e-05,
+      "loss": 1.7415,
+      "step": 6481
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.342701710730855e-05,
+      "loss": 1.4159,
+      "step": 6482
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.3413638482198474e-05,
+      "loss": 1.1729,
+      "step": 6483
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.3400259857088394e-05,
+      "loss": 1.2834,
+      "step": 6484
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.3386881231978314e-05,
+      "loss": 1.2554,
+      "step": 6485
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.337350260686824e-05,
+      "loss": 1.1343,
+      "step": 6486
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.336012398175817e-05,
+      "loss": 1.0007,
+      "step": 6487
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.334674535664809e-05,
+      "loss": 1.1345,
+      "step": 6488
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.333336673153801e-05,
+      "loss": 1.2119,
+      "step": 6489
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.3319988106427934e-05,
+      "loss": 0.8776,
+      "step": 6490
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.330660948131786e-05,
+      "loss": 1.0269,
+      "step": 6491
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.329323085620779e-05,
+      "loss": 1.2212,
+      "step": 6492
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.32798522310977e-05,
+      "loss": 1.2068,
+      "step": 6493
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.326647360598763e-05,
+      "loss": 1.0975,
+      "step": 6494
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.3253094980877554e-05,
+      "loss": 1.0771,
+      "step": 6495
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.323971635576748e-05,
+      "loss": 1.1694,
+      "step": 6496
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.3226337730657394e-05,
+      "loss": 1.1768,
+      "step": 6497
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.321295910554732e-05,
+      "loss": 1.1367,
+      "step": 6498
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.319958048043725e-05,
+      "loss": 1.039,
+      "step": 6499
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.318620185532717e-05,
+      "loss": 1.0363,
+      "step": 6500
+    },
+    {
+      "epoch": 2.01,
+      "eval_bleu": 5.411484461414665e-14,
+      "eval_loss": 1.7418849468231201,
+      "eval_runtime": 2548.0612,
+      "eval_samples_per_second": 5.793,
+      "eval_steps_per_second": 0.724,
+      "step": 6500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.317282323021709e-05,
+      "loss": 1.129,
+      "step": 6501
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.3159444605107015e-05,
+      "loss": 1.1773,
+      "step": 6502
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.314606597999694e-05,
+      "loss": 1.0551,
+      "step": 6503
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.313268735488686e-05,
+      "loss": 1.0376,
+      "step": 6504
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.311930872977678e-05,
+      "loss": 0.9158,
+      "step": 6505
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.310593010466671e-05,
+      "loss": 1.0634,
+      "step": 6506
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.3092551479556635e-05,
+      "loss": 0.9095,
+      "step": 6507
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.3079172854446555e-05,
+      "loss": 0.8493,
+      "step": 6508
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.306579422933648e-05,
+      "loss": 1.001,
+      "step": 6509
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.30524156042264e-05,
+      "loss": 0.9876,
+      "step": 6510
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.303903697911633e-05,
+      "loss": 0.8626,
+      "step": 6511
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.302565835400625e-05,
+      "loss": 0.8142,
+      "step": 6512
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.3012279728896175e-05,
+      "loss": 0.9781,
+      "step": 6513
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.2998901103786095e-05,
+      "loss": 0.733,
+      "step": 6514
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.298552247867602e-05,
+      "loss": 0.8565,
+      "step": 6515
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.297214385356594e-05,
+      "loss": 0.8474,
+      "step": 6516
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.295876522845587e-05,
+      "loss": 0.8806,
+      "step": 6517
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.294538660334579e-05,
+      "loss": 0.7193,
+      "step": 6518
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.2932007978235716e-05,
+      "loss": 0.7227,
+      "step": 6519
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.2918629353125636e-05,
+      "loss": 0.7118,
+      "step": 6520
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.290525072801556e-05,
+      "loss": 0.602,
+      "step": 6521
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.289187210290548e-05,
+      "loss": 0.6237,
+      "step": 6522
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.28784934777954e-05,
+      "loss": 0.7852,
+      "step": 6523
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.286511485268533e-05,
+      "loss": 0.619,
+      "step": 6524
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.2851736227575256e-05,
+      "loss": 0.7978,
+      "step": 6525
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.283835760246518e-05,
+      "loss": 0.6667,
+      "step": 6526
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.2824978977355096e-05,
+      "loss": 0.6758,
+      "step": 6527
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.281160035224502e-05,
+      "loss": 0.6016,
+      "step": 6528
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.279822172713495e-05,
+      "loss": 0.6239,
+      "step": 6529
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.2784843102024876e-05,
+      "loss": 0.7371,
+      "step": 6530
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.277146447691479e-05,
+      "loss": 1.5151,
+      "step": 6531
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.2758085851804716e-05,
+      "loss": 1.416,
+      "step": 6532
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.274470722669464e-05,
+      "loss": 1.3348,
+      "step": 6533
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.273132860158457e-05,
+      "loss": 1.2624,
+      "step": 6534
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.271794997647448e-05,
+      "loss": 1.3101,
+      "step": 6535
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.270457135136441e-05,
+      "loss": 1.2977,
+      "step": 6536
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.2691192726254337e-05,
+      "loss": 1.2022,
+      "step": 6537
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.267781410114426e-05,
+      "loss": 1.2036,
+      "step": 6538
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.2664435476034177e-05,
+      "loss": 1.1464,
+      "step": 6539
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.26510568509241e-05,
+      "loss": 1.2877,
+      "step": 6540
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.263767822581403e-05,
+      "loss": 1.1524,
+      "step": 6541
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.262429960070396e-05,
+      "loss": 1.0612,
+      "step": 6542
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.261092097559388e-05,
+      "loss": 1.2615,
+      "step": 6543
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.25975423504838e-05,
+      "loss": 1.0962,
+      "step": 6544
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.2584163725373724e-05,
+      "loss": 1.1894,
+      "step": 6545
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.257078510026365e-05,
+      "loss": 1.1291,
+      "step": 6546
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.255740647515357e-05,
+      "loss": 1.044,
+      "step": 6547
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.254402785004349e-05,
+      "loss": 1.0775,
+      "step": 6548
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.253064922493342e-05,
+      "loss": 1.0967,
+      "step": 6549
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.251727059982334e-05,
+      "loss": 1.0799,
+      "step": 6550
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.2503891974713264e-05,
+      "loss": 0.9569,
+      "step": 6551
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.2490513349603184e-05,
+      "loss": 1.0316,
+      "step": 6552
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.247713472449311e-05,
+      "loss": 0.908,
+      "step": 6553
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.246375609938303e-05,
+      "loss": 1.0174,
+      "step": 6554
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.245037747427296e-05,
+      "loss": 0.8276,
+      "step": 6555
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.243699884916288e-05,
+      "loss": 0.9012,
+      "step": 6556
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.2423620224052804e-05,
+      "loss": 0.9855,
+      "step": 6557
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.2410241598942724e-05,
+      "loss": 0.9518,
+      "step": 6558
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.239686297383265e-05,
+      "loss": 0.851,
+      "step": 6559
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.238348434872258e-05,
+      "loss": 0.9292,
+      "step": 6560
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.23701057236125e-05,
+      "loss": 0.8234,
+      "step": 6561
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.235672709850242e-05,
+      "loss": 0.863,
+      "step": 6562
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.2343348473392345e-05,
+      "loss": 0.9017,
+      "step": 6563
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.232996984828227e-05,
+      "loss": 0.8498,
+      "step": 6564
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.231659122317219e-05,
+      "loss": 0.922,
+      "step": 6565
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.230321259806211e-05,
+      "loss": 0.7708,
+      "step": 6566
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.228983397295204e-05,
+      "loss": 0.7286,
+      "step": 6567
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.2276455347841965e-05,
+      "loss": 0.7275,
+      "step": 6568
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.2263076722731885e-05,
+      "loss": 0.7897,
+      "step": 6569
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.2249698097621805e-05,
+      "loss": 0.8636,
+      "step": 6570
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.223631947251173e-05,
+      "loss": 0.7634,
+      "step": 6571
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.222294084740166e-05,
+      "loss": 0.6486,
+      "step": 6572
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.2209562222291585e-05,
+      "loss": 0.7131,
+      "step": 6573
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.21961835971815e-05,
+      "loss": 0.6239,
+      "step": 6574
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.2182804972071425e-05,
+      "loss": 0.7023,
+      "step": 6575
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.216942634696135e-05,
+      "loss": 0.7066,
+      "step": 6576
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.215604772185127e-05,
+      "loss": 0.5755,
+      "step": 6577
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.214266909674119e-05,
+      "loss": 0.5842,
+      "step": 6578
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.212929047163112e-05,
+      "loss": 0.5618,
+      "step": 6579
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.2115911846521046e-05,
+      "loss": 0.632,
+      "step": 6580
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.2102533221410966e-05,
+      "loss": 1.678,
+      "step": 6581
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.2089154596300886e-05,
+      "loss": 1.3219,
+      "step": 6582
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.207577597119081e-05,
+      "loss": 1.3871,
+      "step": 6583
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.206239734608074e-05,
+      "loss": 1.1097,
+      "step": 6584
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.204901872097066e-05,
+      "loss": 1.2657,
+      "step": 6585
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.203564009586058e-05,
+      "loss": 1.0717,
+      "step": 6586
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.2022261470750506e-05,
+      "loss": 1.3048,
+      "step": 6587
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.200888284564043e-05,
+      "loss": 1.2288,
+      "step": 6588
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.199550422053035e-05,
+      "loss": 1.2926,
+      "step": 6589
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.198212559542028e-05,
+      "loss": 1.227,
+      "step": 6590
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.19687469703102e-05,
+      "loss": 1.14,
+      "step": 6591
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.1955368345200126e-05,
+      "loss": 1.0861,
+      "step": 6592
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.1941989720090046e-05,
+      "loss": 1.0058,
+      "step": 6593
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.192861109497997e-05,
+      "loss": 1.266,
+      "step": 6594
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.191523246986989e-05,
+      "loss": 1.1687,
+      "step": 6595
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.190185384475982e-05,
+      "loss": 1.0552,
+      "step": 6596
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.188847521964974e-05,
+      "loss": 1.0476,
+      "step": 6597
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.187509659453967e-05,
+      "loss": 1.1884,
+      "step": 6598
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.186171796942959e-05,
+      "loss": 1.0091,
+      "step": 6599
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.1848339344319514e-05,
+      "loss": 1.0852,
+      "step": 6600
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.1834960719209434e-05,
+      "loss": 0.964,
+      "step": 6601
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.182158209409936e-05,
+      "loss": 1.0061,
+      "step": 6602
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.180820346898928e-05,
+      "loss": 0.9918,
+      "step": 6603
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.17948248438792e-05,
+      "loss": 1.0241,
+      "step": 6604
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.178144621876913e-05,
+      "loss": 1.0701,
+      "step": 6605
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.1768067593659054e-05,
+      "loss": 1.1278,
+      "step": 6606
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.175468896854898e-05,
+      "loss": 1.1331,
+      "step": 6607
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.1741310343438894e-05,
+      "loss": 0.8988,
+      "step": 6608
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.172793171832882e-05,
+      "loss": 0.8157,
+      "step": 6609
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.171455309321875e-05,
+      "loss": 0.9534,
+      "step": 6610
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.1701174468108674e-05,
+      "loss": 0.7752,
+      "step": 6611
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.168779584299859e-05,
+      "loss": 0.9455,
+      "step": 6612
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.1674417217888514e-05,
+      "loss": 0.9125,
+      "step": 6613
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.166103859277844e-05,
+      "loss": 0.8434,
+      "step": 6614
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.164765996766837e-05,
+      "loss": 0.8331,
+      "step": 6615
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.163428134255828e-05,
+      "loss": 0.8721,
+      "step": 6616
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.162090271744821e-05,
+      "loss": 0.7847,
+      "step": 6617
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.1607524092338135e-05,
+      "loss": 0.8824,
+      "step": 6618
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.159414546722806e-05,
+      "loss": 0.7162,
+      "step": 6619
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.1580766842117975e-05,
+      "loss": 0.6184,
+      "step": 6620
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.15673882170079e-05,
+      "loss": 0.7236,
+      "step": 6621
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.155400959189783e-05,
+      "loss": 0.7326,
+      "step": 6622
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.1540630966787755e-05,
+      "loss": 0.5164,
+      "step": 6623
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.1527252341677675e-05,
+      "loss": 0.7524,
+      "step": 6624
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.1513873716567595e-05,
+      "loss": 0.5398,
+      "step": 6625
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.150049509145752e-05,
+      "loss": 0.5691,
+      "step": 6626
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.148711646634745e-05,
+      "loss": 0.6966,
+      "step": 6627
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.147373784123737e-05,
+      "loss": 0.5624,
+      "step": 6628
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.146035921612729e-05,
+      "loss": 0.4026,
+      "step": 6629
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.1446980591017215e-05,
+      "loss": 0.5669,
+      "step": 6630
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.1433601965907135e-05,
+      "loss": 1.4225,
+      "step": 6631
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.142022334079706e-05,
+      "loss": 1.5563,
+      "step": 6632
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.140684471568698e-05,
+      "loss": 1.2634,
+      "step": 6633
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.139346609057691e-05,
+      "loss": 1.2744,
+      "step": 6634
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.138008746546683e-05,
+      "loss": 1.1759,
+      "step": 6635
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.1366708840356756e-05,
+      "loss": 1.1416,
+      "step": 6636
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.1353330215246676e-05,
+      "loss": 1.0594,
+      "step": 6637
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.13399515901366e-05,
+      "loss": 1.4057,
+      "step": 6638
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.132657296502652e-05,
+      "loss": 1.1485,
+      "step": 6639
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.131319433991645e-05,
+      "loss": 1.0675,
+      "step": 6640
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.1299815714806376e-05,
+      "loss": 1.035,
+      "step": 6641
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.1286437089696296e-05,
+      "loss": 1.4295,
+      "step": 6642
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.1273058464586216e-05,
+      "loss": 1.25,
+      "step": 6643
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.125967983947614e-05,
+      "loss": 1.0279,
+      "step": 6644
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.124630121436607e-05,
+      "loss": 1.1487,
+      "step": 6645
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.123292258925599e-05,
+      "loss": 0.9965,
+      "step": 6646
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.121954396414591e-05,
+      "loss": 1.0375,
+      "step": 6647
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.1206165339035836e-05,
+      "loss": 1.0321,
+      "step": 6648
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.119278671392576e-05,
+      "loss": 1.1216,
+      "step": 6649
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.117940808881568e-05,
+      "loss": 1.1468,
+      "step": 6650
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.11660294637056e-05,
+      "loss": 0.9646,
+      "step": 6651
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.115265083859553e-05,
+      "loss": 1.0041,
+      "step": 6652
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.113927221348546e-05,
+      "loss": 1.0126,
+      "step": 6653
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.112589358837538e-05,
+      "loss": 1.0214,
+      "step": 6654
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.11125149632653e-05,
+      "loss": 1.0856,
+      "step": 6655
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.1099136338155223e-05,
+      "loss": 0.8337,
+      "step": 6656
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.108575771304515e-05,
+      "loss": 0.935,
+      "step": 6657
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.107237908793507e-05,
+      "loss": 0.869,
+      "step": 6658
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.105900046282499e-05,
+      "loss": 0.9458,
+      "step": 6659
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.104562183771492e-05,
+      "loss": 0.9135,
+      "step": 6660
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.1032243212604844e-05,
+      "loss": 0.9281,
+      "step": 6661
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.1018864587494764e-05,
+      "loss": 0.7598,
+      "step": 6662
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.1005485962384684e-05,
+      "loss": 0.8887,
+      "step": 6663
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.099210733727461e-05,
+      "loss": 0.7995,
+      "step": 6664
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.097872871216454e-05,
+      "loss": 0.8598,
+      "step": 6665
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.096535008705446e-05,
+      "loss": 0.8526,
+      "step": 6666
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.095197146194438e-05,
+      "loss": 0.8062,
+      "step": 6667
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.0938592836834304e-05,
+      "loss": 0.6986,
+      "step": 6668
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.092521421172423e-05,
+      "loss": 0.6915,
+      "step": 6669
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.091183558661415e-05,
+      "loss": 0.8401,
+      "step": 6670
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.089845696150407e-05,
+      "loss": 0.7551,
+      "step": 6671
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.0885078336394e-05,
+      "loss": 0.6359,
+      "step": 6672
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.0871699711283924e-05,
+      "loss": 0.6465,
+      "step": 6673
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.0858321086173844e-05,
+      "loss": 0.7036,
+      "step": 6674
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.084494246106377e-05,
+      "loss": 0.6826,
+      "step": 6675
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.083156383595369e-05,
+      "loss": 0.513,
+      "step": 6676
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.081818521084362e-05,
+      "loss": 0.7095,
+      "step": 6677
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.080480658573354e-05,
+      "loss": 0.6209,
+      "step": 6678
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.0791427960623465e-05,
+      "loss": 0.6377,
+      "step": 6679
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.0778049335513385e-05,
+      "loss": 0.5804,
+      "step": 6680
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.0764670710403305e-05,
+      "loss": 1.5597,
+      "step": 6681
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.075129208529323e-05,
+      "loss": 1.463,
+      "step": 6682
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.073791346018316e-05,
+      "loss": 1.3885,
+      "step": 6683
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.072453483507308e-05,
+      "loss": 1.261,
+      "step": 6684
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.0711156209963e-05,
+      "loss": 1.1861,
+      "step": 6685
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.0697777584852925e-05,
+      "loss": 1.1975,
+      "step": 6686
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.068439895974285e-05,
+      "loss": 1.3451,
+      "step": 6687
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.067102033463278e-05,
+      "loss": 1.1538,
+      "step": 6688
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.065764170952269e-05,
+      "loss": 1.3592,
+      "step": 6689
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.064426308441262e-05,
+      "loss": 1.1333,
+      "step": 6690
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.0630884459302545e-05,
+      "loss": 1.1763,
+      "step": 6691
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.061750583419247e-05,
+      "loss": 1.1707,
+      "step": 6692
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.0604127209082385e-05,
+      "loss": 1.1934,
+      "step": 6693
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.059074858397231e-05,
+      "loss": 1.2666,
+      "step": 6694
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.057736995886224e-05,
+      "loss": 1.139,
+      "step": 6695
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.0563991333752166e-05,
+      "loss": 1.0779,
+      "step": 6696
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.055061270864208e-05,
+      "loss": 1.1133,
+      "step": 6697
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.0537234083532006e-05,
+      "loss": 1.0907,
+      "step": 6698
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.052385545842193e-05,
+      "loss": 1.0536,
+      "step": 6699
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.051047683331186e-05,
+      "loss": 1.021,
+      "step": 6700
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.049709820820177e-05,
+      "loss": 1.0865,
+      "step": 6701
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.04837195830917e-05,
+      "loss": 1.0575,
+      "step": 6702
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.0470340957981626e-05,
+      "loss": 1.0417,
+      "step": 6703
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.045696233287155e-05,
+      "loss": 0.9943,
+      "step": 6704
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.044358370776147e-05,
+      "loss": 1.0101,
+      "step": 6705
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.043020508265139e-05,
+      "loss": 1.0879,
+      "step": 6706
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.041682645754132e-05,
+      "loss": 0.9526,
+      "step": 6707
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.040344783243124e-05,
+      "loss": 0.9627,
+      "step": 6708
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.0390069207321166e-05,
+      "loss": 0.8206,
+      "step": 6709
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.0376690582211086e-05,
+      "loss": 0.9138,
+      "step": 6710
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.036331195710101e-05,
+      "loss": 0.7922,
+      "step": 6711
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.034993333199093e-05,
+      "loss": 0.957,
+      "step": 6712
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.033655470688086e-05,
+      "loss": 0.9999,
+      "step": 6713
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.032317608177078e-05,
+      "loss": 0.7457,
+      "step": 6714
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.030979745666071e-05,
+      "loss": 0.8745,
+      "step": 6715
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.029641883155063e-05,
+      "loss": 0.9116,
+      "step": 6716
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.0283040206440554e-05,
+      "loss": 0.6604,
+      "step": 6717
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.0269661581330474e-05,
+      "loss": 0.8754,
+      "step": 6718
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.02562829562204e-05,
+      "loss": 0.8939,
+      "step": 6719
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.024290433111032e-05,
+      "loss": 0.7195,
+      "step": 6720
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.022952570600025e-05,
+      "loss": 0.8232,
+      "step": 6721
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.0216147080890174e-05,
+      "loss": 0.7084,
+      "step": 6722
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.0202768455780094e-05,
+      "loss": 0.6071,
+      "step": 6723
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.0189389830670014e-05,
+      "loss": 0.5694,
+      "step": 6724
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.017601120555994e-05,
+      "loss": 0.6596,
+      "step": 6725
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.016263258044987e-05,
+      "loss": 0.7254,
+      "step": 6726
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.014925395533979e-05,
+      "loss": 0.5502,
+      "step": 6727
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.013587533022971e-05,
+      "loss": 0.6516,
+      "step": 6728
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.0122496705119634e-05,
+      "loss": 0.5188,
+      "step": 6729
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.010911808000956e-05,
+      "loss": 0.8297,
+      "step": 6730
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.009573945489948e-05,
+      "loss": 1.721,
+      "step": 6731
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.00823608297894e-05,
+      "loss": 1.5481,
+      "step": 6732
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.006898220467933e-05,
+      "loss": 1.1512,
+      "step": 6733
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.0055603579569255e-05,
+      "loss": 1.1156,
+      "step": 6734
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.004222495445917e-05,
+      "loss": 1.2456,
+      "step": 6735
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.0028846329349095e-05,
+      "loss": 1.1758,
+      "step": 6736
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.001546770423902e-05,
+      "loss": 1.2947,
+      "step": 6737
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.000208907912895e-05,
+      "loss": 1.1571,
+      "step": 6738
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.998871045401887e-05,
+      "loss": 1.0143,
+      "step": 6739
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.997533182890879e-05,
+      "loss": 1.1374,
+      "step": 6740
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9961953203798715e-05,
+      "loss": 1.0775,
+      "step": 6741
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.994857457868864e-05,
+      "loss": 1.1404,
+      "step": 6742
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.993519595357856e-05,
+      "loss": 0.9824,
+      "step": 6743
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.992181732846848e-05,
+      "loss": 1.1176,
+      "step": 6744
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.990843870335841e-05,
+      "loss": 1.2284,
+      "step": 6745
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9895060078248335e-05,
+      "loss": 1.2173,
+      "step": 6746
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9881681453138255e-05,
+      "loss": 1.1513,
+      "step": 6747
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9868302828028175e-05,
+      "loss": 1.2024,
+      "step": 6748
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.98549242029181e-05,
+      "loss": 1.2556,
+      "step": 6749
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.984154557780803e-05,
+      "loss": 1.0339,
+      "step": 6750
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.982816695269795e-05,
+      "loss": 1.0329,
+      "step": 6751
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.981478832758787e-05,
+      "loss": 1.0237,
+      "step": 6752
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9801409702477796e-05,
+      "loss": 0.94,
+      "step": 6753
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.978803107736772e-05,
+      "loss": 1.1639,
+      "step": 6754
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.977465245225764e-05,
+      "loss": 0.8393,
+      "step": 6755
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.976127382714757e-05,
+      "loss": 0.8544,
+      "step": 6756
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.974789520203749e-05,
+      "loss": 1.1451,
+      "step": 6757
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.9734516576927416e-05,
+      "loss": 1.0917,
+      "step": 6758
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.9721137951817336e-05,
+      "loss": 0.715,
+      "step": 6759
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.970775932670726e-05,
+      "loss": 0.9857,
+      "step": 6760
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.969438070159718e-05,
+      "loss": 0.7344,
+      "step": 6761
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.96810020764871e-05,
+      "loss": 0.9461,
+      "step": 6762
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.966762345137703e-05,
+      "loss": 0.8342,
+      "step": 6763
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.9654244826266956e-05,
+      "loss": 0.8423,
+      "step": 6764
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.9640866201156876e-05,
+      "loss": 0.9496,
+      "step": 6765
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.9627487576046796e-05,
+      "loss": 0.8261,
+      "step": 6766
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.961410895093672e-05,
+      "loss": 0.8068,
+      "step": 6767
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.960073032582665e-05,
+      "loss": 0.7749,
+      "step": 6768
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.958735170071657e-05,
+      "loss": 0.7156,
+      "step": 6769
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.957397307560649e-05,
+      "loss": 0.6664,
+      "step": 6770
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.956059445049642e-05,
+      "loss": 0.6593,
+      "step": 6771
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.9547215825386343e-05,
+      "loss": 0.7193,
+      "step": 6772
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.953383720027627e-05,
+      "loss": 0.6517,
+      "step": 6773
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.9520458575166183e-05,
+      "loss": 0.5605,
+      "step": 6774
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.950707995005611e-05,
+      "loss": 0.6655,
+      "step": 6775
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.949370132494604e-05,
+      "loss": 0.5867,
+      "step": 6776
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.9480322699835964e-05,
+      "loss": 0.6075,
+      "step": 6777
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.946694407472588e-05,
+      "loss": 0.5935,
+      "step": 6778
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.9453565449615804e-05,
+      "loss": 0.7099,
+      "step": 6779
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.944018682450573e-05,
+      "loss": 0.7526,
+      "step": 6780
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.942680819939566e-05,
+      "loss": 1.5621,
+      "step": 6781
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.941342957428557e-05,
+      "loss": 1.3991,
+      "step": 6782
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.94000509491755e-05,
+      "loss": 1.2309,
+      "step": 6783
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.9386672324065424e-05,
+      "loss": 1.4232,
+      "step": 6784
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.937329369895535e-05,
+      "loss": 1.3017,
+      "step": 6785
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.935991507384527e-05,
+      "loss": 1.2848,
+      "step": 6786
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.934653644873519e-05,
+      "loss": 1.2886,
+      "step": 6787
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.933315782362512e-05,
+      "loss": 1.2414,
+      "step": 6788
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.931977919851504e-05,
+      "loss": 1.3525,
+      "step": 6789
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9306400573404964e-05,
+      "loss": 1.2157,
+      "step": 6790
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9293021948294884e-05,
+      "loss": 1.1083,
+      "step": 6791
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.927964332318481e-05,
+      "loss": 1.179,
+      "step": 6792
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.926626469807473e-05,
+      "loss": 1.1303,
+      "step": 6793
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.925288607296466e-05,
+      "loss": 1.1561,
+      "step": 6794
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.923950744785458e-05,
+      "loss": 1.0836,
+      "step": 6795
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9226128822744505e-05,
+      "loss": 0.9876,
+      "step": 6796
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9212750197634425e-05,
+      "loss": 1.1669,
+      "step": 6797
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.919937157252435e-05,
+      "loss": 1.0761,
+      "step": 6798
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.918599294741427e-05,
+      "loss": 1.1002,
+      "step": 6799
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.91726143223042e-05,
+      "loss": 1.1412,
+      "step": 6800
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.915923569719412e-05,
+      "loss": 1.1567,
+      "step": 6801
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9145857072084045e-05,
+      "loss": 0.9321,
+      "step": 6802
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.913247844697397e-05,
+      "loss": 1.0873,
+      "step": 6803
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.911909982186389e-05,
+      "loss": 1.0278,
+      "step": 6804
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.910572119675381e-05,
+      "loss": 0.9744,
+      "step": 6805
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.909234257164374e-05,
+      "loss": 1.083,
+      "step": 6806
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9078963946533666e-05,
+      "loss": 0.931,
+      "step": 6807
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9065585321423586e-05,
+      "loss": 0.8531,
+      "step": 6808
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9052206696313506e-05,
+      "loss": 0.9787,
+      "step": 6809
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.903882807120343e-05,
+      "loss": 0.9943,
+      "step": 6810
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.902544944609336e-05,
+      "loss": 0.9163,
+      "step": 6811
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.901207082098328e-05,
+      "loss": 0.8433,
+      "step": 6812
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.89986921958732e-05,
+      "loss": 0.7684,
+      "step": 6813
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.8985313570763126e-05,
+      "loss": 0.9809,
+      "step": 6814
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.897193494565305e-05,
+      "loss": 0.9446,
+      "step": 6815
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.8958556320542966e-05,
+      "loss": 0.9021,
+      "step": 6816
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.894517769543289e-05,
+      "loss": 0.851,
+      "step": 6817
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.893179907032282e-05,
+      "loss": 0.8051,
+      "step": 6818
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.8918420445212746e-05,
+      "loss": 0.8204,
+      "step": 6819
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.8905041820102666e-05,
+      "loss": 0.7256,
+      "step": 6820
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.8891663194992586e-05,
+      "loss": 0.802,
+      "step": 6821
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.887828456988251e-05,
+      "loss": 0.6097,
+      "step": 6822
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.886490594477244e-05,
+      "loss": 0.736,
+      "step": 6823
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.885152731966236e-05,
+      "loss": 0.5776,
+      "step": 6824
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.883814869455228e-05,
+      "loss": 0.6873,
+      "step": 6825
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.8824770069442207e-05,
+      "loss": 0.5682,
+      "step": 6826
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.881139144433213e-05,
+      "loss": 0.7105,
+      "step": 6827
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.879801281922205e-05,
+      "loss": 0.555,
+      "step": 6828
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.878463419411197e-05,
+      "loss": 0.5551,
+      "step": 6829
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.87712555690019e-05,
+      "loss": 0.6257,
+      "step": 6830
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.875787694389183e-05,
+      "loss": 1.5235,
+      "step": 6831
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.874449831878175e-05,
+      "loss": 1.3307,
+      "step": 6832
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.873111969367167e-05,
+      "loss": 1.5046,
+      "step": 6833
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.8717741068561594e-05,
+      "loss": 1.2997,
+      "step": 6834
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.870436244345152e-05,
+      "loss": 1.1131,
+      "step": 6835
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.869098381834144e-05,
+      "loss": 1.004,
+      "step": 6836
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.867760519323137e-05,
+      "loss": 1.2741,
+      "step": 6837
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.866422656812129e-05,
+      "loss": 1.2581,
+      "step": 6838
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.8650847943011214e-05,
+      "loss": 1.2203,
+      "step": 6839
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.8637469317901134e-05,
+      "loss": 1.1632,
+      "step": 6840
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.862409069279106e-05,
+      "loss": 1.0338,
+      "step": 6841
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.861071206768098e-05,
+      "loss": 1.2031,
+      "step": 6842
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.85973334425709e-05,
+      "loss": 1.1767,
+      "step": 6843
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.858395481746083e-05,
+      "loss": 1.1692,
+      "step": 6844
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.8570576192350754e-05,
+      "loss": 1.2577,
+      "step": 6845
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.8557197567240674e-05,
+      "loss": 1.2042,
+      "step": 6846
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.8543818942130594e-05,
+      "loss": 1.0757,
+      "step": 6847
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.853044031702052e-05,
+      "loss": 1.0267,
+      "step": 6848
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.851706169191045e-05,
+      "loss": 1.1837,
+      "step": 6849
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.850368306680037e-05,
+      "loss": 1.0467,
+      "step": 6850
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.849030444169029e-05,
+      "loss": 1.0168,
+      "step": 6851
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.8476925816580215e-05,
+      "loss": 1.0632,
+      "step": 6852
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.846354719147014e-05,
+      "loss": 1.0169,
+      "step": 6853
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.845016856636007e-05,
+      "loss": 1.1413,
+      "step": 6854
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.843678994124998e-05,
+      "loss": 0.9899,
+      "step": 6855
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.842341131613991e-05,
+      "loss": 1.1852,
+      "step": 6856
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.8410032691029835e-05,
+      "loss": 0.9685,
+      "step": 6857
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.839665406591976e-05,
+      "loss": 0.9503,
+      "step": 6858
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.8383275440809675e-05,
+      "loss": 1.0076,
+      "step": 6859
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.83698968156996e-05,
+      "loss": 0.8531,
+      "step": 6860
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.835651819058953e-05,
+      "loss": 0.8653,
+      "step": 6861
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.8343139565479455e-05,
+      "loss": 0.7717,
+      "step": 6862
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.832976094036937e-05,
+      "loss": 0.8163,
+      "step": 6863
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.8316382315259295e-05,
+      "loss": 0.9578,
+      "step": 6864
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.830300369014922e-05,
+      "loss": 0.9561,
+      "step": 6865
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.828962506503915e-05,
+      "loss": 0.945,
+      "step": 6866
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.827624643992906e-05,
+      "loss": 0.8506,
+      "step": 6867
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.826286781481899e-05,
+      "loss": 0.7819,
+      "step": 6868
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.8249489189708916e-05,
+      "loss": 0.6625,
+      "step": 6869
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.8236110564598836e-05,
+      "loss": 0.7179,
+      "step": 6870
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.822273193948876e-05,
+      "loss": 0.6207,
+      "step": 6871
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.820935331437868e-05,
+      "loss": 0.7359,
+      "step": 6872
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.819597468926861e-05,
+      "loss": 0.7244,
+      "step": 6873
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.818259606415853e-05,
+      "loss": 0.5981,
+      "step": 6874
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.8169217439048456e-05,
+      "loss": 0.6123,
+      "step": 6875
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.8155838813938376e-05,
+      "loss": 0.7489,
+      "step": 6876
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.81424601888283e-05,
+      "loss": 0.606,
+      "step": 6877
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.812908156371822e-05,
+      "loss": 0.5494,
+      "step": 6878
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.811570293860815e-05,
+      "loss": 0.6242,
+      "step": 6879
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.810232431349807e-05,
+      "loss": 0.6081,
+      "step": 6880
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.8088945688387996e-05,
+      "loss": 1.5611,
+      "step": 6881
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.8075567063277916e-05,
+      "loss": 1.4712,
+      "step": 6882
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.806218843816784e-05,
+      "loss": 1.3131,
+      "step": 6883
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.804880981305777e-05,
+      "loss": 1.0889,
+      "step": 6884
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.803543118794769e-05,
+      "loss": 1.24,
+      "step": 6885
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.802205256283761e-05,
+      "loss": 1.2212,
+      "step": 6886
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.800867393772754e-05,
+      "loss": 1.3297,
+      "step": 6887
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.7995295312617463e-05,
+      "loss": 1.1184,
+      "step": 6888
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.7981916687507383e-05,
+      "loss": 1.0883,
+      "step": 6889
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.7968538062397303e-05,
+      "loss": 1.2042,
+      "step": 6890
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.795515943728723e-05,
+      "loss": 1.1556,
+      "step": 6891
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.794178081217716e-05,
+      "loss": 1.1485,
+      "step": 6892
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.792840218706707e-05,
+      "loss": 1.1226,
+      "step": 6893
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.7915023561957e-05,
+      "loss": 0.9817,
+      "step": 6894
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.7901644936846924e-05,
+      "loss": 1.0763,
+      "step": 6895
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.788826631173685e-05,
+      "loss": 1.017,
+      "step": 6896
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.7874887686626764e-05,
+      "loss": 1.0135,
+      "step": 6897
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.786150906151669e-05,
+      "loss": 1.1597,
+      "step": 6898
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.784813043640662e-05,
+      "loss": 1.1343,
+      "step": 6899
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.7834751811296544e-05,
+      "loss": 1.0321,
+      "step": 6900
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.7821373186186464e-05,
+      "loss": 0.975,
+      "step": 6901
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.7807994561076384e-05,
+      "loss": 1.1819,
+      "step": 6902
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.779461593596631e-05,
+      "loss": 1.0535,
+      "step": 6903
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.778123731085624e-05,
+      "loss": 0.9207,
+      "step": 6904
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.776785868574616e-05,
+      "loss": 1.0397,
+      "step": 6905
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.775448006063608e-05,
+      "loss": 0.9199,
+      "step": 6906
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.7741101435526005e-05,
+      "loss": 1.0132,
+      "step": 6907
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.772772281041593e-05,
+      "loss": 0.9936,
+      "step": 6908
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.771434418530585e-05,
+      "loss": 0.9116,
+      "step": 6909
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.770096556019577e-05,
+      "loss": 1.0133,
+      "step": 6910
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.76875869350857e-05,
+      "loss": 0.8512,
+      "step": 6911
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.7674208309975625e-05,
+      "loss": 0.9382,
+      "step": 6912
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.7660829684865545e-05,
+      "loss": 0.9312,
+      "step": 6913
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.7647451059755465e-05,
+      "loss": 0.8767,
+      "step": 6914
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.763407243464539e-05,
+      "loss": 0.8133,
+      "step": 6915
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.762069380953532e-05,
+      "loss": 0.7413,
+      "step": 6916
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.760731518442524e-05,
+      "loss": 0.7296,
+      "step": 6917
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.7593936559315165e-05,
+      "loss": 0.8119,
+      "step": 6918
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.7580557934205085e-05,
+      "loss": 0.6747,
+      "step": 6919
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.7567179309095005e-05,
+      "loss": 0.9279,
+      "step": 6920
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.755380068398493e-05,
+      "loss": 0.845,
+      "step": 6921
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.754042205887486e-05,
+      "loss": 0.7627,
+      "step": 6922
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.752704343376478e-05,
+      "loss": 0.6516,
+      "step": 6923
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.75136648086547e-05,
+      "loss": 0.7802,
+      "step": 6924
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.7500286183544626e-05,
+      "loss": 0.7558,
+      "step": 6925
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.748690755843455e-05,
+      "loss": 0.6285,
+      "step": 6926
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.747352893332447e-05,
+      "loss": 0.6046,
+      "step": 6927
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.746015030821439e-05,
+      "loss": 0.5556,
+      "step": 6928
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.744677168310432e-05,
+      "loss": 0.6589,
+      "step": 6929
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.7433393057994246e-05,
+      "loss": 0.7325,
+      "step": 6930
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.7420014432884166e-05,
+      "loss": 1.5147,
+      "step": 6931
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.7406635807774086e-05,
+      "loss": 1.5163,
+      "step": 6932
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.739325718266401e-05,
+      "loss": 1.5061,
+      "step": 6933
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.737987855755394e-05,
+      "loss": 1.1215,
+      "step": 6934
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.7366499932443866e-05,
+      "loss": 1.306,
+      "step": 6935
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.735312130733378e-05,
+      "loss": 1.1633,
+      "step": 6936
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.7339742682223706e-05,
+      "loss": 1.1174,
+      "step": 6937
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.732636405711363e-05,
+      "loss": 1.2039,
+      "step": 6938
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.731298543200356e-05,
+      "loss": 1.1571,
+      "step": 6939
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.729960680689347e-05,
+      "loss": 1.0725,
+      "step": 6940
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.72862281817834e-05,
+      "loss": 1.1991,
+      "step": 6941
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.7272849556673327e-05,
+      "loss": 1.1643,
+      "step": 6942
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.725947093156325e-05,
+      "loss": 1.1677,
+      "step": 6943
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.7246092306453167e-05,
+      "loss": 1.0192,
+      "step": 6944
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.723271368134309e-05,
+      "loss": 1.0763,
+      "step": 6945
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.721933505623302e-05,
+      "loss": 1.1716,
+      "step": 6946
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.720595643112294e-05,
+      "loss": 0.9784,
+      "step": 6947
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.719257780601286e-05,
+      "loss": 1.0079,
+      "step": 6948
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.717919918090279e-05,
+      "loss": 1.2802,
+      "step": 6949
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.7165820555792714e-05,
+      "loss": 0.9786,
+      "step": 6950
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.7152441930682634e-05,
+      "loss": 1.1208,
+      "step": 6951
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.713906330557256e-05,
+      "loss": 0.911,
+      "step": 6952
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.712568468046248e-05,
+      "loss": 0.8922,
+      "step": 6953
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.711230605535241e-05,
+      "loss": 1.1131,
+      "step": 6954
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.709892743024233e-05,
+      "loss": 1.0737,
+      "step": 6955
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.7085548805132254e-05,
+      "loss": 0.9523,
+      "step": 6956
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.7072170180022174e-05,
+      "loss": 0.88,
+      "step": 6957
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.70587915549121e-05,
+      "loss": 0.8862,
+      "step": 6958
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.704541292980202e-05,
+      "loss": 1.0809,
+      "step": 6959
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.703203430469195e-05,
+      "loss": 0.8999,
+      "step": 6960
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.701865567958187e-05,
+      "loss": 0.9,
+      "step": 6961
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.7005277054471794e-05,
+      "loss": 0.9172,
+      "step": 6962
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.6991898429361714e-05,
+      "loss": 0.8571,
+      "step": 6963
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.697851980425164e-05,
+      "loss": 0.8241,
+      "step": 6964
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.696514117914156e-05,
+      "loss": 0.7978,
+      "step": 6965
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.695176255403149e-05,
+      "loss": 0.6746,
+      "step": 6966
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.693838392892141e-05,
+      "loss": 0.845,
+      "step": 6967
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.6925005303811335e-05,
+      "loss": 0.7676,
+      "step": 6968
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.691162667870126e-05,
+      "loss": 0.7087,
+      "step": 6969
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.689824805359118e-05,
+      "loss": 0.9036,
+      "step": 6970
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.68848694284811e-05,
+      "loss": 0.854,
+      "step": 6971
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.687149080337103e-05,
+      "loss": 0.7364,
+      "step": 6972
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.6858112178260955e-05,
+      "loss": 0.5165,
+      "step": 6973
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.684473355315087e-05,
+      "loss": 0.5035,
+      "step": 6974
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.6831354928040795e-05,
+      "loss": 0.7538,
+      "step": 6975
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.681797630293072e-05,
+      "loss": 0.6033,
+      "step": 6976
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.680459767782065e-05,
+      "loss": 0.7813,
+      "step": 6977
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.679121905271056e-05,
+      "loss": 0.5978,
+      "step": 6978
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.677784042760049e-05,
+      "loss": 0.574,
+      "step": 6979
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.6764461802490415e-05,
+      "loss": 0.6579,
+      "step": 6980
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.675108317738034e-05,
+      "loss": 1.3393,
+      "step": 6981
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.673770455227026e-05,
+      "loss": 1.4335,
+      "step": 6982
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.672432592716018e-05,
+      "loss": 1.4023,
+      "step": 6983
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.671094730205011e-05,
+      "loss": 1.1169,
+      "step": 6984
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.6697568676940036e-05,
+      "loss": 1.2473,
+      "step": 6985
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.6684190051829956e-05,
+      "loss": 1.2686,
+      "step": 6986
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.6670811426719876e-05,
+      "loss": 1.101,
+      "step": 6987
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.66574328016098e-05,
+      "loss": 1.0557,
+      "step": 6988
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.664405417649973e-05,
+      "loss": 1.0952,
+      "step": 6989
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.663067555138965e-05,
+      "loss": 1.1812,
+      "step": 6990
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.661729692627957e-05,
+      "loss": 1.1397,
+      "step": 6991
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.6603918301169496e-05,
+      "loss": 1.2522,
+      "step": 6992
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.659053967605942e-05,
+      "loss": 1.1637,
+      "step": 6993
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.657716105094934e-05,
+      "loss": 1.0268,
+      "step": 6994
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.656378242583926e-05,
+      "loss": 1.0377,
+      "step": 6995
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.655040380072919e-05,
+      "loss": 1.0597,
+      "step": 6996
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.6537025175619116e-05,
+      "loss": 1.0137,
+      "step": 6997
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.6523646550509036e-05,
+      "loss": 0.9125,
+      "step": 6998
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.651026792539896e-05,
+      "loss": 1.0091,
+      "step": 6999
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.649688930028888e-05,
+      "loss": 0.9667,
+      "step": 7000
+    },
+    {
+      "epoch": 2.16,
+      "eval_bleu": 1.1125921921971483e-14,
+      "eval_loss": 1.730904221534729,
+      "eval_runtime": 2559.8735,
+      "eval_samples_per_second": 5.766,
+      "eval_steps_per_second": 0.721,
+      "step": 7000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.64835106751788e-05,
+      "loss": 1.0139,
+      "step": 7001
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.647013205006873e-05,
+      "loss": 1.176,
+      "step": 7002
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.645675342495866e-05,
+      "loss": 1.0364,
+      "step": 7003
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.644337479984858e-05,
+      "loss": 1.1071,
+      "step": 7004
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.64299961747385e-05,
+      "loss": 0.8803,
+      "step": 7005
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.6416617549628424e-05,
+      "loss": 1.0674,
+      "step": 7006
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.640323892451835e-05,
+      "loss": 0.9719,
+      "step": 7007
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.638986029940827e-05,
+      "loss": 0.9324,
+      "step": 7008
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.637648167429819e-05,
+      "loss": 0.9149,
+      "step": 7009
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.636310304918812e-05,
+      "loss": 1.032,
+      "step": 7010
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.6349724424078044e-05,
+      "loss": 0.8289,
+      "step": 7011
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.6336345798967964e-05,
+      "loss": 1.0188,
+      "step": 7012
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.6322967173857884e-05,
+      "loss": 0.634,
+      "step": 7013
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.630958854874781e-05,
+      "loss": 0.8398,
+      "step": 7014
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.629620992363774e-05,
+      "loss": 0.9213,
+      "step": 7015
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.6282831298527664e-05,
+      "loss": 0.9316,
+      "step": 7016
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.626945267341758e-05,
+      "loss": 0.5852,
+      "step": 7017
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.6256074048307504e-05,
+      "loss": 0.7384,
+      "step": 7018
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.624269542319743e-05,
+      "loss": 0.6567,
+      "step": 7019
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.622931679808736e-05,
+      "loss": 0.6671,
+      "step": 7020
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.621593817297727e-05,
+      "loss": 0.7068,
+      "step": 7021
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.62025595478672e-05,
+      "loss": 0.7193,
+      "step": 7022
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.6189180922757125e-05,
+      "loss": 0.6962,
+      "step": 7023
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.617580229764705e-05,
+      "loss": 0.7266,
+      "step": 7024
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.6162423672536965e-05,
+      "loss": 0.5921,
+      "step": 7025
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.614904504742689e-05,
+      "loss": 0.692,
+      "step": 7026
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.613566642231682e-05,
+      "loss": 0.6484,
+      "step": 7027
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.612228779720674e-05,
+      "loss": 0.685,
+      "step": 7028
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.610890917209666e-05,
+      "loss": 0.466,
+      "step": 7029
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.6095530546986585e-05,
+      "loss": 0.735,
+      "step": 7030
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.608215192187651e-05,
+      "loss": 1.7453,
+      "step": 7031
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.606877329676643e-05,
+      "loss": 1.6398,
+      "step": 7032
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.605539467165636e-05,
+      "loss": 1.2721,
+      "step": 7033
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.604201604654628e-05,
+      "loss": 1.1045,
+      "step": 7034
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.6028637421436205e-05,
+      "loss": 1.2896,
+      "step": 7035
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.6015258796326125e-05,
+      "loss": 1.121,
+      "step": 7036
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.600188017121605e-05,
+      "loss": 1.3386,
+      "step": 7037
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.598850154610597e-05,
+      "loss": 1.1616,
+      "step": 7038
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.59751229209959e-05,
+      "loss": 1.1464,
+      "step": 7039
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.596174429588582e-05,
+      "loss": 1.0888,
+      "step": 7040
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.5948365670775746e-05,
+      "loss": 1.1194,
+      "step": 7041
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.5934987045665666e-05,
+      "loss": 1.1242,
+      "step": 7042
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.592160842055559e-05,
+      "loss": 1.2994,
+      "step": 7043
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.590822979544551e-05,
+      "loss": 1.1154,
+      "step": 7044
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.589485117033544e-05,
+      "loss": 1.0997,
+      "step": 7045
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.588147254522536e-05,
+      "loss": 1.274,
+      "step": 7046
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.5868093920115286e-05,
+      "loss": 0.9771,
+      "step": 7047
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.5854715295005206e-05,
+      "loss": 1.0079,
+      "step": 7048
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.584133666989513e-05,
+      "loss": 1.05,
+      "step": 7049
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.582795804478506e-05,
+      "loss": 0.9639,
+      "step": 7050
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.581457941967498e-05,
+      "loss": 0.8624,
+      "step": 7051
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.58012007945649e-05,
+      "loss": 1.0716,
+      "step": 7052
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.5787822169454826e-05,
+      "loss": 1.1316,
+      "step": 7053
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.577444354434475e-05,
+      "loss": 1.0766,
+      "step": 7054
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.5761064919234666e-05,
+      "loss": 1.0395,
+      "step": 7055
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.574768629412459e-05,
+      "loss": 0.9741,
+      "step": 7056
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.573430766901452e-05,
+      "loss": 1.1168,
+      "step": 7057
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.5720929043904447e-05,
+      "loss": 0.8737,
+      "step": 7058
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.570755041879436e-05,
+      "loss": 0.8128,
+      "step": 7059
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.5694171793684287e-05,
+      "loss": 0.9805,
+      "step": 7060
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.568079316857421e-05,
+      "loss": 0.7963,
+      "step": 7061
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.566741454346414e-05,
+      "loss": 0.8786,
+      "step": 7062
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.565403591835405e-05,
+      "loss": 0.8671,
+      "step": 7063
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.564065729324398e-05,
+      "loss": 0.7789,
+      "step": 7064
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.562727866813391e-05,
+      "loss": 0.9099,
+      "step": 7065
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.5613900043023834e-05,
+      "loss": 0.7971,
+      "step": 7066
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.5600521417913754e-05,
+      "loss": 0.6738,
+      "step": 7067
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.5587142792803674e-05,
+      "loss": 0.6903,
+      "step": 7068
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.55737641676936e-05,
+      "loss": 0.7707,
+      "step": 7069
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.556038554258353e-05,
+      "loss": 0.7859,
+      "step": 7070
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.554700691747345e-05,
+      "loss": 0.6429,
+      "step": 7071
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.553362829236337e-05,
+      "loss": 0.7513,
+      "step": 7072
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.5520249667253294e-05,
+      "loss": 0.668,
+      "step": 7073
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.550687104214322e-05,
+      "loss": 0.7834,
+      "step": 7074
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.549349241703314e-05,
+      "loss": 0.6132,
+      "step": 7075
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.548011379192306e-05,
+      "loss": 0.735,
+      "step": 7076
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.546673516681299e-05,
+      "loss": 0.6141,
+      "step": 7077
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.5453356541702914e-05,
+      "loss": 0.709,
+      "step": 7078
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.5439977916592834e-05,
+      "loss": 0.6313,
+      "step": 7079
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.542659929148276e-05,
+      "loss": 0.5688,
+      "step": 7080
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.541322066637268e-05,
+      "loss": 1.5295,
+      "step": 7081
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.53998420412626e-05,
+      "loss": 1.41,
+      "step": 7082
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.538646341615253e-05,
+      "loss": 1.1862,
+      "step": 7083
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5373084791042455e-05,
+      "loss": 1.4112,
+      "step": 7084
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5359706165932375e-05,
+      "loss": 1.3746,
+      "step": 7085
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5346327540822295e-05,
+      "loss": 1.1246,
+      "step": 7086
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.533294891571222e-05,
+      "loss": 1.2346,
+      "step": 7087
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.531957029060215e-05,
+      "loss": 1.1645,
+      "step": 7088
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.530619166549207e-05,
+      "loss": 1.3326,
+      "step": 7089
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.529281304038199e-05,
+      "loss": 1.2002,
+      "step": 7090
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5279434415271915e-05,
+      "loss": 1.1644,
+      "step": 7091
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.526605579016184e-05,
+      "loss": 1.0724,
+      "step": 7092
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.525267716505176e-05,
+      "loss": 1.2246,
+      "step": 7093
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.523929853994168e-05,
+      "loss": 1.0388,
+      "step": 7094
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.522591991483161e-05,
+      "loss": 0.9028,
+      "step": 7095
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5212541289721535e-05,
+      "loss": 1.0402,
+      "step": 7096
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.519916266461146e-05,
+      "loss": 1.0821,
+      "step": 7097
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5185784039501375e-05,
+      "loss": 1.0001,
+      "step": 7098
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.51724054143913e-05,
+      "loss": 0.9012,
+      "step": 7099
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.515902678928123e-05,
+      "loss": 1.1903,
+      "step": 7100
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5145648164171156e-05,
+      "loss": 1.0857,
+      "step": 7101
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.513226953906107e-05,
+      "loss": 1.1281,
+      "step": 7102
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5118890913950996e-05,
+      "loss": 0.8623,
+      "step": 7103
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.510551228884092e-05,
+      "loss": 1.1247,
+      "step": 7104
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.509213366373085e-05,
+      "loss": 1.0085,
+      "step": 7105
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.507875503862076e-05,
+      "loss": 0.8943,
+      "step": 7106
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.506537641351069e-05,
+      "loss": 0.8289,
+      "step": 7107
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5051997788400616e-05,
+      "loss": 0.8594,
+      "step": 7108
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5038619163290536e-05,
+      "loss": 0.8218,
+      "step": 7109
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5025240538180456e-05,
+      "loss": 0.7668,
+      "step": 7110
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.501186191307038e-05,
+      "loss": 0.712,
+      "step": 7111
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.499848328796031e-05,
+      "loss": 0.7139,
+      "step": 7112
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.498510466285023e-05,
+      "loss": 0.8957,
+      "step": 7113
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4971726037740156e-05,
+      "loss": 0.8603,
+      "step": 7114
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4958347412630076e-05,
+      "loss": 0.6055,
+      "step": 7115
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.494496878752e-05,
+      "loss": 0.8423,
+      "step": 7116
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.493159016240992e-05,
+      "loss": 0.7011,
+      "step": 7117
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.491821153729985e-05,
+      "loss": 0.7819,
+      "step": 7118
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.490483291218977e-05,
+      "loss": 0.8063,
+      "step": 7119
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.48914542870797e-05,
+      "loss": 0.8154,
+      "step": 7120
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.487807566196962e-05,
+      "loss": 0.8414,
+      "step": 7121
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4864697036859544e-05,
+      "loss": 0.6264,
+      "step": 7122
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4851318411749464e-05,
+      "loss": 0.6519,
+      "step": 7123
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.483793978663939e-05,
+      "loss": 0.7407,
+      "step": 7124
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.482456116152931e-05,
+      "loss": 0.7024,
+      "step": 7125
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.481118253641924e-05,
+      "loss": 0.6912,
+      "step": 7126
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.479780391130916e-05,
+      "loss": 0.6052,
+      "step": 7127
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4784425286199084e-05,
+      "loss": 0.6392,
+      "step": 7128
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4771046661089004e-05,
+      "loss": 0.5763,
+      "step": 7129
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.475766803597893e-05,
+      "loss": 0.5518,
+      "step": 7130
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.474428941086886e-05,
+      "loss": 1.4579,
+      "step": 7131
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.473091078575877e-05,
+      "loss": 1.4787,
+      "step": 7132
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.47175321606487e-05,
+      "loss": 1.1465,
+      "step": 7133
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4704153535538624e-05,
+      "loss": 1.3154,
+      "step": 7134
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.469077491042855e-05,
+      "loss": 1.2207,
+      "step": 7135
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4677396285318464e-05,
+      "loss": 1.191,
+      "step": 7136
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.466401766020839e-05,
+      "loss": 1.0758,
+      "step": 7137
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.465063903509832e-05,
+      "loss": 1.1823,
+      "step": 7138
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4637260409988245e-05,
+      "loss": 1.1547,
+      "step": 7139
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.462388178487816e-05,
+      "loss": 1.1476,
+      "step": 7140
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4610503159768085e-05,
+      "loss": 1.1198,
+      "step": 7141
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.459712453465801e-05,
+      "loss": 1.0969,
+      "step": 7142
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.458374590954794e-05,
+      "loss": 1.1886,
+      "step": 7143
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.457036728443785e-05,
+      "loss": 1.1853,
+      "step": 7144
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.455698865932778e-05,
+      "loss": 0.8926,
+      "step": 7145
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4543610034217705e-05,
+      "loss": 1.1602,
+      "step": 7146
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.453023140910763e-05,
+      "loss": 1.0647,
+      "step": 7147
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.451685278399755e-05,
+      "loss": 0.9724,
+      "step": 7148
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.450347415888747e-05,
+      "loss": 1.0656,
+      "step": 7149
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.44900955337774e-05,
+      "loss": 1.2559,
+      "step": 7150
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4476716908667325e-05,
+      "loss": 1.0205,
+      "step": 7151
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4463338283557245e-05,
+      "loss": 0.9099,
+      "step": 7152
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4449959658447165e-05,
+      "loss": 1.0816,
+      "step": 7153
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.443658103333709e-05,
+      "loss": 1.0522,
+      "step": 7154
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.442320240822702e-05,
+      "loss": 0.9687,
+      "step": 7155
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.440982378311694e-05,
+      "loss": 1.013,
+      "step": 7156
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.439644515800686e-05,
+      "loss": 1.0674,
+      "step": 7157
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4383066532896786e-05,
+      "loss": 1.0277,
+      "step": 7158
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4369687907786706e-05,
+      "loss": 0.945,
+      "step": 7159
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.435630928267663e-05,
+      "loss": 0.8249,
+      "step": 7160
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.434293065756655e-05,
+      "loss": 0.8315,
+      "step": 7161
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.432955203245648e-05,
+      "loss": 0.8303,
+      "step": 7162
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.43161734073464e-05,
+      "loss": 0.9511,
+      "step": 7163
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4302794782236326e-05,
+      "loss": 0.8453,
+      "step": 7164
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.428941615712625e-05,
+      "loss": 0.8181,
+      "step": 7165
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.427603753201617e-05,
+      "loss": 0.7684,
+      "step": 7166
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.426265890690609e-05,
+      "loss": 0.7417,
+      "step": 7167
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.424928028179602e-05,
+      "loss": 0.8027,
+      "step": 7168
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4235901656685946e-05,
+      "loss": 0.9217,
+      "step": 7169
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4222523031575866e-05,
+      "loss": 0.7261,
+      "step": 7170
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4209144406465786e-05,
+      "loss": 0.7201,
+      "step": 7171
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.419576578135571e-05,
+      "loss": 0.5711,
+      "step": 7172
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.418238715624564e-05,
+      "loss": 0.7022,
+      "step": 7173
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.416900853113556e-05,
+      "loss": 0.6937,
+      "step": 7174
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.415562990602548e-05,
+      "loss": 0.684,
+      "step": 7175
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.414225128091541e-05,
+      "loss": 0.68,
+      "step": 7176
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.4128872655805333e-05,
+      "loss": 0.4891,
+      "step": 7177
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.411549403069526e-05,
+      "loss": 0.5751,
+      "step": 7178
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.4102115405585173e-05,
+      "loss": 0.534,
+      "step": 7179
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.40887367804751e-05,
+      "loss": 0.5469,
+      "step": 7180
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.407535815536503e-05,
+      "loss": 1.5253,
+      "step": 7181
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.4061979530254954e-05,
+      "loss": 1.4668,
+      "step": 7182
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.404860090514487e-05,
+      "loss": 1.1464,
+      "step": 7183
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.4035222280034794e-05,
+      "loss": 1.3859,
+      "step": 7184
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.402184365492472e-05,
+      "loss": 1.1203,
+      "step": 7185
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.400846502981464e-05,
+      "loss": 0.9608,
+      "step": 7186
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.399508640470456e-05,
+      "loss": 1.2779,
+      "step": 7187
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.398170777959449e-05,
+      "loss": 1.1334,
+      "step": 7188
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3968329154484414e-05,
+      "loss": 1.0324,
+      "step": 7189
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3954950529374334e-05,
+      "loss": 0.9767,
+      "step": 7190
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3941571904264254e-05,
+      "loss": 1.2203,
+      "step": 7191
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.392819327915418e-05,
+      "loss": 1.2785,
+      "step": 7192
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.391481465404411e-05,
+      "loss": 1.1675,
+      "step": 7193
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.390143602893403e-05,
+      "loss": 1.0019,
+      "step": 7194
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3888057403823954e-05,
+      "loss": 1.0468,
+      "step": 7195
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3874678778713874e-05,
+      "loss": 1.1735,
+      "step": 7196
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.38613001536038e-05,
+      "loss": 1.089,
+      "step": 7197
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.384792152849372e-05,
+      "loss": 1.1346,
+      "step": 7198
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.383454290338365e-05,
+      "loss": 1.0765,
+      "step": 7199
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.382116427827357e-05,
+      "loss": 0.9738,
+      "step": 7200
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3807785653163495e-05,
+      "loss": 1.0065,
+      "step": 7201
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3794407028053415e-05,
+      "loss": 1.0103,
+      "step": 7202
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.378102840294334e-05,
+      "loss": 1.0963,
+      "step": 7203
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.376764977783326e-05,
+      "loss": 0.8637,
+      "step": 7204
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.375427115272319e-05,
+      "loss": 0.945,
+      "step": 7205
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.374089252761311e-05,
+      "loss": 0.851,
+      "step": 7206
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3727513902503035e-05,
+      "loss": 0.8702,
+      "step": 7207
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3714135277392955e-05,
+      "loss": 0.9941,
+      "step": 7208
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.370075665228288e-05,
+      "loss": 0.8277,
+      "step": 7209
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.36873780271728e-05,
+      "loss": 0.91,
+      "step": 7210
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.367399940206273e-05,
+      "loss": 0.7882,
+      "step": 7211
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.3660620776952655e-05,
+      "loss": 0.9294,
+      "step": 7212
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.364724215184257e-05,
+      "loss": 0.9312,
+      "step": 7213
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.3633863526732495e-05,
+      "loss": 0.7856,
+      "step": 7214
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.362048490162242e-05,
+      "loss": 0.8815,
+      "step": 7215
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.360710627651235e-05,
+      "loss": 0.7824,
+      "step": 7216
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.359372765140226e-05,
+      "loss": 0.766,
+      "step": 7217
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.358034902629219e-05,
+      "loss": 0.9123,
+      "step": 7218
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.3566970401182116e-05,
+      "loss": 0.7742,
+      "step": 7219
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.355359177607204e-05,
+      "loss": 0.7233,
+      "step": 7220
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.3540213150961956e-05,
+      "loss": 0.7657,
+      "step": 7221
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.352683452585188e-05,
+      "loss": 0.6762,
+      "step": 7222
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.351345590074181e-05,
+      "loss": 0.7069,
+      "step": 7223
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.3500077275631736e-05,
+      "loss": 0.6905,
+      "step": 7224
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.348669865052165e-05,
+      "loss": 0.5143,
+      "step": 7225
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.3473320025411576e-05,
+      "loss": 0.5823,
+      "step": 7226
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.34599414003015e-05,
+      "loss": 0.5753,
+      "step": 7227
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.344656277519143e-05,
+      "loss": 0.5108,
+      "step": 7228
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.343318415008135e-05,
+      "loss": 0.7634,
+      "step": 7229
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.341980552497127e-05,
+      "loss": 0.6829,
+      "step": 7230
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.3406426899861196e-05,
+      "loss": 1.6003,
+      "step": 7231
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.339304827475112e-05,
+      "loss": 1.4841,
+      "step": 7232
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.337966964964104e-05,
+      "loss": 1.3437,
+      "step": 7233
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.336629102453096e-05,
+      "loss": 1.2251,
+      "step": 7234
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.335291239942089e-05,
+      "loss": 1.1117,
+      "step": 7235
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.333953377431082e-05,
+      "loss": 1.2354,
+      "step": 7236
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.332615514920074e-05,
+      "loss": 1.0207,
+      "step": 7237
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.331277652409066e-05,
+      "loss": 1.1083,
+      "step": 7238
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.3299397898980584e-05,
+      "loss": 1.0949,
+      "step": 7239
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.3286019273870504e-05,
+      "loss": 1.0703,
+      "step": 7240
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.327264064876043e-05,
+      "loss": 0.9654,
+      "step": 7241
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.325926202365035e-05,
+      "loss": 1.1197,
+      "step": 7242
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.324588339854028e-05,
+      "loss": 0.9241,
+      "step": 7243
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.32325047734302e-05,
+      "loss": 1.136,
+      "step": 7244
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.3219126148320124e-05,
+      "loss": 1.1186,
+      "step": 7245
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.320574752321005e-05,
+      "loss": 1.24,
+      "step": 7246
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.319236889809997e-05,
+      "loss": 1.0669,
+      "step": 7247
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.317899027298989e-05,
+      "loss": 1.0433,
+      "step": 7248
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.316561164787982e-05,
+      "loss": 1.0698,
+      "step": 7249
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.3152233022769744e-05,
+      "loss": 1.01,
+      "step": 7250
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.3138854397659664e-05,
+      "loss": 0.9602,
+      "step": 7251
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.3125475772549584e-05,
+      "loss": 0.9729,
+      "step": 7252
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.311209714743951e-05,
+      "loss": 1.1448,
+      "step": 7253
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.309871852232944e-05,
+      "loss": 1.0443,
+      "step": 7254
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.308533989721936e-05,
+      "loss": 0.957,
+      "step": 7255
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.307196127210928e-05,
+      "loss": 0.9367,
+      "step": 7256
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.3058582646999205e-05,
+      "loss": 1.0076,
+      "step": 7257
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.304520402188913e-05,
+      "loss": 0.8006,
+      "step": 7258
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.303182539677905e-05,
+      "loss": 0.8299,
+      "step": 7259
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.301844677166897e-05,
+      "loss": 0.827,
+      "step": 7260
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.30050681465589e-05,
+      "loss": 0.9103,
+      "step": 7261
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2991689521448825e-05,
+      "loss": 0.7603,
+      "step": 7262
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.297831089633875e-05,
+      "loss": 0.8501,
+      "step": 7263
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2964932271228665e-05,
+      "loss": 0.8005,
+      "step": 7264
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.295155364611859e-05,
+      "loss": 0.9057,
+      "step": 7265
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.293817502100852e-05,
+      "loss": 0.8062,
+      "step": 7266
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.292479639589844e-05,
+      "loss": 0.8895,
+      "step": 7267
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.291141777078836e-05,
+      "loss": 0.7231,
+      "step": 7268
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2898039145678285e-05,
+      "loss": 0.6525,
+      "step": 7269
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.288466052056821e-05,
+      "loss": 0.7908,
+      "step": 7270
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.287128189545813e-05,
+      "loss": 0.6068,
+      "step": 7271
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.285790327034805e-05,
+      "loss": 0.6937,
+      "step": 7272
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.284452464523798e-05,
+      "loss": 0.816,
+      "step": 7273
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.2831146020127906e-05,
+      "loss": 0.6412,
+      "step": 7274
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.2817767395017826e-05,
+      "loss": 0.5583,
+      "step": 7275
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.280438876990775e-05,
+      "loss": 0.7396,
+      "step": 7276
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.279101014479767e-05,
+      "loss": 0.5807,
+      "step": 7277
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.27776315196876e-05,
+      "loss": 0.5504,
+      "step": 7278
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.276425289457752e-05,
+      "loss": 0.5858,
+      "step": 7279
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.2750874269467446e-05,
+      "loss": 0.7067,
+      "step": 7280
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.2737495644357366e-05,
+      "loss": 1.5355,
+      "step": 7281
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.272411701924729e-05,
+      "loss": 1.2614,
+      "step": 7282
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.271073839413721e-05,
+      "loss": 1.2448,
+      "step": 7283
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.269735976902714e-05,
+      "loss": 1.2374,
+      "step": 7284
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.268398114391706e-05,
+      "loss": 1.1022,
+      "step": 7285
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.2670602518806986e-05,
+      "loss": 1.2299,
+      "step": 7286
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.2657223893696906e-05,
+      "loss": 1.1754,
+      "step": 7287
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.264384526858683e-05,
+      "loss": 1.0975,
+      "step": 7288
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.263046664347675e-05,
+      "loss": 1.087,
+      "step": 7289
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.261708801836668e-05,
+      "loss": 1.091,
+      "step": 7290
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.26037093932566e-05,
+      "loss": 1.1086,
+      "step": 7291
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.259033076814653e-05,
+      "loss": 1.0412,
+      "step": 7292
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.2576952143036453e-05,
+      "loss": 1.0844,
+      "step": 7293
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.256357351792637e-05,
+      "loss": 1.0389,
+      "step": 7294
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.2550194892816293e-05,
+      "loss": 1.1818,
+      "step": 7295
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.253681626770622e-05,
+      "loss": 0.8904,
+      "step": 7296
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.252343764259615e-05,
+      "loss": 1.0233,
+      "step": 7297
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.251005901748606e-05,
+      "loss": 1.2132,
+      "step": 7298
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.249668039237599e-05,
+      "loss": 0.9953,
+      "step": 7299
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.2483301767265914e-05,
+      "loss": 1.0126,
+      "step": 7300
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.246992314215584e-05,
+      "loss": 1.0262,
+      "step": 7301
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.2456544517045754e-05,
+      "loss": 0.9086,
+      "step": 7302
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.244316589193568e-05,
+      "loss": 1.1498,
+      "step": 7303
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.242978726682561e-05,
+      "loss": 0.8846,
+      "step": 7304
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.2416408641715534e-05,
+      "loss": 0.9127,
+      "step": 7305
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.240303001660545e-05,
+      "loss": 0.9585,
+      "step": 7306
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.2389651391495374e-05,
+      "loss": 0.943,
+      "step": 7307
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.23762727663853e-05,
+      "loss": 0.8588,
+      "step": 7308
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.236289414127523e-05,
+      "loss": 0.8455,
+      "step": 7309
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.234951551616515e-05,
+      "loss": 0.842,
+      "step": 7310
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.233613689105507e-05,
+      "loss": 0.8283,
+      "step": 7311
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.2322758265944994e-05,
+      "loss": 0.8875,
+      "step": 7312
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.230937964083492e-05,
+      "loss": 1.0453,
+      "step": 7313
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.229600101572484e-05,
+      "loss": 0.8668,
+      "step": 7314
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.228262239061476e-05,
+      "loss": 0.6723,
+      "step": 7315
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.226924376550469e-05,
+      "loss": 0.9045,
+      "step": 7316
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.2255865140394615e-05,
+      "loss": 0.8514,
+      "step": 7317
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.2242486515284535e-05,
+      "loss": 0.6899,
+      "step": 7318
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.2229107890174455e-05,
+      "loss": 0.6587,
+      "step": 7319
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.221572926506438e-05,
+      "loss": 0.9463,
+      "step": 7320
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.22023506399543e-05,
+      "loss": 0.7282,
+      "step": 7321
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.218897201484423e-05,
+      "loss": 0.6357,
+      "step": 7322
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.217559338973415e-05,
+      "loss": 0.6447,
+      "step": 7323
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.2162214764624075e-05,
+      "loss": 0.6387,
+      "step": 7324
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.2148836139513995e-05,
+      "loss": 0.7311,
+      "step": 7325
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.213545751440392e-05,
+      "loss": 0.5005,
+      "step": 7326
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.212207888929385e-05,
+      "loss": 0.6997,
+      "step": 7327
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.210870026418377e-05,
+      "loss": 0.6372,
+      "step": 7328
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.209532163907369e-05,
+      "loss": 0.6484,
+      "step": 7329
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.2081943013963616e-05,
+      "loss": 0.6354,
+      "step": 7330
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.206856438885354e-05,
+      "loss": 1.6424,
+      "step": 7331
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.205518576374346e-05,
+      "loss": 1.2789,
+      "step": 7332
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.204180713863338e-05,
+      "loss": 1.3611,
+      "step": 7333
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.202842851352331e-05,
+      "loss": 1.3258,
+      "step": 7334
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.2015049888413236e-05,
+      "loss": 1.2889,
+      "step": 7335
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.2001671263303156e-05,
+      "loss": 1.1956,
+      "step": 7336
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.1988292638193076e-05,
+      "loss": 1.182,
+      "step": 7337
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.1974914013083e-05,
+      "loss": 1.3918,
+      "step": 7338
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.196153538797293e-05,
+      "loss": 1.1088,
+      "step": 7339
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.194815676286285e-05,
+      "loss": 1.2491,
+      "step": 7340
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.193477813775277e-05,
+      "loss": 1.0911,
+      "step": 7341
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.1921399512642696e-05,
+      "loss": 0.9541,
+      "step": 7342
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.190802088753262e-05,
+      "loss": 1.1905,
+      "step": 7343
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.189464226242255e-05,
+      "loss": 1.0657,
+      "step": 7344
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.188126363731246e-05,
+      "loss": 1.0563,
+      "step": 7345
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.186788501220239e-05,
+      "loss": 1.2515,
+      "step": 7346
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.1854506387092317e-05,
+      "loss": 1.1269,
+      "step": 7347
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.1841127761982237e-05,
+      "loss": 1.0261,
+      "step": 7348
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.1827749136872157e-05,
+      "loss": 0.9306,
+      "step": 7349
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.181437051176208e-05,
+      "loss": 1.056,
+      "step": 7350
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.180099188665201e-05,
+      "loss": 0.9342,
+      "step": 7351
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.178761326154193e-05,
+      "loss": 1.0635,
+      "step": 7352
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.177423463643185e-05,
+      "loss": 1.0594,
+      "step": 7353
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.176085601132178e-05,
+      "loss": 0.9874,
+      "step": 7354
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.1747477386211704e-05,
+      "loss": 1.0434,
+      "step": 7355
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.1734098761101624e-05,
+      "loss": 0.9303,
+      "step": 7356
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.1720720135991544e-05,
+      "loss": 0.9966,
+      "step": 7357
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.170734151088147e-05,
+      "loss": 1.0176,
+      "step": 7358
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.16939628857714e-05,
+      "loss": 0.9793,
+      "step": 7359
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.168058426066132e-05,
+      "loss": 0.896,
+      "step": 7360
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.1667205635551244e-05,
+      "loss": 0.9214,
+      "step": 7361
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.1653827010441164e-05,
+      "loss": 0.8541,
+      "step": 7362
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.164044838533109e-05,
+      "loss": 0.8153,
+      "step": 7363
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.162706976022101e-05,
+      "loss": 0.9418,
+      "step": 7364
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.161369113511094e-05,
+      "loss": 0.8177,
+      "step": 7365
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.160031251000086e-05,
+      "loss": 0.7162,
+      "step": 7366
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.1586933884890784e-05,
+      "loss": 0.9109,
+      "step": 7367
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.1573555259780704e-05,
+      "loss": 0.7704,
+      "step": 7368
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.156017663467063e-05,
+      "loss": 0.7799,
+      "step": 7369
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.154679800956055e-05,
+      "loss": 0.747,
+      "step": 7370
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.153341938445047e-05,
+      "loss": 0.8043,
+      "step": 7371
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.15200407593404e-05,
+      "loss": 0.6429,
+      "step": 7372
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.1506662134230325e-05,
+      "loss": 0.694,
+      "step": 7373
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.149328350912025e-05,
+      "loss": 0.569,
+      "step": 7374
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.1479904884010165e-05,
+      "loss": 0.6241,
+      "step": 7375
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.146652625890009e-05,
+      "loss": 0.6007,
+      "step": 7376
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.145314763379002e-05,
+      "loss": 0.5008,
+      "step": 7377
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.1439769008679945e-05,
+      "loss": 0.5627,
+      "step": 7378
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.142639038356986e-05,
+      "loss": 0.5485,
+      "step": 7379
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.1413011758459785e-05,
+      "loss": 0.5782,
+      "step": 7380
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.139963313334971e-05,
+      "loss": 1.5979,
+      "step": 7381
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.138625450823964e-05,
+      "loss": 1.2915,
+      "step": 7382
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.137287588312955e-05,
+      "loss": 1.4301,
+      "step": 7383
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.135949725801948e-05,
+      "loss": 1.4248,
+      "step": 7384
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.1346118632909405e-05,
+      "loss": 1.3432,
+      "step": 7385
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.133274000779933e-05,
+      "loss": 1.1563,
+      "step": 7386
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.1319361382689245e-05,
+      "loss": 1.1354,
+      "step": 7387
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.130598275757917e-05,
+      "loss": 1.1936,
+      "step": 7388
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.12926041324691e-05,
+      "loss": 1.1016,
+      "step": 7389
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.1279225507359026e-05,
+      "loss": 1.0891,
+      "step": 7390
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.1265846882248946e-05,
+      "loss": 1.0934,
+      "step": 7391
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.1252468257138866e-05,
+      "loss": 1.1101,
+      "step": 7392
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.123908963202879e-05,
+      "loss": 1.0429,
+      "step": 7393
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.122571100691872e-05,
+      "loss": 1.055,
+      "step": 7394
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.121233238180864e-05,
+      "loss": 1.1558,
+      "step": 7395
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.119895375669856e-05,
+      "loss": 1.0936,
+      "step": 7396
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.1185575131588486e-05,
+      "loss": 1.0597,
+      "step": 7397
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.1172196506478406e-05,
+      "loss": 1.1176,
+      "step": 7398
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.115881788136833e-05,
+      "loss": 1.0384,
+      "step": 7399
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.114543925625825e-05,
+      "loss": 1.1162,
+      "step": 7400
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.113206063114818e-05,
+      "loss": 0.9358,
+      "step": 7401
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.11186820060381e-05,
+      "loss": 0.9558,
+      "step": 7402
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.1105303380928026e-05,
+      "loss": 1.0582,
+      "step": 7403
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.1091924755817946e-05,
+      "loss": 0.9398,
+      "step": 7404
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.107854613070787e-05,
+      "loss": 0.8982,
+      "step": 7405
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.106516750559779e-05,
+      "loss": 1.0058,
+      "step": 7406
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.105178888048772e-05,
+      "loss": 0.9021,
+      "step": 7407
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.103841025537765e-05,
+      "loss": 0.8983,
+      "step": 7408
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.102503163026757e-05,
+      "loss": 0.7809,
+      "step": 7409
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.101165300515749e-05,
+      "loss": 0.7282,
+      "step": 7410
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.0998274380047413e-05,
+      "loss": 0.8027,
+      "step": 7411
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.098489575493734e-05,
+      "loss": 0.931,
+      "step": 7412
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.097151712982726e-05,
+      "loss": 0.6744,
+      "step": 7413
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.095813850471718e-05,
+      "loss": 0.844,
+      "step": 7414
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.094475987960711e-05,
+      "loss": 0.8916,
+      "step": 7415
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.0931381254497034e-05,
+      "loss": 0.7385,
+      "step": 7416
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.0918002629386954e-05,
+      "loss": 0.7153,
+      "step": 7417
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.0904624004276874e-05,
+      "loss": 0.6202,
+      "step": 7418
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.08912453791668e-05,
+      "loss": 0.8125,
+      "step": 7419
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.087786675405673e-05,
+      "loss": 0.7671,
+      "step": 7420
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.086448812894665e-05,
+      "loss": 0.8818,
+      "step": 7421
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.085110950383657e-05,
+      "loss": 0.7371,
+      "step": 7422
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.0837730878726494e-05,
+      "loss": 0.6818,
+      "step": 7423
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.082435225361642e-05,
+      "loss": 0.6716,
+      "step": 7424
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.081097362850634e-05,
+      "loss": 0.6954,
+      "step": 7425
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.079759500339627e-05,
+      "loss": 0.4466,
+      "step": 7426
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.078421637828619e-05,
+      "loss": 0.5327,
+      "step": 7427
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.0770837753176115e-05,
+      "loss": 0.4949,
+      "step": 7428
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.0757459128066035e-05,
+      "loss": 0.5583,
+      "step": 7429
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.074408050295596e-05,
+      "loss": 0.5285,
+      "step": 7430
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.073070187784588e-05,
+      "loss": 1.5354,
+      "step": 7431
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.071732325273581e-05,
+      "loss": 1.4137,
+      "step": 7432
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.070394462762573e-05,
+      "loss": 1.3854,
+      "step": 7433
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.0690566002515655e-05,
+      "loss": 1.1383,
+      "step": 7434
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.0677187377405575e-05,
+      "loss": 1.2819,
+      "step": 7435
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.06638087522955e-05,
+      "loss": 1.148,
+      "step": 7436
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.065043012718542e-05,
+      "loss": 1.0959,
+      "step": 7437
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.063705150207535e-05,
+      "loss": 1.1048,
+      "step": 7438
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.062367287696527e-05,
+      "loss": 1.0831,
+      "step": 7439
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.0610294251855195e-05,
+      "loss": 1.061,
+      "step": 7440
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.0596915626745115e-05,
+      "loss": 0.9995,
+      "step": 7441
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.058353700163504e-05,
+      "loss": 1.1688,
+      "step": 7442
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.057015837652496e-05,
+      "loss": 1.0333,
+      "step": 7443
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.055677975141489e-05,
+      "loss": 1.1007,
+      "step": 7444
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.054340112630481e-05,
+      "loss": 1.1386,
+      "step": 7445
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.0530022501194736e-05,
+      "loss": 1.1587,
+      "step": 7446
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.051664387608466e-05,
+      "loss": 1.0186,
+      "step": 7447
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.050326525097458e-05,
+      "loss": 0.9138,
+      "step": 7448
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.0489886625864506e-05,
+      "loss": 0.9896,
+      "step": 7449
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.0476508000754426e-05,
+      "loss": 1.0269,
+      "step": 7450
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.046312937564435e-05,
+      "loss": 1.1056,
+      "step": 7451
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.0449750750534272e-05,
+      "loss": 1.031,
+      "step": 7452
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.0436372125424196e-05,
+      "loss": 0.9592,
+      "step": 7453
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.042299350031412e-05,
+      "loss": 1.0629,
+      "step": 7454
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.0409614875204043e-05,
+      "loss": 0.9698,
+      "step": 7455
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.039623625009397e-05,
+      "loss": 0.9824,
+      "step": 7456
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.038285762498389e-05,
+      "loss": 0.988,
+      "step": 7457
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.0369478999873816e-05,
+      "loss": 1.0302,
+      "step": 7458
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.0356100374763736e-05,
+      "loss": 0.9399,
+      "step": 7459
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.0342721749653663e-05,
+      "loss": 0.9241,
+      "step": 7460
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.0329343124543583e-05,
+      "loss": 0.8985,
+      "step": 7461
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.031596449943351e-05,
+      "loss": 0.9222,
+      "step": 7462
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.030258587432343e-05,
+      "loss": 0.9267,
+      "step": 7463
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.0289207249213357e-05,
+      "loss": 0.8118,
+      "step": 7464
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.0275828624103277e-05,
+      "loss": 0.8582,
+      "step": 7465
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.0262449998993203e-05,
+      "loss": 0.8774,
+      "step": 7466
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.0249071373883123e-05,
+      "loss": 0.7429,
+      "step": 7467
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.023569274877305e-05,
+      "loss": 0.737,
+      "step": 7468
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.022231412366297e-05,
+      "loss": 0.676,
+      "step": 7469
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.0208935498552897e-05,
+      "loss": 0.8011,
+      "step": 7470
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.0195556873442817e-05,
+      "loss": 0.8063,
+      "step": 7471
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.0182178248332744e-05,
+      "loss": 0.6745,
+      "step": 7472
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.0168799623222667e-05,
+      "loss": 0.6775,
+      "step": 7473
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.015542099811259e-05,
+      "loss": 0.6183,
+      "step": 7474
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.0142042373002514e-05,
+      "loss": 0.6968,
+      "step": 7475
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.0128663747892434e-05,
+      "loss": 0.523,
+      "step": 7476
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.011528512278236e-05,
+      "loss": 0.6124,
+      "step": 7477
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.010190649767228e-05,
+      "loss": 0.6889,
+      "step": 7478
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.0088527872562207e-05,
+      "loss": 0.4515,
+      "step": 7479
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.0075149247452127e-05,
+      "loss": 0.5617,
+      "step": 7480
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.0061770622342054e-05,
+      "loss": 1.6947,
+      "step": 7481
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.0048391997231974e-05,
+      "loss": 1.3469,
+      "step": 7482
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.00350133721219e-05,
+      "loss": 1.2267,
+      "step": 7483
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.002163474701182e-05,
+      "loss": 1.3494,
+      "step": 7484
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.0008256121901748e-05,
+      "loss": 1.0759,
+      "step": 7485
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9994877496791668e-05,
+      "loss": 1.2372,
+      "step": 7486
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9981498871681595e-05,
+      "loss": 1.1086,
+      "step": 7487
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9968120246571515e-05,
+      "loss": 1.1675,
+      "step": 7488
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.995474162146144e-05,
+      "loss": 1.2022,
+      "step": 7489
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9941362996351365e-05,
+      "loss": 0.9195,
+      "step": 7490
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9927984371241288e-05,
+      "loss": 1.0919,
+      "step": 7491
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.991460574613121e-05,
+      "loss": 1.1655,
+      "step": 7492
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9901227121021135e-05,
+      "loss": 1.2152,
+      "step": 7493
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9887848495911058e-05,
+      "loss": 1.0089,
+      "step": 7494
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.987446987080098e-05,
+      "loss": 1.0969,
+      "step": 7495
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9861091245690905e-05,
+      "loss": 1.0794,
+      "step": 7496
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.984771262058083e-05,
+      "loss": 1.1413,
+      "step": 7497
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9834333995470752e-05,
+      "loss": 1.1355,
+      "step": 7498
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9820955370360675e-05,
+      "loss": 1.1267,
+      "step": 7499
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.98075767452506e-05,
+      "loss": 0.9147,
+      "step": 7500
+    },
+    {
+      "epoch": 2.31,
+      "eval_bleu": 5.2261887618934477e-14,
+      "eval_loss": 1.7244117259979248,
+      "eval_runtime": 2554.8731,
+      "eval_samples_per_second": 5.777,
+      "eval_steps_per_second": 0.722,
+      "step": 7500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9794198120140522e-05,
+      "loss": 1.2907,
+      "step": 7501
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9780819495030445e-05,
+      "loss": 1.002,
+      "step": 7502
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9767440869920365e-05,
+      "loss": 1.0535,
+      "step": 7503
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9754062244810292e-05,
+      "loss": 0.8953,
+      "step": 7504
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9740683619700216e-05,
+      "loss": 0.9731,
+      "step": 7505
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.972730499459014e-05,
+      "loss": 0.8797,
+      "step": 7506
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9713926369480062e-05,
+      "loss": 0.9275,
+      "step": 7507
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9700547744369986e-05,
+      "loss": 0.9941,
+      "step": 7508
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.968716911925991e-05,
+      "loss": 0.8833,
+      "step": 7509
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9673790494149833e-05,
+      "loss": 1.0558,
+      "step": 7510
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9660411869039756e-05,
+      "loss": 1.0534,
+      "step": 7511
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.964703324392968e-05,
+      "loss": 0.7368,
+      "step": 7512
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9633654618819603e-05,
+      "loss": 0.804,
+      "step": 7513
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9620275993709526e-05,
+      "loss": 0.8007,
+      "step": 7514
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.960689736859945e-05,
+      "loss": 0.7322,
+      "step": 7515
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9593518743489373e-05,
+      "loss": 0.7477,
+      "step": 7516
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9580140118379296e-05,
+      "loss": 0.9016,
+      "step": 7517
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.956676149326922e-05,
+      "loss": 0.7715,
+      "step": 7518
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9553382868159143e-05,
+      "loss": 0.812,
+      "step": 7519
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9540004243049066e-05,
+      "loss": 0.6579,
+      "step": 7520
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.952662561793899e-05,
+      "loss": 0.5786,
+      "step": 7521
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9513246992828917e-05,
+      "loss": 0.9182,
+      "step": 7522
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9499868367718837e-05,
+      "loss": 0.7538,
+      "step": 7523
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9486489742608763e-05,
+      "loss": 0.6593,
+      "step": 7524
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9473111117498683e-05,
+      "loss": 0.6845,
+      "step": 7525
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.945973249238861e-05,
+      "loss": 0.4929,
+      "step": 7526
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.944635386727853e-05,
+      "loss": 0.5503,
+      "step": 7527
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9432975242168457e-05,
+      "loss": 0.5079,
+      "step": 7528
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9419596617058377e-05,
+      "loss": 0.5205,
+      "step": 7529
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.94062179919483e-05,
+      "loss": 0.6077,
+      "step": 7530
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9392839366838224e-05,
+      "loss": 1.486,
+      "step": 7531
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9379460741728147e-05,
+      "loss": 1.5466,
+      "step": 7532
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.936608211661807e-05,
+      "loss": 1.3668,
+      "step": 7533
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.9352703491507994e-05,
+      "loss": 1.1895,
+      "step": 7534
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.9339324866397917e-05,
+      "loss": 1.1544,
+      "step": 7535
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.932594624128784e-05,
+      "loss": 1.1052,
+      "step": 7536
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.9312567616177764e-05,
+      "loss": 1.1199,
+      "step": 7537
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.9299188991067687e-05,
+      "loss": 0.9428,
+      "step": 7538
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.9285810365957614e-05,
+      "loss": 1.085,
+      "step": 7539
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.9272431740847534e-05,
+      "loss": 1.0642,
+      "step": 7540
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.925905311573746e-05,
+      "loss": 1.096,
+      "step": 7541
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.924567449062738e-05,
+      "loss": 1.3067,
+      "step": 7542
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.9232295865517308e-05,
+      "loss": 1.1276,
+      "step": 7543
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.9218917240407228e-05,
+      "loss": 1.0257,
+      "step": 7544
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.9205538615297155e-05,
+      "loss": 1.0209,
+      "step": 7545
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.9192159990187075e-05,
+      "loss": 0.9088,
+      "step": 7546
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.9178781365077e-05,
+      "loss": 1.0483,
+      "step": 7547
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.916540273996692e-05,
+      "loss": 1.0957,
+      "step": 7548
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.9152024114856848e-05,
+      "loss": 1.0841,
+      "step": 7549
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.9138645489746768e-05,
+      "loss": 1.0801,
+      "step": 7550
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.9125266864636695e-05,
+      "loss": 0.9215,
+      "step": 7551
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.9111888239526615e-05,
+      "loss": 0.9273,
+      "step": 7552
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.909850961441654e-05,
+      "loss": 1.0548,
+      "step": 7553
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.9085130989306465e-05,
+      "loss": 1.0527,
+      "step": 7554
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.907175236419639e-05,
+      "loss": 0.9554,
+      "step": 7555
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.9058373739086312e-05,
+      "loss": 0.8764,
+      "step": 7556
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.9044995113976232e-05,
+      "loss": 0.8328,
+      "step": 7557
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.903161648886616e-05,
+      "loss": 0.8976,
+      "step": 7558
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.901823786375608e-05,
+      "loss": 1.0308,
+      "step": 7559
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.9004859238646005e-05,
+      "loss": 0.8525,
+      "step": 7560
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8991480613535925e-05,
+      "loss": 0.784,
+      "step": 7561
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8978101988425852e-05,
+      "loss": 0.8618,
+      "step": 7562
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8964723363315772e-05,
+      "loss": 0.9452,
+      "step": 7563
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.89513447382057e-05,
+      "loss": 0.8139,
+      "step": 7564
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.893796611309562e-05,
+      "loss": 0.8368,
+      "step": 7565
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8924587487985546e-05,
+      "loss": 0.7915,
+      "step": 7566
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8911208862875466e-05,
+      "loss": 0.7217,
+      "step": 7567
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8897830237765393e-05,
+      "loss": 0.7938,
+      "step": 7568
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8884451612655313e-05,
+      "loss": 0.5874,
+      "step": 7569
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.887107298754524e-05,
+      "loss": 0.5933,
+      "step": 7570
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8857694362435163e-05,
+      "loss": 0.6565,
+      "step": 7571
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8844315737325086e-05,
+      "loss": 0.6992,
+      "step": 7572
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.883093711221501e-05,
+      "loss": 0.646,
+      "step": 7573
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8817558487104933e-05,
+      "loss": 0.6372,
+      "step": 7574
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8804179861994856e-05,
+      "loss": 0.7293,
+      "step": 7575
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.879080123688478e-05,
+      "loss": 0.5484,
+      "step": 7576
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8777422611774703e-05,
+      "loss": 0.6706,
+      "step": 7577
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8764043986664626e-05,
+      "loss": 0.6173,
+      "step": 7578
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.875066536155455e-05,
+      "loss": 0.6646,
+      "step": 7579
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8737286736444473e-05,
+      "loss": 0.7451,
+      "step": 7580
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8723908111334397e-05,
+      "loss": 1.6387,
+      "step": 7581
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8710529486224317e-05,
+      "loss": 1.2844,
+      "step": 7582
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8697150861114243e-05,
+      "loss": 1.4149,
+      "step": 7583
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8683772236004163e-05,
+      "loss": 1.1931,
+      "step": 7584
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.867039361089409e-05,
+      "loss": 1.1686,
+      "step": 7585
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.865701498578401e-05,
+      "loss": 1.1636,
+      "step": 7586
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8643636360673937e-05,
+      "loss": 1.2647,
+      "step": 7587
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.863025773556386e-05,
+      "loss": 1.0841,
+      "step": 7588
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8616879110453784e-05,
+      "loss": 1.1169,
+      "step": 7589
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8603500485343707e-05,
+      "loss": 1.0499,
+      "step": 7590
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.859012186023363e-05,
+      "loss": 1.038,
+      "step": 7591
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8576743235123554e-05,
+      "loss": 1.162,
+      "step": 7592
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8563364610013477e-05,
+      "loss": 1.0421,
+      "step": 7593
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.85499859849034e-05,
+      "loss": 0.9354,
+      "step": 7594
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8536607359793324e-05,
+      "loss": 1.0632,
+      "step": 7595
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8523228734683247e-05,
+      "loss": 0.9699,
+      "step": 7596
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.850985010957317e-05,
+      "loss": 1.0448,
+      "step": 7597
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8496471484463094e-05,
+      "loss": 1.0379,
+      "step": 7598
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8483092859353018e-05,
+      "loss": 0.9325,
+      "step": 7599
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.846971423424294e-05,
+      "loss": 0.9811,
+      "step": 7600
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8456335609132864e-05,
+      "loss": 1.0306,
+      "step": 7601
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8442956984022788e-05,
+      "loss": 0.8636,
+      "step": 7602
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8429578358912715e-05,
+      "loss": 0.9509,
+      "step": 7603
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8416199733802635e-05,
+      "loss": 1.0998,
+      "step": 7604
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.840282110869256e-05,
+      "loss": 0.8301,
+      "step": 7605
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.838944248358248e-05,
+      "loss": 0.9948,
+      "step": 7606
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8376063858472408e-05,
+      "loss": 0.7652,
+      "step": 7607
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8362685233362328e-05,
+      "loss": 0.9701,
+      "step": 7608
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.834930660825225e-05,
+      "loss": 0.8548,
+      "step": 7609
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8335927983142175e-05,
+      "loss": 0.8062,
+      "step": 7610
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8322549358032098e-05,
+      "loss": 0.8652,
+      "step": 7611
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8309170732922022e-05,
+      "loss": 0.7978,
+      "step": 7612
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8295792107811945e-05,
+      "loss": 0.8252,
+      "step": 7613
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.828241348270187e-05,
+      "loss": 0.6636,
+      "step": 7614
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8269034857591792e-05,
+      "loss": 0.855,
+      "step": 7615
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8255656232481715e-05,
+      "loss": 0.6808,
+      "step": 7616
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.824227760737164e-05,
+      "loss": 0.7561,
+      "step": 7617
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8228898982261562e-05,
+      "loss": 0.6567,
+      "step": 7618
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8215520357151485e-05,
+      "loss": 0.6376,
+      "step": 7619
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8202141732041412e-05,
+      "loss": 0.8292,
+      "step": 7620
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8188763106931332e-05,
+      "loss": 0.6935,
+      "step": 7621
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.817538448182126e-05,
+      "loss": 0.6721,
+      "step": 7622
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.816200585671118e-05,
+      "loss": 0.7051,
+      "step": 7623
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8148627231601106e-05,
+      "loss": 0.7055,
+      "step": 7624
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8135248606491026e-05,
+      "loss": 0.6737,
+      "step": 7625
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8121869981380953e-05,
+      "loss": 0.6491,
+      "step": 7626
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8108491356270873e-05,
+      "loss": 0.6741,
+      "step": 7627
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.80951127311608e-05,
+      "loss": 0.597,
+      "step": 7628
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.808173410605072e-05,
+      "loss": 0.4568,
+      "step": 7629
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8068355480940646e-05,
+      "loss": 0.7466,
+      "step": 7630
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8054976855830566e-05,
+      "loss": 1.7187,
+      "step": 7631
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8041598230720493e-05,
+      "loss": 1.4436,
+      "step": 7632
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8028219605610413e-05,
+      "loss": 1.3369,
+      "step": 7633
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.801484098050034e-05,
+      "loss": 1.0548,
+      "step": 7634
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.800146235539026e-05,
+      "loss": 1.1923,
+      "step": 7635
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.7988083730280183e-05,
+      "loss": 1.1203,
+      "step": 7636
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.797470510517011e-05,
+      "loss": 1.2274,
+      "step": 7637
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.796132648006003e-05,
+      "loss": 1.2254,
+      "step": 7638
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.7947947854949957e-05,
+      "loss": 1.0924,
+      "step": 7639
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.7934569229839877e-05,
+      "loss": 1.0363,
+      "step": 7640
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.7921190604729803e-05,
+      "loss": 1.1682,
+      "step": 7641
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.7907811979619723e-05,
+      "loss": 1.0501,
+      "step": 7642
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.789443335450965e-05,
+      "loss": 1.1423,
+      "step": 7643
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.788105472939957e-05,
+      "loss": 1.092,
+      "step": 7644
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.7867676104289497e-05,
+      "loss": 1.0133,
+      "step": 7645
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.7854297479179417e-05,
+      "loss": 1.0865,
+      "step": 7646
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.7840918854069344e-05,
+      "loss": 1.188,
+      "step": 7647
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.7827540228959264e-05,
+      "loss": 1.111,
+      "step": 7648
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.781416160384919e-05,
+      "loss": 1.0234,
+      "step": 7649
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.780078297873911e-05,
+      "loss": 0.9939,
+      "step": 7650
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.7787404353629037e-05,
+      "loss": 0.927,
+      "step": 7651
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.777402572851896e-05,
+      "loss": 0.9824,
+      "step": 7652
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.7760647103408884e-05,
+      "loss": 1.1155,
+      "step": 7653
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.7747268478298807e-05,
+      "loss": 0.9877,
+      "step": 7654
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.773388985318873e-05,
+      "loss": 0.9346,
+      "step": 7655
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.7720511228078654e-05,
+      "loss": 0.9306,
+      "step": 7656
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.7707132602968578e-05,
+      "loss": 0.8775,
+      "step": 7657
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.76937539778585e-05,
+      "loss": 0.879,
+      "step": 7658
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.7680375352748424e-05,
+      "loss": 0.8126,
+      "step": 7659
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.7666996727638348e-05,
+      "loss": 0.9381,
+      "step": 7660
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.765361810252827e-05,
+      "loss": 0.765,
+      "step": 7661
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.7640239477418195e-05,
+      "loss": 0.8275,
+      "step": 7662
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.7626860852308115e-05,
+      "loss": 0.9309,
+      "step": 7663
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.761348222719804e-05,
+      "loss": 0.7101,
+      "step": 7664
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.760010360208796e-05,
+      "loss": 0.7006,
+      "step": 7665
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.7586724976977888e-05,
+      "loss": 0.8577,
+      "step": 7666
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.7573346351867808e-05,
+      "loss": 0.7352,
+      "step": 7667
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.7559967726757735e-05,
+      "loss": 0.6014,
+      "step": 7668
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.754658910164766e-05,
+      "loss": 0.8058,
+      "step": 7669
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.7533210476537582e-05,
+      "loss": 0.5755,
+      "step": 7670
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.7519831851427505e-05,
+      "loss": 0.6967,
+      "step": 7671
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.750645322631743e-05,
+      "loss": 0.629,
+      "step": 7672
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.7493074601207352e-05,
+      "loss": 0.797,
+      "step": 7673
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.7479695976097275e-05,
+      "loss": 0.6677,
+      "step": 7674
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.74663173509872e-05,
+      "loss": 0.5616,
+      "step": 7675
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.7452938725877122e-05,
+      "loss": 0.6214,
+      "step": 7676
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.7439560100767045e-05,
+      "loss": 0.5624,
+      "step": 7677
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.742618147565697e-05,
+      "loss": 0.5974,
+      "step": 7678
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.7412802850546892e-05,
+      "loss": 0.5463,
+      "step": 7679
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.7399424225436816e-05,
+      "loss": 0.5317,
+      "step": 7680
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.738604560032674e-05,
+      "loss": 1.3346,
+      "step": 7681
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.7372666975216662e-05,
+      "loss": 1.3361,
+      "step": 7682
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.7359288350106586e-05,
+      "loss": 1.4647,
+      "step": 7683
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.734590972499651e-05,
+      "loss": 1.2465,
+      "step": 7684
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.7332531099886433e-05,
+      "loss": 1.1563,
+      "step": 7685
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.731915247477636e-05,
+      "loss": 1.0807,
+      "step": 7686
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.730577384966628e-05,
+      "loss": 1.2475,
+      "step": 7687
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.7292395224556206e-05,
+      "loss": 1.088,
+      "step": 7688
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.7279016599446126e-05,
+      "loss": 1.2194,
+      "step": 7689
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.726563797433605e-05,
+      "loss": 1.0473,
+      "step": 7690
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.7252259349225973e-05,
+      "loss": 1.1001,
+      "step": 7691
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.7238880724115896e-05,
+      "loss": 1.0292,
+      "step": 7692
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.722550209900582e-05,
+      "loss": 1.0763,
+      "step": 7693
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.7212123473895743e-05,
+      "loss": 1.0181,
+      "step": 7694
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.7198744848785666e-05,
+      "loss": 1.1239,
+      "step": 7695
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.718536622367559e-05,
+      "loss": 1.0099,
+      "step": 7696
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.7171987598565513e-05,
+      "loss": 1.0761,
+      "step": 7697
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.7158608973455437e-05,
+      "loss": 0.9317,
+      "step": 7698
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.714523034834536e-05,
+      "loss": 1.2779,
+      "step": 7699
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.7131851723235283e-05,
+      "loss": 0.9899,
+      "step": 7700
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.711847309812521e-05,
+      "loss": 0.9655,
+      "step": 7701
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.710509447301513e-05,
+      "loss": 0.8076,
+      "step": 7702
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.7091715847905057e-05,
+      "loss": 0.8885,
+      "step": 7703
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.7078337222794977e-05,
+      "loss": 0.928,
+      "step": 7704
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.7064958597684904e-05,
+      "loss": 0.9703,
+      "step": 7705
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.7051579972574824e-05,
+      "loss": 1.155,
+      "step": 7706
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.703820134746475e-05,
+      "loss": 1.0822,
+      "step": 7707
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.702482272235467e-05,
+      "loss": 0.9079,
+      "step": 7708
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.7011444097244597e-05,
+      "loss": 0.9911,
+      "step": 7709
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.6998065472134517e-05,
+      "loss": 0.8254,
+      "step": 7710
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.6984686847024444e-05,
+      "loss": 0.723,
+      "step": 7711
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.6971308221914364e-05,
+      "loss": 0.9,
+      "step": 7712
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.695792959680429e-05,
+      "loss": 0.9476,
+      "step": 7713
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.694455097169421e-05,
+      "loss": 0.947,
+      "step": 7714
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.6931172346584134e-05,
+      "loss": 0.713,
+      "step": 7715
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.6917793721474058e-05,
+      "loss": 0.7513,
+      "step": 7716
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.690441509636398e-05,
+      "loss": 0.8317,
+      "step": 7717
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.6891036471253908e-05,
+      "loss": 0.9393,
+      "step": 7718
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.6877657846143828e-05,
+      "loss": 0.8278,
+      "step": 7719
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.6864279221033755e-05,
+      "loss": 0.6563,
+      "step": 7720
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.6850900595923675e-05,
+      "loss": 0.6862,
+      "step": 7721
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.68375219708136e-05,
+      "loss": 0.8649,
+      "step": 7722
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.682414334570352e-05,
+      "loss": 0.7226,
+      "step": 7723
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.6810764720593448e-05,
+      "loss": 0.5469,
+      "step": 7724
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.6797386095483368e-05,
+      "loss": 0.6847,
+      "step": 7725
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.6784007470373295e-05,
+      "loss": 0.6496,
+      "step": 7726
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.6770628845263215e-05,
+      "loss": 0.5508,
+      "step": 7727
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6757250220153142e-05,
+      "loss": 0.5293,
+      "step": 7728
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6743871595043062e-05,
+      "loss": 0.4791,
+      "step": 7729
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.673049296993299e-05,
+      "loss": 0.4091,
+      "step": 7730
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.671711434482291e-05,
+      "loss": 1.355,
+      "step": 7731
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6703735719712835e-05,
+      "loss": 1.3709,
+      "step": 7732
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6690357094602755e-05,
+      "loss": 1.2336,
+      "step": 7733
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6676978469492682e-05,
+      "loss": 1.2833,
+      "step": 7734
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6663599844382605e-05,
+      "loss": 1.1062,
+      "step": 7735
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.665022121927253e-05,
+      "loss": 1.3202,
+      "step": 7736
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6636842594162452e-05,
+      "loss": 1.1716,
+      "step": 7737
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6623463969052376e-05,
+      "loss": 1.2339,
+      "step": 7738
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.66100853439423e-05,
+      "loss": 1.1794,
+      "step": 7739
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6596706718832222e-05,
+      "loss": 1.1966,
+      "step": 7740
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6583328093722146e-05,
+      "loss": 1.0586,
+      "step": 7741
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6569949468612066e-05,
+      "loss": 1.1441,
+      "step": 7742
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6556570843501993e-05,
+      "loss": 1.0323,
+      "step": 7743
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6543192218391913e-05,
+      "loss": 1.1337,
+      "step": 7744
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.652981359328184e-05,
+      "loss": 1.002,
+      "step": 7745
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.651643496817176e-05,
+      "loss": 1.073,
+      "step": 7746
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6503056343061686e-05,
+      "loss": 1.0172,
+      "step": 7747
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6489677717951606e-05,
+      "loss": 1.1561,
+      "step": 7748
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6476299092841533e-05,
+      "loss": 1.0869,
+      "step": 7749
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6462920467731456e-05,
+      "loss": 1.0754,
+      "step": 7750
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.644954184262138e-05,
+      "loss": 0.9126,
+      "step": 7751
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6436163217511303e-05,
+      "loss": 0.8753,
+      "step": 7752
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6422784592401226e-05,
+      "loss": 0.9831,
+      "step": 7753
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.640940596729115e-05,
+      "loss": 1.0024,
+      "step": 7754
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6396027342181073e-05,
+      "loss": 1.0238,
+      "step": 7755
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6382648717070997e-05,
+      "loss": 1.1096,
+      "step": 7756
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.636927009196092e-05,
+      "loss": 0.918,
+      "step": 7757
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6355891466850843e-05,
+      "loss": 0.9703,
+      "step": 7758
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6342512841740767e-05,
+      "loss": 0.9548,
+      "step": 7759
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.632913421663069e-05,
+      "loss": 0.8736,
+      "step": 7760
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.6315755591520614e-05,
+      "loss": 0.81,
+      "step": 7761
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.6302376966410537e-05,
+      "loss": 0.8386,
+      "step": 7762
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.628899834130046e-05,
+      "loss": 0.7826,
+      "step": 7763
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.6275619716190384e-05,
+      "loss": 0.8499,
+      "step": 7764
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.6262241091080307e-05,
+      "loss": 0.7868,
+      "step": 7765
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.624886246597023e-05,
+      "loss": 0.9909,
+      "step": 7766
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.6235483840860157e-05,
+      "loss": 0.7654,
+      "step": 7767
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.6222105215750077e-05,
+      "loss": 0.6914,
+      "step": 7768
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.620872659064e-05,
+      "loss": 0.8347,
+      "step": 7769
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.6195347965529924e-05,
+      "loss": 0.6505,
+      "step": 7770
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.6181969340419848e-05,
+      "loss": 0.6808,
+      "step": 7771
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.616859071530977e-05,
+      "loss": 0.8236,
+      "step": 7772
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.6155212090199694e-05,
+      "loss": 0.6738,
+      "step": 7773
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.6141833465089618e-05,
+      "loss": 0.7243,
+      "step": 7774
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.612845483997954e-05,
+      "loss": 0.6476,
+      "step": 7775
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.6115076214869464e-05,
+      "loss": 0.6976,
+      "step": 7776
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.6101697589759388e-05,
+      "loss": 0.5452,
+      "step": 7777
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.608831896464931e-05,
+      "loss": 0.5487,
+      "step": 7778
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.6074940339539235e-05,
+      "loss": 0.6147,
+      "step": 7779
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.6061561714429158e-05,
+      "loss": 0.6345,
+      "step": 7780
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.604818308931908e-05,
+      "loss": 1.5343,
+      "step": 7781
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.6034804464209005e-05,
+      "loss": 1.223,
+      "step": 7782
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.6021425839098928e-05,
+      "loss": 1.1243,
+      "step": 7783
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.6008047213988855e-05,
+      "loss": 1.1696,
+      "step": 7784
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.5994668588878775e-05,
+      "loss": 1.1048,
+      "step": 7785
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.5981289963768702e-05,
+      "loss": 1.1441,
+      "step": 7786
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.5967911338658622e-05,
+      "loss": 1.1357,
+      "step": 7787
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.595453271354855e-05,
+      "loss": 1.2213,
+      "step": 7788
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.594115408843847e-05,
+      "loss": 1.2952,
+      "step": 7789
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.5927775463328395e-05,
+      "loss": 1.078,
+      "step": 7790
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.5914396838218315e-05,
+      "loss": 1.0695,
+      "step": 7791
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.5901018213108242e-05,
+      "loss": 1.0359,
+      "step": 7792
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.5887639587998162e-05,
+      "loss": 1.1362,
+      "step": 7793
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.587426096288809e-05,
+      "loss": 0.9489,
+      "step": 7794
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.586088233777801e-05,
+      "loss": 1.103,
+      "step": 7795
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.5847503712667932e-05,
+      "loss": 0.9405,
+      "step": 7796
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.5834125087557856e-05,
+      "loss": 1.0872,
+      "step": 7797
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.582074646244778e-05,
+      "loss": 1.0536,
+      "step": 7798
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.5807367837337706e-05,
+      "loss": 0.994,
+      "step": 7799
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.5793989212227626e-05,
+      "loss": 0.9519,
+      "step": 7800
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.5780610587117553e-05,
+      "loss": 0.906,
+      "step": 7801
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.5767231962007473e-05,
+      "loss": 0.9451,
+      "step": 7802
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.57538533368974e-05,
+      "loss": 0.9015,
+      "step": 7803
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.574047471178732e-05,
+      "loss": 0.933,
+      "step": 7804
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.5727096086677246e-05,
+      "loss": 0.9443,
+      "step": 7805
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.5713717461567166e-05,
+      "loss": 0.8896,
+      "step": 7806
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.5700338836457093e-05,
+      "loss": 1.0495,
+      "step": 7807
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.5686960211347013e-05,
+      "loss": 0.698,
+      "step": 7808
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.567358158623694e-05,
+      "loss": 0.8497,
+      "step": 7809
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.566020296112686e-05,
+      "loss": 0.8264,
+      "step": 7810
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.5646824336016787e-05,
+      "loss": 0.7309,
+      "step": 7811
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.5633445710906707e-05,
+      "loss": 0.712,
+      "step": 7812
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.5620067085796633e-05,
+      "loss": 0.7345,
+      "step": 7813
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.5606688460686553e-05,
+      "loss": 0.7139,
+      "step": 7814
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.559330983557648e-05,
+      "loss": 0.6554,
+      "step": 7815
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.5579931210466403e-05,
+      "loss": 0.7789,
+      "step": 7816
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.5566552585356327e-05,
+      "loss": 0.7827,
+      "step": 7817
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.555317396024625e-05,
+      "loss": 0.7759,
+      "step": 7818
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.5539795335136174e-05,
+      "loss": 0.6485,
+      "step": 7819
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.5526416710026097e-05,
+      "loss": 0.6262,
+      "step": 7820
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.5513038084916017e-05,
+      "loss": 0.5137,
+      "step": 7821
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.5499659459805944e-05,
+      "loss": 0.615,
+      "step": 7822
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.5486280834695864e-05,
+      "loss": 0.9524,
+      "step": 7823
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.547290220958579e-05,
+      "loss": 0.6005,
+      "step": 7824
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.545952358447571e-05,
+      "loss": 0.6513,
+      "step": 7825
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5446144959365637e-05,
+      "loss": 0.6359,
+      "step": 7826
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5432766334255557e-05,
+      "loss": 0.4811,
+      "step": 7827
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5419387709145484e-05,
+      "loss": 0.6497,
+      "step": 7828
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5406009084035404e-05,
+      "loss": 0.56,
+      "step": 7829
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.539263045892533e-05,
+      "loss": 0.5013,
+      "step": 7830
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.537925183381525e-05,
+      "loss": 1.5302,
+      "step": 7831
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5365873208705178e-05,
+      "loss": 1.2277,
+      "step": 7832
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.53524945835951e-05,
+      "loss": 1.192,
+      "step": 7833
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5339115958485024e-05,
+      "loss": 1.2899,
+      "step": 7834
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5325737333374948e-05,
+      "loss": 1.182,
+      "step": 7835
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.531235870826487e-05,
+      "loss": 0.9956,
+      "step": 7836
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5298980083154795e-05,
+      "loss": 1.2471,
+      "step": 7837
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5285601458044718e-05,
+      "loss": 1.1777,
+      "step": 7838
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.527222283293464e-05,
+      "loss": 1.1064,
+      "step": 7839
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5258844207824565e-05,
+      "loss": 1.1985,
+      "step": 7840
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5245465582714488e-05,
+      "loss": 1.1799,
+      "step": 7841
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.523208695760441e-05,
+      "loss": 1.1384,
+      "step": 7842
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5218708332494335e-05,
+      "loss": 1.0629,
+      "step": 7843
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.520532970738426e-05,
+      "loss": 1.1522,
+      "step": 7844
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5191951082274182e-05,
+      "loss": 1.0087,
+      "step": 7845
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5178572457164105e-05,
+      "loss": 1.0605,
+      "step": 7846
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.516519383205403e-05,
+      "loss": 1.1867,
+      "step": 7847
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.515181520694395e-05,
+      "loss": 1.1548,
+      "step": 7848
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5138436581833875e-05,
+      "loss": 0.9412,
+      "step": 7849
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.51250579567238e-05,
+      "loss": 1.0681,
+      "step": 7850
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5111679331613722e-05,
+      "loss": 1.0289,
+      "step": 7851
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5098300706503645e-05,
+      "loss": 0.9311,
+      "step": 7852
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.508492208139357e-05,
+      "loss": 0.9697,
+      "step": 7853
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5071543456283492e-05,
+      "loss": 0.9627,
+      "step": 7854
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5058164831173416e-05,
+      "loss": 0.9863,
+      "step": 7855
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.504478620606334e-05,
+      "loss": 0.8693,
+      "step": 7856
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5031407580953262e-05,
+      "loss": 0.9943,
+      "step": 7857
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.5018028955843186e-05,
+      "loss": 0.9919,
+      "step": 7858
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.500465033073311e-05,
+      "loss": 0.9225,
+      "step": 7859
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.4991271705623033e-05,
+      "loss": 0.9206,
+      "step": 7860
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.4977893080512956e-05,
+      "loss": 1.0181,
+      "step": 7861
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.496451445540288e-05,
+      "loss": 0.7897,
+      "step": 7862
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.4951135830292803e-05,
+      "loss": 0.9418,
+      "step": 7863
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.4937757205182726e-05,
+      "loss": 0.6633,
+      "step": 7864
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.4924378580072653e-05,
+      "loss": 0.7289,
+      "step": 7865
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.4910999954962573e-05,
+      "loss": 0.8066,
+      "step": 7866
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.48976213298525e-05,
+      "loss": 0.7803,
+      "step": 7867
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.488424270474242e-05,
+      "loss": 0.7494,
+      "step": 7868
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.4870864079632347e-05,
+      "loss": 0.7317,
+      "step": 7869
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.4857485454522267e-05,
+      "loss": 0.7596,
+      "step": 7870
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.4844106829412193e-05,
+      "loss": 0.6677,
+      "step": 7871
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.4830728204302113e-05,
+      "loss": 0.7609,
+      "step": 7872
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.481734957919204e-05,
+      "loss": 0.5409,
+      "step": 7873
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.480397095408196e-05,
+      "loss": 0.6102,
+      "step": 7874
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.4790592328971883e-05,
+      "loss": 0.6315,
+      "step": 7875
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.4777213703861807e-05,
+      "loss": 0.6689,
+      "step": 7876
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.476383507875173e-05,
+      "loss": 0.6203,
+      "step": 7877
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.4750456453641654e-05,
+      "loss": 0.5477,
+      "step": 7878
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.4737077828531577e-05,
+      "loss": 0.5429,
+      "step": 7879
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.47236992034215e-05,
+      "loss": 0.5562,
+      "step": 7880
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.4710320578311424e-05,
+      "loss": 1.6016,
+      "step": 7881
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.469694195320135e-05,
+      "loss": 1.3746,
+      "step": 7882
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.468356332809127e-05,
+      "loss": 1.4648,
+      "step": 7883
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.4670184702981197e-05,
+      "loss": 1.3487,
+      "step": 7884
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.4656806077871117e-05,
+      "loss": 1.0818,
+      "step": 7885
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.4643427452761044e-05,
+      "loss": 1.2912,
+      "step": 7886
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.4630048827650964e-05,
+      "loss": 1.2594,
+      "step": 7887
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.461667020254089e-05,
+      "loss": 1.1016,
+      "step": 7888
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.460329157743081e-05,
+      "loss": 1.1711,
+      "step": 7889
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.4589912952320738e-05,
+      "loss": 1.1629,
+      "step": 7890
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.4576534327210658e-05,
+      "loss": 1.0503,
+      "step": 7891
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.4563155702100584e-05,
+      "loss": 1.1221,
+      "step": 7892
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.4549777076990504e-05,
+      "loss": 1.0691,
+      "step": 7893
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.453639845188043e-05,
+      "loss": 0.9879,
+      "step": 7894
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.452301982677035e-05,
+      "loss": 1.0579,
+      "step": 7895
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.4509641201660278e-05,
+      "loss": 0.9966,
+      "step": 7896
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.4496262576550198e-05,
+      "loss": 0.9507,
+      "step": 7897
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.4482883951440125e-05,
+      "loss": 0.9957,
+      "step": 7898
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.4469505326330048e-05,
+      "loss": 0.938,
+      "step": 7899
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.4456126701219968e-05,
+      "loss": 1.022,
+      "step": 7900
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.4442748076109895e-05,
+      "loss": 0.9208,
+      "step": 7901
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.4429369450999815e-05,
+      "loss": 0.9325,
+      "step": 7902
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.4415990825889742e-05,
+      "loss": 0.9225,
+      "step": 7903
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.4402612200779662e-05,
+      "loss": 1.139,
+      "step": 7904
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.438923357566959e-05,
+      "loss": 1.0692,
+      "step": 7905
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.437585495055951e-05,
+      "loss": 1.0301,
+      "step": 7906
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.4362476325449435e-05,
+      "loss": 0.9639,
+      "step": 7907
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.4349097700339355e-05,
+      "loss": 1.0165,
+      "step": 7908
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.4335719075229282e-05,
+      "loss": 0.8444,
+      "step": 7909
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.4322340450119202e-05,
+      "loss": 0.9499,
+      "step": 7910
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.430896182500913e-05,
+      "loss": 0.8552,
+      "step": 7911
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.429558319989905e-05,
+      "loss": 1.0214,
+      "step": 7912
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.4282204574788976e-05,
+      "loss": 0.9222,
+      "step": 7913
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.42688259496789e-05,
+      "loss": 0.7762,
+      "step": 7914
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.4255447324568822e-05,
+      "loss": 0.749,
+      "step": 7915
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.4242068699458746e-05,
+      "loss": 0.7145,
+      "step": 7916
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.422869007434867e-05,
+      "loss": 0.883,
+      "step": 7917
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.4215311449238593e-05,
+      "loss": 0.6375,
+      "step": 7918
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.4201932824128516e-05,
+      "loss": 0.6623,
+      "step": 7919
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.418855419901844e-05,
+      "loss": 0.7391,
+      "step": 7920
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.4175175573908363e-05,
+      "loss": 0.5787,
+      "step": 7921
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.4161796948798286e-05,
+      "loss": 0.6765,
+      "step": 7922
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.414841832368821e-05,
+      "loss": 0.6816,
+      "step": 7923
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.4135039698578133e-05,
+      "loss": 0.6288,
+      "step": 7924
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.4121661073468056e-05,
+      "loss": 0.6021,
+      "step": 7925
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.410828244835798e-05,
+      "loss": 0.4603,
+      "step": 7926
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.40949038232479e-05,
+      "loss": 0.5092,
+      "step": 7927
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.4081525198137827e-05,
+      "loss": 0.4557,
+      "step": 7928
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.4068146573027747e-05,
+      "loss": 0.5515,
+      "step": 7929
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.4054767947917673e-05,
+      "loss": 0.5697,
+      "step": 7930
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.4041389322807597e-05,
+      "loss": 1.6036,
+      "step": 7931
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.402801069769752e-05,
+      "loss": 1.6694,
+      "step": 7932
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.4014632072587443e-05,
+      "loss": 1.154,
+      "step": 7933
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.4001253447477367e-05,
+      "loss": 1.2331,
+      "step": 7934
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.398787482236729e-05,
+      "loss": 1.2734,
+      "step": 7935
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.3974496197257214e-05,
+      "loss": 1.1781,
+      "step": 7936
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.3961117572147137e-05,
+      "loss": 1.0544,
+      "step": 7937
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.394773894703706e-05,
+      "loss": 1.1501,
+      "step": 7938
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.3934360321926984e-05,
+      "loss": 1.4017,
+      "step": 7939
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.3920981696816907e-05,
+      "loss": 1.1715,
+      "step": 7940
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.390760307170683e-05,
+      "loss": 0.9531,
+      "step": 7941
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.3894224446596754e-05,
+      "loss": 1.111,
+      "step": 7942
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.3880845821486677e-05,
+      "loss": 1.1025,
+      "step": 7943
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.38674671963766e-05,
+      "loss": 1.0324,
+      "step": 7944
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.3854088571266524e-05,
+      "loss": 1.1032,
+      "step": 7945
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.3840709946156448e-05,
+      "loss": 1.0371,
+      "step": 7946
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.382733132104637e-05,
+      "loss": 1.0225,
+      "step": 7947
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.3813952695936298e-05,
+      "loss": 1.0712,
+      "step": 7948
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.3800574070826218e-05,
+      "loss": 0.9537,
+      "step": 7949
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.3787195445716145e-05,
+      "loss": 1.0199,
+      "step": 7950
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.3773816820606065e-05,
+      "loss": 1.1945,
+      "step": 7951
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.376043819549599e-05,
+      "loss": 0.9508,
+      "step": 7952
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.374705957038591e-05,
+      "loss": 0.9586,
+      "step": 7953
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.3733680945275835e-05,
+      "loss": 0.8802,
+      "step": 7954
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.3720302320165758e-05,
+      "loss": 1.0402,
+      "step": 7955
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.370692369505568e-05,
+      "loss": 0.9066,
+      "step": 7956
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.3693545069945605e-05,
+      "loss": 1.029,
+      "step": 7957
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.3680166444835528e-05,
+      "loss": 0.9556,
+      "step": 7958
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.366678781972545e-05,
+      "loss": 1.0198,
+      "step": 7959
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.3653409194615375e-05,
+      "loss": 0.9725,
+      "step": 7960
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.36400305695053e-05,
+      "loss": 1.0215,
+      "step": 7961
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.3626651944395222e-05,
+      "loss": 0.801,
+      "step": 7962
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.361327331928515e-05,
+      "loss": 0.6619,
+      "step": 7963
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.359989469417507e-05,
+      "loss": 0.716,
+      "step": 7964
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.3586516069064995e-05,
+      "loss": 0.6822,
+      "step": 7965
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.3573137443954915e-05,
+      "loss": 0.8459,
+      "step": 7966
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.3559758818844842e-05,
+      "loss": 0.6566,
+      "step": 7967
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.3546380193734762e-05,
+      "loss": 0.7896,
+      "step": 7968
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.353300156862469e-05,
+      "loss": 0.7029,
+      "step": 7969
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.351962294351461e-05,
+      "loss": 0.74,
+      "step": 7970
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.3506244318404536e-05,
+      "loss": 0.5674,
+      "step": 7971
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.3492865693294456e-05,
+      "loss": 0.5895,
+      "step": 7972
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.3479487068184382e-05,
+      "loss": 0.6439,
+      "step": 7973
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.3466108443074302e-05,
+      "loss": 0.5842,
+      "step": 7974
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.345272981796423e-05,
+      "loss": 0.6519,
+      "step": 7975
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.343935119285415e-05,
+      "loss": 0.6028,
+      "step": 7976
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.3425972567744076e-05,
+      "loss": 0.5208,
+      "step": 7977
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.3412593942633996e-05,
+      "loss": 0.5146,
+      "step": 7978
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.3399215317523923e-05,
+      "loss": 0.5155,
+      "step": 7979
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.3385836692413846e-05,
+      "loss": 0.6652,
+      "step": 7980
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.3372458067303766e-05,
+      "loss": 1.6791,
+      "step": 7981
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.3359079442193693e-05,
+      "loss": 1.4805,
+      "step": 7982
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.3345700817083613e-05,
+      "loss": 1.2353,
+      "step": 7983
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.333232219197354e-05,
+      "loss": 1.1751,
+      "step": 7984
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.331894356686346e-05,
+      "loss": 1.3655,
+      "step": 7985
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.3305564941753387e-05,
+      "loss": 1.1862,
+      "step": 7986
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.3292186316643307e-05,
+      "loss": 1.1784,
+      "step": 7987
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.3278807691533233e-05,
+      "loss": 1.0884,
+      "step": 7988
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.3265429066423153e-05,
+      "loss": 1.137,
+      "step": 7989
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.325205044131308e-05,
+      "loss": 1.1877,
+      "step": 7990
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.3238671816203e-05,
+      "loss": 1.2565,
+      "step": 7991
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.3225293191092927e-05,
+      "loss": 1.1612,
+      "step": 7992
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.3211914565982847e-05,
+      "loss": 1.202,
+      "step": 7993
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.3198535940872774e-05,
+      "loss": 1.1145,
+      "step": 7994
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.3185157315762694e-05,
+      "loss": 1.0558,
+      "step": 7995
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.317177869065262e-05,
+      "loss": 1.0678,
+      "step": 7996
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.3158400065542544e-05,
+      "loss": 1.0708,
+      "step": 7997
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.3145021440432467e-05,
+      "loss": 0.9813,
+      "step": 7998
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.313164281532239e-05,
+      "loss": 0.9198,
+      "step": 7999
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.3118264190212314e-05,
+      "loss": 1.1975,
+      "step": 8000
+    },
+    {
+      "epoch": 2.47,
+      "eval_bleu": 6.400614109364979e-17,
+      "eval_loss": 1.6715521812438965,
+      "eval_runtime": 2557.718,
+      "eval_samples_per_second": 5.771,
+      "eval_steps_per_second": 0.721,
+      "step": 8000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.3104885565102237e-05,
+      "loss": 1.044,
+      "step": 8001
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.309150693999216e-05,
+      "loss": 0.948,
+      "step": 8002
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.3078128314882084e-05,
+      "loss": 0.9991,
+      "step": 8003
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.3064749689772008e-05,
+      "loss": 1.0284,
+      "step": 8004
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.305137106466193e-05,
+      "loss": 0.9829,
+      "step": 8005
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.303799243955185e-05,
+      "loss": 0.9731,
+      "step": 8006
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.3024613814441778e-05,
+      "loss": 0.8856,
+      "step": 8007
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.3011235189331698e-05,
+      "loss": 0.9295,
+      "step": 8008
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.2997856564221625e-05,
+      "loss": 1.0023,
+      "step": 8009
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.2984477939111545e-05,
+      "loss": 0.8191,
+      "step": 8010
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.297109931400147e-05,
+      "loss": 0.799,
+      "step": 8011
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.2957720688891395e-05,
+      "loss": 0.9057,
+      "step": 8012
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.2944342063781318e-05,
+      "loss": 0.8067,
+      "step": 8013
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.293096343867124e-05,
+      "loss": 0.9123,
+      "step": 8014
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.2917584813561165e-05,
+      "loss": 0.7374,
+      "step": 8015
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.2904206188451088e-05,
+      "loss": 0.8937,
+      "step": 8016
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.289082756334101e-05,
+      "loss": 0.7236,
+      "step": 8017
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.2877448938230935e-05,
+      "loss": 0.7776,
+      "step": 8018
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.286407031312086e-05,
+      "loss": 0.667,
+      "step": 8019
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.2850691688010782e-05,
+      "loss": 0.6555,
+      "step": 8020
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.2837313062900705e-05,
+      "loss": 0.5795,
+      "step": 8021
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.282393443779063e-05,
+      "loss": 0.658,
+      "step": 8022
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.2810555812680552e-05,
+      "loss": 0.7538,
+      "step": 8023
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.2797177187570475e-05,
+      "loss": 0.6257,
+      "step": 8024
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.27837985624604e-05,
+      "loss": 0.6036,
+      "step": 8025
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.2770419937350322e-05,
+      "loss": 0.6616,
+      "step": 8026
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.2757041312240246e-05,
+      "loss": 0.5385,
+      "step": 8027
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.274366268713017e-05,
+      "loss": 0.5978,
+      "step": 8028
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.2730284062020096e-05,
+      "loss": 0.579,
+      "step": 8029
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.2716905436910016e-05,
+      "loss": 0.5417,
+      "step": 8030
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.2703526811799943e-05,
+      "loss": 1.476,
+      "step": 8031
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.2690148186689863e-05,
+      "loss": 1.3941,
+      "step": 8032
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.2676769561579786e-05,
+      "loss": 1.2951,
+      "step": 8033
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.266339093646971e-05,
+      "loss": 1.1679,
+      "step": 8034
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.2650012311359633e-05,
+      "loss": 1.1017,
+      "step": 8035
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.2636633686249556e-05,
+      "loss": 1.1259,
+      "step": 8036
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.262325506113948e-05,
+      "loss": 1.0556,
+      "step": 8037
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.2609876436029403e-05,
+      "loss": 1.2467,
+      "step": 8038
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.2596497810919326e-05,
+      "loss": 1.2205,
+      "step": 8039
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.258311918580925e-05,
+      "loss": 1.0942,
+      "step": 8040
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.2569740560699173e-05,
+      "loss": 1.1623,
+      "step": 8041
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.2556361935589096e-05,
+      "loss": 1.0921,
+      "step": 8042
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.254298331047902e-05,
+      "loss": 1.163,
+      "step": 8043
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.2529604685368943e-05,
+      "loss": 1.1756,
+      "step": 8044
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.2516226060258867e-05,
+      "loss": 1.1059,
+      "step": 8045
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.2502847435148793e-05,
+      "loss": 1.1495,
+      "step": 8046
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.2489468810038713e-05,
+      "loss": 1.005,
+      "step": 8047
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.247609018492864e-05,
+      "loss": 1.0847,
+      "step": 8048
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.246271155981856e-05,
+      "loss": 1.0993,
+      "step": 8049
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.2449332934708487e-05,
+      "loss": 1.0321,
+      "step": 8050
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.2435954309598407e-05,
+      "loss": 1.0332,
+      "step": 8051
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.2422575684488334e-05,
+      "loss": 0.9177,
+      "step": 8052
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.2409197059378254e-05,
+      "loss": 1.022,
+      "step": 8053
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.239581843426818e-05,
+      "loss": 0.8557,
+      "step": 8054
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.23824398091581e-05,
+      "loss": 1.0434,
+      "step": 8055
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.2369061184048027e-05,
+      "loss": 1.0137,
+      "step": 8056
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.2355682558937947e-05,
+      "loss": 0.783,
+      "step": 8057
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.2342303933827874e-05,
+      "loss": 0.8916,
+      "step": 8058
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.2328925308717794e-05,
+      "loss": 0.9136,
+      "step": 8059
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.2315546683607717e-05,
+      "loss": 0.9282,
+      "step": 8060
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.2302168058497644e-05,
+      "loss": 0.8099,
+      "step": 8061
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.2288789433387564e-05,
+      "loss": 0.8526,
+      "step": 8062
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.227541080827749e-05,
+      "loss": 0.8269,
+      "step": 8063
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.226203218316741e-05,
+      "loss": 0.6972,
+      "step": 8064
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.2248653558057338e-05,
+      "loss": 0.8749,
+      "step": 8065
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.2235274932947258e-05,
+      "loss": 0.9062,
+      "step": 8066
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.2221896307837185e-05,
+      "loss": 0.7913,
+      "step": 8067
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.2208517682727105e-05,
+      "loss": 0.6981,
+      "step": 8068
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.219513905761703e-05,
+      "loss": 0.6559,
+      "step": 8069
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.218176043250695e-05,
+      "loss": 0.6847,
+      "step": 8070
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.2168381807396878e-05,
+      "loss": 0.6107,
+      "step": 8071
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.2155003182286798e-05,
+      "loss": 0.8048,
+      "step": 8072
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.2141624557176725e-05,
+      "loss": 0.8199,
+      "step": 8073
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.2128245932066645e-05,
+      "loss": 0.5489,
+      "step": 8074
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.211486730695657e-05,
+      "loss": 0.6247,
+      "step": 8075
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.210148868184649e-05,
+      "loss": 0.5443,
+      "step": 8076
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.208811005673642e-05,
+      "loss": 0.6222,
+      "step": 8077
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.2074731431626342e-05,
+      "loss": 0.4768,
+      "step": 8078
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.2061352806516265e-05,
+      "loss": 0.6232,
+      "step": 8079
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.204797418140619e-05,
+      "loss": 0.4881,
+      "step": 8080
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.2034595556296112e-05,
+      "loss": 1.6001,
+      "step": 8081
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.2021216931186035e-05,
+      "loss": 1.2992,
+      "step": 8082
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.200783830607596e-05,
+      "loss": 1.0496,
+      "step": 8083
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.1994459680965882e-05,
+      "loss": 1.1395,
+      "step": 8084
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.1981081055855806e-05,
+      "loss": 1.1049,
+      "step": 8085
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.196770243074573e-05,
+      "loss": 1.1659,
+      "step": 8086
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.195432380563565e-05,
+      "loss": 1.1918,
+      "step": 8087
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.1940945180525576e-05,
+      "loss": 1.0737,
+      "step": 8088
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.1927566555415496e-05,
+      "loss": 1.1365,
+      "step": 8089
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.1914187930305423e-05,
+      "loss": 1.1084,
+      "step": 8090
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.1900809305195343e-05,
+      "loss": 1.1084,
+      "step": 8091
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.188743068008527e-05,
+      "loss": 0.8498,
+      "step": 8092
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.187405205497519e-05,
+      "loss": 0.9706,
+      "step": 8093
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.1860673429865116e-05,
+      "loss": 0.9501,
+      "step": 8094
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.184729480475504e-05,
+      "loss": 1.1033,
+      "step": 8095
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.1833916179644963e-05,
+      "loss": 1.1686,
+      "step": 8096
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.1820537554534886e-05,
+      "loss": 1.1514,
+      "step": 8097
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.180715892942481e-05,
+      "loss": 0.9158,
+      "step": 8098
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.1793780304314733e-05,
+      "loss": 0.9813,
+      "step": 8099
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.1780401679204656e-05,
+      "loss": 0.8909,
+      "step": 8100
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.176702305409458e-05,
+      "loss": 1.0198,
+      "step": 8101
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.1753644428984503e-05,
+      "loss": 1.0002,
+      "step": 8102
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.1740265803874427e-05,
+      "loss": 1.0772,
+      "step": 8103
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.172688717876435e-05,
+      "loss": 0.9563,
+      "step": 8104
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.1713508553654273e-05,
+      "loss": 0.9124,
+      "step": 8105
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.1700129928544197e-05,
+      "loss": 0.9711,
+      "step": 8106
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.168675130343412e-05,
+      "loss": 0.8307,
+      "step": 8107
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.1673372678324044e-05,
+      "loss": 0.8298,
+      "step": 8108
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.1659994053213967e-05,
+      "loss": 0.9038,
+      "step": 8109
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.1646615428103894e-05,
+      "loss": 0.7819,
+      "step": 8110
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.1633236802993814e-05,
+      "loss": 0.9534,
+      "step": 8111
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.161985817788374e-05,
+      "loss": 0.6608,
+      "step": 8112
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.160647955277366e-05,
+      "loss": 0.7887,
+      "step": 8113
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.1593100927663584e-05,
+      "loss": 0.7852,
+      "step": 8114
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.1579722302553507e-05,
+      "loss": 0.751,
+      "step": 8115
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.156634367744343e-05,
+      "loss": 0.7485,
+      "step": 8116
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.1552965052333354e-05,
+      "loss": 0.7326,
+      "step": 8117
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.1539586427223277e-05,
+      "loss": 0.8647,
+      "step": 8118
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.15262078021132e-05,
+      "loss": 0.6844,
+      "step": 8119
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.1512829177003124e-05,
+      "loss": 0.5489,
+      "step": 8120
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.1499450551893048e-05,
+      "loss": 0.5139,
+      "step": 8121
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.148607192678297e-05,
+      "loss": 0.6405,
+      "step": 8122
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.1472693301672894e-05,
+      "loss": 0.6703,
+      "step": 8123
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.1459314676562818e-05,
+      "loss": 0.5435,
+      "step": 8124
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.144593605145274e-05,
+      "loss": 0.6415,
+      "step": 8125
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.1432557426342665e-05,
+      "loss": 0.5526,
+      "step": 8126
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.141917880123259e-05,
+      "loss": 0.5018,
+      "step": 8127
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.140580017612251e-05,
+      "loss": 0.4139,
+      "step": 8128
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.1392421551012438e-05,
+      "loss": 0.5253,
+      "step": 8129
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.1379042925902358e-05,
+      "loss": 0.6268,
+      "step": 8130
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.1365664300792285e-05,
+      "loss": 1.3825,
+      "step": 8131
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.1352285675682205e-05,
+      "loss": 1.33,
+      "step": 8132
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.133890705057213e-05,
+      "loss": 1.1281,
+      "step": 8133
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.132552842546205e-05,
+      "loss": 1.1484,
+      "step": 8134
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.131214980035198e-05,
+      "loss": 1.0502,
+      "step": 8135
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.12987711752419e-05,
+      "loss": 1.1594,
+      "step": 8136
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.1285392550131825e-05,
+      "loss": 1.0562,
+      "step": 8137
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.1272013925021745e-05,
+      "loss": 1.0613,
+      "step": 8138
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.125863529991167e-05,
+      "loss": 1.2057,
+      "step": 8139
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.1245256674801592e-05,
+      "loss": 0.9985,
+      "step": 8140
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.1231878049691515e-05,
+      "loss": 1.0741,
+      "step": 8141
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.121849942458144e-05,
+      "loss": 1.1712,
+      "step": 8142
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.1205120799471362e-05,
+      "loss": 1.0204,
+      "step": 8143
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.119174217436129e-05,
+      "loss": 0.8647,
+      "step": 8144
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.117836354925121e-05,
+      "loss": 0.9749,
+      "step": 8145
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.1164984924141136e-05,
+      "loss": 1.0906,
+      "step": 8146
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.1151606299031056e-05,
+      "loss": 1.1382,
+      "step": 8147
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.1138227673920983e-05,
+      "loss": 1.0849,
+      "step": 8148
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.1124849048810903e-05,
+      "loss": 1.0864,
+      "step": 8149
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.111147042370083e-05,
+      "loss": 0.9483,
+      "step": 8150
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.109809179859075e-05,
+      "loss": 1.0154,
+      "step": 8151
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.1084713173480676e-05,
+      "loss": 0.9393,
+      "step": 8152
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.1071334548370596e-05,
+      "loss": 1.0419,
+      "step": 8153
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.1057955923260523e-05,
+      "loss": 0.9222,
+      "step": 8154
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.1044577298150443e-05,
+      "loss": 0.7994,
+      "step": 8155
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.103119867304037e-05,
+      "loss": 1.0804,
+      "step": 8156
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.101782004793029e-05,
+      "loss": 0.9381,
+      "step": 8157
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.1004441422820216e-05,
+      "loss": 0.7859,
+      "step": 8158
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.099106279771014e-05,
+      "loss": 0.8632,
+      "step": 8159
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0977684172600063e-05,
+      "loss": 0.8124,
+      "step": 8160
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0964305547489987e-05,
+      "loss": 0.882,
+      "step": 8161
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.095092692237991e-05,
+      "loss": 0.8448,
+      "step": 8162
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0937548297269833e-05,
+      "loss": 0.8316,
+      "step": 8163
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0924169672159757e-05,
+      "loss": 0.81,
+      "step": 8164
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.091079104704968e-05,
+      "loss": 0.9277,
+      "step": 8165
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.08974124219396e-05,
+      "loss": 0.8487,
+      "step": 8166
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0884033796829527e-05,
+      "loss": 0.7942,
+      "step": 8167
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0870655171719447e-05,
+      "loss": 0.683,
+      "step": 8168
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0857276546609374e-05,
+      "loss": 0.7457,
+      "step": 8169
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0843897921499294e-05,
+      "loss": 0.633,
+      "step": 8170
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.083051929638922e-05,
+      "loss": 0.6169,
+      "step": 8171
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.081714067127914e-05,
+      "loss": 0.7144,
+      "step": 8172
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0803762046169067e-05,
+      "loss": 0.8062,
+      "step": 8173
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0790383421058987e-05,
+      "loss": 0.5936,
+      "step": 8174
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0777004795948914e-05,
+      "loss": 0.6141,
+      "step": 8175
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0763626170838837e-05,
+      "loss": 0.5361,
+      "step": 8176
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.075024754572876e-05,
+      "loss": 0.5461,
+      "step": 8177
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0736868920618684e-05,
+      "loss": 0.5263,
+      "step": 8178
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0723490295508608e-05,
+      "loss": 0.3661,
+      "step": 8179
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.071011167039853e-05,
+      "loss": 0.4037,
+      "step": 8180
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0696733045288454e-05,
+      "loss": 1.5033,
+      "step": 8181
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0683354420178378e-05,
+      "loss": 1.4252,
+      "step": 8182
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.06699757950683e-05,
+      "loss": 1.1588,
+      "step": 8183
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0656597169958225e-05,
+      "loss": 1.2063,
+      "step": 8184
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0643218544848148e-05,
+      "loss": 1.2503,
+      "step": 8185
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.062983991973807e-05,
+      "loss": 1.2263,
+      "step": 8186
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0616461294627995e-05,
+      "loss": 1.1042,
+      "step": 8187
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0603082669517918e-05,
+      "loss": 1.208,
+      "step": 8188
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.058970404440784e-05,
+      "loss": 1.3036,
+      "step": 8189
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0576325419297765e-05,
+      "loss": 1.2974,
+      "step": 8190
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.056294679418769e-05,
+      "loss": 1.1253,
+      "step": 8191
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0549568169077612e-05,
+      "loss": 1.1002,
+      "step": 8192
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0536189543967535e-05,
+      "loss": 1.0172,
+      "step": 8193
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.052281091885746e-05,
+      "loss": 0.8676,
+      "step": 8194
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0509432293747382e-05,
+      "loss": 1.0617,
+      "step": 8195
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0496053668637305e-05,
+      "loss": 1.1399,
+      "step": 8196
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.048267504352723e-05,
+      "loss": 1.016,
+      "step": 8197
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0469296418417152e-05,
+      "loss": 0.9794,
+      "step": 8198
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0455917793307075e-05,
+      "loss": 1.0518,
+      "step": 8199
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0442539168197e-05,
+      "loss": 1.0283,
+      "step": 8200
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0429160543086922e-05,
+      "loss": 0.8772,
+      "step": 8201
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0415781917976846e-05,
+      "loss": 0.8367,
+      "step": 8202
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.040240329286677e-05,
+      "loss": 0.9095,
+      "step": 8203
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0389024667756692e-05,
+      "loss": 0.9279,
+      "step": 8204
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0375646042646616e-05,
+      "loss": 1.0224,
+      "step": 8205
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.036226741753654e-05,
+      "loss": 0.7889,
+      "step": 8206
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0348888792426463e-05,
+      "loss": 0.922,
+      "step": 8207
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.033551016731639e-05,
+      "loss": 0.7199,
+      "step": 8208
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.032213154220631e-05,
+      "loss": 0.8419,
+      "step": 8209
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0308752917096236e-05,
+      "loss": 0.8354,
+      "step": 8210
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0295374291986156e-05,
+      "loss": 0.7856,
+      "step": 8211
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0281995666876083e-05,
+      "loss": 0.8773,
+      "step": 8212
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0268617041766003e-05,
+      "loss": 0.7704,
+      "step": 8213
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.025523841665593e-05,
+      "loss": 0.7343,
+      "step": 8214
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.024185979154585e-05,
+      "loss": 0.8973,
+      "step": 8215
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0228481166435776e-05,
+      "loss": 0.8993,
+      "step": 8216
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0215102541325696e-05,
+      "loss": 0.7896,
+      "step": 8217
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.020172391621562e-05,
+      "loss": 0.8303,
+      "step": 8218
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0188345291105543e-05,
+      "loss": 0.6677,
+      "step": 8219
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0174966665995467e-05,
+      "loss": 0.5911,
+      "step": 8220
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.016158804088539e-05,
+      "loss": 0.7033,
+      "step": 8221
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0148209415775313e-05,
+      "loss": 0.6087,
+      "step": 8222
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0134830790665237e-05,
+      "loss": 0.7209,
+      "step": 8223
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.012145216555516e-05,
+      "loss": 0.6351,
+      "step": 8224
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0108073540445087e-05,
+      "loss": 0.6292,
+      "step": 8225
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0094694915335007e-05,
+      "loss": 0.5238,
+      "step": 8226
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0081316290224934e-05,
+      "loss": 0.5486,
+      "step": 8227
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0067937665114854e-05,
+      "loss": 0.5914,
+      "step": 8228
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.005455904000478e-05,
+      "loss": 0.4146,
+      "step": 8229
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.00411804148947e-05,
+      "loss": 0.6557,
+      "step": 8230
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0027801789784627e-05,
+      "loss": 1.4997,
+      "step": 8231
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0014423164674547e-05,
+      "loss": 1.2478,
+      "step": 8232
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0001044539564474e-05,
+      "loss": 1.126,
+      "step": 8233
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9987665914454394e-05,
+      "loss": 1.2347,
+      "step": 8234
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.997428728934432e-05,
+      "loss": 1.4393,
+      "step": 8235
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.996090866423424e-05,
+      "loss": 1.2304,
+      "step": 8236
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9947530039124168e-05,
+      "loss": 1.062,
+      "step": 8237
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9934151414014088e-05,
+      "loss": 1.1152,
+      "step": 8238
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9920772788904014e-05,
+      "loss": 1.257,
+      "step": 8239
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9907394163793934e-05,
+      "loss": 1.2155,
+      "step": 8240
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.989401553868386e-05,
+      "loss": 1.1583,
+      "step": 8241
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9880636913573785e-05,
+      "loss": 1.0925,
+      "step": 8242
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9867258288463708e-05,
+      "loss": 1.1963,
+      "step": 8243
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.985387966335363e-05,
+      "loss": 1.0991,
+      "step": 8244
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.984050103824355e-05,
+      "loss": 1.0517,
+      "step": 8245
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9827122413133478e-05,
+      "loss": 1.0853,
+      "step": 8246
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9813743788023398e-05,
+      "loss": 1.0101,
+      "step": 8247
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9800365162913325e-05,
+      "loss": 0.9618,
+      "step": 8248
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9786986537803245e-05,
+      "loss": 0.9373,
+      "step": 8249
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9773607912693172e-05,
+      "loss": 1.0594,
+      "step": 8250
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9760229287583092e-05,
+      "loss": 1.0135,
+      "step": 8251
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.974685066247302e-05,
+      "loss": 0.9462,
+      "step": 8252
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.973347203736294e-05,
+      "loss": 0.9558,
+      "step": 8253
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9720093412252865e-05,
+      "loss": 0.9767,
+      "step": 8254
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9706714787142785e-05,
+      "loss": 0.9363,
+      "step": 8255
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9693336162032712e-05,
+      "loss": 0.882,
+      "step": 8256
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9679957536922635e-05,
+      "loss": 1.014,
+      "step": 8257
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.966657891181256e-05,
+      "loss": 0.9029,
+      "step": 8258
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9653200286702482e-05,
+      "loss": 0.9305,
+      "step": 8259
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9639821661592406e-05,
+      "loss": 0.9218,
+      "step": 8260
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.962644303648233e-05,
+      "loss": 0.8057,
+      "step": 8261
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9613064411372252e-05,
+      "loss": 0.879,
+      "step": 8262
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9599685786262176e-05,
+      "loss": 0.895,
+      "step": 8263
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.95863071611521e-05,
+      "loss": 0.8713,
+      "step": 8264
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9572928536042023e-05,
+      "loss": 0.6426,
+      "step": 8265
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9559549910931946e-05,
+      "loss": 0.5781,
+      "step": 8266
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.954617128582187e-05,
+      "loss": 1.0014,
+      "step": 8267
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9532792660711793e-05,
+      "loss": 0.8041,
+      "step": 8268
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9519414035601716e-05,
+      "loss": 0.7054,
+      "step": 8269
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.950603541049164e-05,
+      "loss": 0.7098,
+      "step": 8270
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9492656785381563e-05,
+      "loss": 0.749,
+      "step": 8271
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9479278160271483e-05,
+      "loss": 0.4907,
+      "step": 8272
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.946589953516141e-05,
+      "loss": 0.6805,
+      "step": 8273
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9452520910051333e-05,
+      "loss": 0.6166,
+      "step": 8274
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9439142284941256e-05,
+      "loss": 0.558,
+      "step": 8275
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.942576365983118e-05,
+      "loss": 0.5429,
+      "step": 8276
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9412385034721103e-05,
+      "loss": 0.5188,
+      "step": 8277
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9399006409611027e-05,
+      "loss": 0.6604,
+      "step": 8278
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.938562778450095e-05,
+      "loss": 0.4876,
+      "step": 8279
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9372249159390873e-05,
+      "loss": 0.513,
+      "step": 8280
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9358870534280797e-05,
+      "loss": 1.3911,
+      "step": 8281
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.934549190917072e-05,
+      "loss": 1.2728,
+      "step": 8282
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9332113284060644e-05,
+      "loss": 1.273,
+      "step": 8283
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9318734658950567e-05,
+      "loss": 1.2606,
+      "step": 8284
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.930535603384049e-05,
+      "loss": 1.0561,
+      "step": 8285
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9291977408730414e-05,
+      "loss": 1.1237,
+      "step": 8286
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9278598783620337e-05,
+      "loss": 1.1249,
+      "step": 8287
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.926522015851026e-05,
+      "loss": 1.2546,
+      "step": 8288
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9251841533400184e-05,
+      "loss": 1.0798,
+      "step": 8289
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9238462908290107e-05,
+      "loss": 1.1002,
+      "step": 8290
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9225084283180034e-05,
+      "loss": 0.9171,
+      "step": 8291
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9211705658069954e-05,
+      "loss": 1.0274,
+      "step": 8292
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.919832703295988e-05,
+      "loss": 1.0115,
+      "step": 8293
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.91849484078498e-05,
+      "loss": 0.9387,
+      "step": 8294
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9171569782739728e-05,
+      "loss": 1.0728,
+      "step": 8295
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9158191157629648e-05,
+      "loss": 1.058,
+      "step": 8296
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9144812532519574e-05,
+      "loss": 1.0517,
+      "step": 8297
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9131433907409494e-05,
+      "loss": 1.1701,
+      "step": 8298
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9118055282299418e-05,
+      "loss": 0.9979,
+      "step": 8299
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.910467665718934e-05,
+      "loss": 0.9009,
+      "step": 8300
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9091298032079265e-05,
+      "loss": 0.8525,
+      "step": 8301
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9077919406969188e-05,
+      "loss": 0.9516,
+      "step": 8302
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.906454078185911e-05,
+      "loss": 0.9798,
+      "step": 8303
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9051162156749035e-05,
+      "loss": 0.9526,
+      "step": 8304
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9037783531638958e-05,
+      "loss": 0.9293,
+      "step": 8305
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9024404906528885e-05,
+      "loss": 0.994,
+      "step": 8306
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9011026281418805e-05,
+      "loss": 0.9177,
+      "step": 8307
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.8997647656308732e-05,
+      "loss": 1.0131,
+      "step": 8308
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.8984269031198652e-05,
+      "loss": 0.9793,
+      "step": 8309
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.897089040608858e-05,
+      "loss": 0.8678,
+      "step": 8310
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.89575117809785e-05,
+      "loss": 0.8314,
+      "step": 8311
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.8944133155868425e-05,
+      "loss": 0.8073,
+      "step": 8312
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.8930754530758345e-05,
+      "loss": 0.8356,
+      "step": 8313
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.8917375905648272e-05,
+      "loss": 0.846,
+      "step": 8314
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.8903997280538192e-05,
+      "loss": 0.7131,
+      "step": 8315
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.889061865542812e-05,
+      "loss": 0.9412,
+      "step": 8316
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.887724003031804e-05,
+      "loss": 0.7842,
+      "step": 8317
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.8863861405207966e-05,
+      "loss": 0.7548,
+      "step": 8318
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.8850482780097886e-05,
+      "loss": 0.7011,
+      "step": 8319
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.8837104154987812e-05,
+      "loss": 0.5833,
+      "step": 8320
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.8823725529877732e-05,
+      "loss": 0.7141,
+      "step": 8321
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.881034690476766e-05,
+      "loss": 0.7192,
+      "step": 8322
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.8796968279657583e-05,
+      "loss": 0.6554,
+      "step": 8323
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.8783589654547503e-05,
+      "loss": 0.629,
+      "step": 8324
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.877021102943743e-05,
+      "loss": 0.6083,
+      "step": 8325
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.875683240432735e-05,
+      "loss": 0.6606,
+      "step": 8326
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.8743453779217276e-05,
+      "loss": 0.5409,
+      "step": 8327
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.8730075154107196e-05,
+      "loss": 0.5224,
+      "step": 8328
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.8716696528997123e-05,
+      "loss": 0.4628,
+      "step": 8329
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.8703317903887043e-05,
+      "loss": 0.7397,
+      "step": 8330
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.868993927877697e-05,
+      "loss": 1.5484,
+      "step": 8331
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.867656065366689e-05,
+      "loss": 1.3027,
+      "step": 8332
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.8663182028556817e-05,
+      "loss": 1.1458,
+      "step": 8333
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.8649803403446737e-05,
+      "loss": 1.0846,
+      "step": 8334
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.8636424778336663e-05,
+      "loss": 1.3006,
+      "step": 8335
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.8623046153226583e-05,
+      "loss": 1.1116,
+      "step": 8336
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.860966752811651e-05,
+      "loss": 1.0303,
+      "step": 8337
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.859628890300643e-05,
+      "loss": 0.952,
+      "step": 8338
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.8582910277896357e-05,
+      "loss": 1.179,
+      "step": 8339
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.856953165278628e-05,
+      "loss": 1.1206,
+      "step": 8340
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.8556153027676204e-05,
+      "loss": 1.0443,
+      "step": 8341
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.8542774402566127e-05,
+      "loss": 1.1428,
+      "step": 8342
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.852939577745605e-05,
+      "loss": 0.9562,
+      "step": 8343
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.8516017152345974e-05,
+      "loss": 0.9589,
+      "step": 8344
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.8502638527235897e-05,
+      "loss": 1.0118,
+      "step": 8345
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.848925990212582e-05,
+      "loss": 1.0661,
+      "step": 8346
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.8475881277015744e-05,
+      "loss": 1.0502,
+      "step": 8347
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.8462502651905667e-05,
+      "loss": 1.0692,
+      "step": 8348
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.844912402679559e-05,
+      "loss": 1.0297,
+      "step": 8349
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.8435745401685514e-05,
+      "loss": 1.0894,
+      "step": 8350
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.8422366776575434e-05,
+      "loss": 1.0117,
+      "step": 8351
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.840898815146536e-05,
+      "loss": 1.0979,
+      "step": 8352
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.839560952635528e-05,
+      "loss": 0.9128,
+      "step": 8353
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.8382230901245208e-05,
+      "loss": 1.037,
+      "step": 8354
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.836885227613513e-05,
+      "loss": 0.8026,
+      "step": 8355
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.8355473651025054e-05,
+      "loss": 0.9285,
+      "step": 8356
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.8342095025914978e-05,
+      "loss": 0.8885,
+      "step": 8357
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.83287164008049e-05,
+      "loss": 0.9219,
+      "step": 8358
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.8315337775694825e-05,
+      "loss": 0.8406,
+      "step": 8359
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.8301959150584748e-05,
+      "loss": 0.8327,
+      "step": 8360
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.828858052547467e-05,
+      "loss": 0.8848,
+      "step": 8361
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.8275201900364595e-05,
+      "loss": 0.7917,
+      "step": 8362
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.8261823275254518e-05,
+      "loss": 0.8319,
+      "step": 8363
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.824844465014444e-05,
+      "loss": 0.815,
+      "step": 8364
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.8235066025034365e-05,
+      "loss": 0.6846,
+      "step": 8365
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.822168739992429e-05,
+      "loss": 0.666,
+      "step": 8366
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.8208308774814212e-05,
+      "loss": 0.7904,
+      "step": 8367
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.8194930149704135e-05,
+      "loss": 0.8444,
+      "step": 8368
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.818155152459406e-05,
+      "loss": 0.7443,
+      "step": 8369
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.8168172899483982e-05,
+      "loss": 0.7391,
+      "step": 8370
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.8154794274373905e-05,
+      "loss": 0.6805,
+      "step": 8371
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.8141415649263832e-05,
+      "loss": 0.5897,
+      "step": 8372
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.8128037024153752e-05,
+      "loss": 0.6701,
+      "step": 8373
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.811465839904368e-05,
+      "loss": 0.4767,
+      "step": 8374
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.81012797739336e-05,
+      "loss": 0.5054,
+      "step": 8375
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.8087901148823526e-05,
+      "loss": 0.5102,
+      "step": 8376
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.8074522523713446e-05,
+      "loss": 0.6315,
+      "step": 8377
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.806114389860337e-05,
+      "loss": 0.5047,
+      "step": 8378
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.8047765273493292e-05,
+      "loss": 0.5277,
+      "step": 8379
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.8034386648383216e-05,
+      "loss": 0.6215,
+      "step": 8380
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.802100802327314e-05,
+      "loss": 1.3347,
+      "step": 8381
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.8007629398163063e-05,
+      "loss": 1.0705,
+      "step": 8382
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.7994250773052986e-05,
+      "loss": 1.4032,
+      "step": 8383
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.798087214794291e-05,
+      "loss": 1.1722,
+      "step": 8384
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.7967493522832833e-05,
+      "loss": 1.1924,
+      "step": 8385
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.7954114897722756e-05,
+      "loss": 1.1233,
+      "step": 8386
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.794073627261268e-05,
+      "loss": 1.2113,
+      "step": 8387
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.7927357647502603e-05,
+      "loss": 1.0654,
+      "step": 8388
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.791397902239253e-05,
+      "loss": 1.1519,
+      "step": 8389
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.790060039728245e-05,
+      "loss": 1.0743,
+      "step": 8390
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.7887221772172377e-05,
+      "loss": 1.1689,
+      "step": 8391
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.7873843147062297e-05,
+      "loss": 1.0853,
+      "step": 8392
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.7860464521952223e-05,
+      "loss": 0.9015,
+      "step": 8393
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.7847085896842143e-05,
+      "loss": 0.9401,
+      "step": 8394
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.783370727173207e-05,
+      "loss": 1.0713,
+      "step": 8395
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.782032864662199e-05,
+      "loss": 1.0311,
+      "step": 8396
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.7806950021511917e-05,
+      "loss": 0.9842,
+      "step": 8397
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.7793571396401837e-05,
+      "loss": 0.9073,
+      "step": 8398
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.7780192771291764e-05,
+      "loss": 0.7972,
+      "step": 8399
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.7766814146181684e-05,
+      "loss": 0.7863,
+      "step": 8400
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.775343552107161e-05,
+      "loss": 0.9316,
+      "step": 8401
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.774005689596153e-05,
+      "loss": 0.9838,
+      "step": 8402
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.7726678270851457e-05,
+      "loss": 0.9206,
+      "step": 8403
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.771329964574138e-05,
+      "loss": 0.8555,
+      "step": 8404
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.76999210206313e-05,
+      "loss": 0.8746,
+      "step": 8405
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.7686542395521227e-05,
+      "loss": 0.8521,
+      "step": 8406
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.7673163770411147e-05,
+      "loss": 0.8688,
+      "step": 8407
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.7659785145301074e-05,
+      "loss": 0.8612,
+      "step": 8408
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.7646406520190994e-05,
+      "loss": 0.8074,
+      "step": 8409
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.763302789508092e-05,
+      "loss": 0.9919,
+      "step": 8410
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.761964926997084e-05,
+      "loss": 0.6984,
+      "step": 8411
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.7606270644860768e-05,
+      "loss": 0.7824,
+      "step": 8412
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.7592892019750688e-05,
+      "loss": 0.6123,
+      "step": 8413
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.7579513394640614e-05,
+      "loss": 0.8582,
+      "step": 8414
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.7566134769530534e-05,
+      "loss": 0.7839,
+      "step": 8415
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.755275614442046e-05,
+      "loss": 0.659,
+      "step": 8416
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.753937751931038e-05,
+      "loss": 0.7437,
+      "step": 8417
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.7525998894200308e-05,
+      "loss": 0.6491,
+      "step": 8418
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.7512620269090228e-05,
+      "loss": 0.8047,
+      "step": 8419
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.7499241643980155e-05,
+      "loss": 0.7331,
+      "step": 8420
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.7485863018870078e-05,
+      "loss": 0.6305,
+      "step": 8421
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.747248439376e-05,
+      "loss": 0.609,
+      "step": 8422
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.7459105768649925e-05,
+      "loss": 0.6377,
+      "step": 8423
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.744572714353985e-05,
+      "loss": 0.5308,
+      "step": 8424
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.7432348518429772e-05,
+      "loss": 0.5767,
+      "step": 8425
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.7418969893319695e-05,
+      "loss": 0.5565,
+      "step": 8426
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.740559126820962e-05,
+      "loss": 0.5826,
+      "step": 8427
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.7392212643099542e-05,
+      "loss": 0.5192,
+      "step": 8428
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.7378834017989465e-05,
+      "loss": 0.517,
+      "step": 8429
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.7365455392879385e-05,
+      "loss": 0.6176,
+      "step": 8430
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.7352076767769312e-05,
+      "loss": 1.4827,
+      "step": 8431
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.7338698142659232e-05,
+      "loss": 1.4213,
+      "step": 8432
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.732531951754916e-05,
+      "loss": 1.09,
+      "step": 8433
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.731194089243908e-05,
+      "loss": 1.3326,
+      "step": 8434
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.7298562267329006e-05,
+      "loss": 1.1226,
+      "step": 8435
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.7285183642218926e-05,
+      "loss": 1.1902,
+      "step": 8436
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.7271805017108852e-05,
+      "loss": 1.1056,
+      "step": 8437
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.7258426391998776e-05,
+      "loss": 1.1325,
+      "step": 8438
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.72450477668887e-05,
+      "loss": 1.0843,
+      "step": 8439
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.7231669141778623e-05,
+      "loss": 1.0996,
+      "step": 8440
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7218290516668546e-05,
+      "loss": 1.0649,
+      "step": 8441
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.720491189155847e-05,
+      "loss": 0.9621,
+      "step": 8442
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7191533266448393e-05,
+      "loss": 1.0496,
+      "step": 8443
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7178154641338316e-05,
+      "loss": 0.8446,
+      "step": 8444
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.716477601622824e-05,
+      "loss": 1.0894,
+      "step": 8445
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7151397391118163e-05,
+      "loss": 0.9743,
+      "step": 8446
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7138018766008086e-05,
+      "loss": 0.9531,
+      "step": 8447
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.712464014089801e-05,
+      "loss": 1.0576,
+      "step": 8448
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7111261515787933e-05,
+      "loss": 0.9541,
+      "step": 8449
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7097882890677857e-05,
+      "loss": 0.9622,
+      "step": 8450
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.708450426556778e-05,
+      "loss": 0.9424,
+      "step": 8451
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7071125640457703e-05,
+      "loss": 0.9708,
+      "step": 8452
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.705774701534763e-05,
+      "loss": 1.0439,
+      "step": 8453
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.704436839023755e-05,
+      "loss": 0.903,
+      "step": 8454
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7030989765127477e-05,
+      "loss": 0.8482,
+      "step": 8455
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7017611140017397e-05,
+      "loss": 0.9174,
+      "step": 8456
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.700423251490732e-05,
+      "loss": 0.8156,
+      "step": 8457
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.6990853889797244e-05,
+      "loss": 1.1579,
+      "step": 8458
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.6977475264687167e-05,
+      "loss": 0.8391,
+      "step": 8459
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.696409663957709e-05,
+      "loss": 0.8351,
+      "step": 8460
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.6950718014467014e-05,
+      "loss": 0.9053,
+      "step": 8461
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.6937339389356937e-05,
+      "loss": 0.7446,
+      "step": 8462
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.692396076424686e-05,
+      "loss": 0.7558,
+      "step": 8463
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.6910582139136784e-05,
+      "loss": 0.8508,
+      "step": 8464
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.6897203514026707e-05,
+      "loss": 0.8228,
+      "step": 8465
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.688382488891663e-05,
+      "loss": 0.8592,
+      "step": 8466
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.6870446263806554e-05,
+      "loss": 0.6817,
+      "step": 8467
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.6857067638696478e-05,
+      "loss": 0.7973,
+      "step": 8468
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.68436890135864e-05,
+      "loss": 0.6798,
+      "step": 8469
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.6830310388476328e-05,
+      "loss": 0.6935,
+      "step": 8470
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.6816931763366248e-05,
+      "loss": 0.605,
+      "step": 8471
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.6803553138256175e-05,
+      "loss": 0.6602,
+      "step": 8472
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6790174513146095e-05,
+      "loss": 0.727,
+      "step": 8473
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.677679588803602e-05,
+      "loss": 0.6243,
+      "step": 8474
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.676341726292594e-05,
+      "loss": 0.5882,
+      "step": 8475
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6750038637815868e-05,
+      "loss": 0.6304,
+      "step": 8476
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6736660012705788e-05,
+      "loss": 0.4072,
+      "step": 8477
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6723281387595715e-05,
+      "loss": 0.4412,
+      "step": 8478
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6709902762485635e-05,
+      "loss": 0.5193,
+      "step": 8479
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.669652413737556e-05,
+      "loss": 0.5297,
+      "step": 8480
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.668314551226548e-05,
+      "loss": 1.5569,
+      "step": 8481
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.666976688715541e-05,
+      "loss": 1.2644,
+      "step": 8482
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.665638826204533e-05,
+      "loss": 1.1854,
+      "step": 8483
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6643009636935252e-05,
+      "loss": 1.3186,
+      "step": 8484
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6629631011825175e-05,
+      "loss": 1.1768,
+      "step": 8485
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.66162523867151e-05,
+      "loss": 1.15,
+      "step": 8486
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6602873761605025e-05,
+      "loss": 1.2543,
+      "step": 8487
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6589495136494945e-05,
+      "loss": 0.9906,
+      "step": 8488
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6576116511384872e-05,
+      "loss": 0.9892,
+      "step": 8489
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6562737886274792e-05,
+      "loss": 1.3003,
+      "step": 8490
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.654935926116472e-05,
+      "loss": 1.024,
+      "step": 8491
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.653598063605464e-05,
+      "loss": 1.0105,
+      "step": 8492
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6522602010944566e-05,
+      "loss": 0.9664,
+      "step": 8493
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6509223385834486e-05,
+      "loss": 1.1236,
+      "step": 8494
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6495844760724412e-05,
+      "loss": 0.9769,
+      "step": 8495
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6482466135614332e-05,
+      "loss": 0.9316,
+      "step": 8496
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.646908751050426e-05,
+      "loss": 1.0777,
+      "step": 8497
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.645570888539418e-05,
+      "loss": 1.0341,
+      "step": 8498
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6442330260284106e-05,
+      "loss": 0.8937,
+      "step": 8499
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6428951635174026e-05,
+      "loss": 0.9071,
+      "step": 8500
+    },
+    {
+      "epoch": 2.62,
+      "eval_bleu": 5.759184720503681e-14,
+      "eval_loss": 1.68861985206604,
+      "eval_runtime": 2574.1367,
+      "eval_samples_per_second": 5.734,
+      "eval_steps_per_second": 0.717,
+      "step": 8500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6415573010063953e-05,
+      "loss": 0.9315,
+      "step": 8501
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6402194384953876e-05,
+      "loss": 1.0361,
+      "step": 8502
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.63888157598438e-05,
+      "loss": 0.9708,
+      "step": 8503
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6375437134733723e-05,
+      "loss": 1.1955,
+      "step": 8504
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6362058509623646e-05,
+      "loss": 0.8979,
+      "step": 8505
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.634867988451357e-05,
+      "loss": 0.941,
+      "step": 8506
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.6335301259403493e-05,
+      "loss": 0.8597,
+      "step": 8507
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.6321922634293417e-05,
+      "loss": 0.902,
+      "step": 8508
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.630854400918334e-05,
+      "loss": 1.0012,
+      "step": 8509
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.6295165384073263e-05,
+      "loss": 0.8545,
+      "step": 8510
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.6281786758963183e-05,
+      "loss": 0.7875,
+      "step": 8511
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.626840813385311e-05,
+      "loss": 0.7282,
+      "step": 8512
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.625502950874303e-05,
+      "loss": 0.6821,
+      "step": 8513
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.6241650883632957e-05,
+      "loss": 0.8852,
+      "step": 8514
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.6228272258522877e-05,
+      "loss": 0.8281,
+      "step": 8515
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.6214893633412804e-05,
+      "loss": 0.8121,
+      "step": 8516
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.6201515008302724e-05,
+      "loss": 0.7991,
+      "step": 8517
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.618813638319265e-05,
+      "loss": 0.6642,
+      "step": 8518
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.6174757758082574e-05,
+      "loss": 0.6062,
+      "step": 8519
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.6161379132972497e-05,
+      "loss": 0.7109,
+      "step": 8520
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.614800050786242e-05,
+      "loss": 0.5572,
+      "step": 8521
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.6134621882752344e-05,
+      "loss": 0.7335,
+      "step": 8522
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.6121243257642267e-05,
+      "loss": 0.6703,
+      "step": 8523
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.610786463253219e-05,
+      "loss": 0.7494,
+      "step": 8524
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.6094486007422114e-05,
+      "loss": 0.5231,
+      "step": 8525
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.6081107382312038e-05,
+      "loss": 0.5597,
+      "step": 8526
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.606772875720196e-05,
+      "loss": 0.4543,
+      "step": 8527
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.6054350132091884e-05,
+      "loss": 0.4955,
+      "step": 8528
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.6040971506981808e-05,
+      "loss": 0.4458,
+      "step": 8529
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.602759288187173e-05,
+      "loss": 0.4883,
+      "step": 8530
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.6014214256761655e-05,
+      "loss": 1.3296,
+      "step": 8531
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.6000835631651578e-05,
+      "loss": 1.5676,
+      "step": 8532
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.59874570065415e-05,
+      "loss": 1.0863,
+      "step": 8533
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5974078381431425e-05,
+      "loss": 1.1133,
+      "step": 8534
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5960699756321348e-05,
+      "loss": 1.175,
+      "step": 8535
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5947321131211275e-05,
+      "loss": 1.2424,
+      "step": 8536
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5933942506101195e-05,
+      "loss": 0.9384,
+      "step": 8537
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5920563880991118e-05,
+      "loss": 1.018,
+      "step": 8538
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.590718525588104e-05,
+      "loss": 1.1077,
+      "step": 8539
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5893806630770965e-05,
+      "loss": 1.07,
+      "step": 8540
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.588042800566089e-05,
+      "loss": 1.0722,
+      "step": 8541
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5867049380550812e-05,
+      "loss": 1.1063,
+      "step": 8542
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5853670755440735e-05,
+      "loss": 0.9856,
+      "step": 8543
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.584029213033066e-05,
+      "loss": 1.1623,
+      "step": 8544
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5826913505220582e-05,
+      "loss": 1.0399,
+      "step": 8545
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5813534880110505e-05,
+      "loss": 0.9905,
+      "step": 8546
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.580015625500043e-05,
+      "loss": 0.9589,
+      "step": 8547
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5786777629890352e-05,
+      "loss": 0.9489,
+      "step": 8548
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5773399004780276e-05,
+      "loss": 0.9924,
+      "step": 8549
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.57600203796702e-05,
+      "loss": 1.0751,
+      "step": 8550
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5746641754560126e-05,
+      "loss": 0.9276,
+      "step": 8551
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5733263129450046e-05,
+      "loss": 0.9196,
+      "step": 8552
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5719884504339973e-05,
+      "loss": 0.8688,
+      "step": 8553
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5706505879229893e-05,
+      "loss": 1.0456,
+      "step": 8554
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.569312725411982e-05,
+      "loss": 0.9451,
+      "step": 8555
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.567974862900974e-05,
+      "loss": 0.9013,
+      "step": 8556
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5666370003899666e-05,
+      "loss": 0.943,
+      "step": 8557
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5652991378789586e-05,
+      "loss": 0.961,
+      "step": 8558
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5639612753679513e-05,
+      "loss": 0.8554,
+      "step": 8559
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5626234128569433e-05,
+      "loss": 0.8485,
+      "step": 8560
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.561285550345936e-05,
+      "loss": 0.7954,
+      "step": 8561
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.559947687834928e-05,
+      "loss": 0.7791,
+      "step": 8562
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5586098253239203e-05,
+      "loss": 0.8715,
+      "step": 8563
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5572719628129126e-05,
+      "loss": 0.6347,
+      "step": 8564
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.555934100301905e-05,
+      "loss": 0.7544,
+      "step": 8565
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5545962377908973e-05,
+      "loss": 0.8037,
+      "step": 8566
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5532583752798897e-05,
+      "loss": 0.7541,
+      "step": 8567
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5519205127688823e-05,
+      "loss": 0.6553,
+      "step": 8568
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5505826502578743e-05,
+      "loss": 0.7681,
+      "step": 8569
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.549244787746867e-05,
+      "loss": 0.7539,
+      "step": 8570
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.547906925235859e-05,
+      "loss": 0.7226,
+      "step": 8571
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5465690627248517e-05,
+      "loss": 0.5547,
+      "step": 8572
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5452312002138437e-05,
+      "loss": 0.6987,
+      "step": 8573
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5438933377028364e-05,
+      "loss": 0.6176,
+      "step": 8574
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5425554751918284e-05,
+      "loss": 0.6275,
+      "step": 8575
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.541217612680821e-05,
+      "loss": 0.4872,
+      "step": 8576
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5398797501698134e-05,
+      "loss": 0.5302,
+      "step": 8577
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5385418876588057e-05,
+      "loss": 0.4833,
+      "step": 8578
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.537204025147798e-05,
+      "loss": 0.4509,
+      "step": 8579
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5358661626367904e-05,
+      "loss": 0.6693,
+      "step": 8580
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5345283001257827e-05,
+      "loss": 1.6054,
+      "step": 8581
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.533190437614775e-05,
+      "loss": 1.3069,
+      "step": 8582
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5318525751037674e-05,
+      "loss": 1.2814,
+      "step": 8583
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5305147125927598e-05,
+      "loss": 1.1272,
+      "step": 8584
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.529176850081752e-05,
+      "loss": 1.0989,
+      "step": 8585
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5278389875707444e-05,
+      "loss": 1.1148,
+      "step": 8586
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5265011250597368e-05,
+      "loss": 1.0261,
+      "step": 8587
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.525163262548729e-05,
+      "loss": 1.1557,
+      "step": 8588
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5238254000377213e-05,
+      "loss": 1.1343,
+      "step": 8589
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5224875375267136e-05,
+      "loss": 1.0687,
+      "step": 8590
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.521149675015706e-05,
+      "loss": 0.9971,
+      "step": 8591
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5198118125046985e-05,
+      "loss": 1.1088,
+      "step": 8592
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5184739499936908e-05,
+      "loss": 0.9794,
+      "step": 8593
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5171360874826832e-05,
+      "loss": 1.1495,
+      "step": 8594
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5157982249716755e-05,
+      "loss": 0.957,
+      "step": 8595
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5144603624606678e-05,
+      "loss": 1.1194,
+      "step": 8596
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5131224999496602e-05,
+      "loss": 1.0374,
+      "step": 8597
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5117846374386525e-05,
+      "loss": 1.0362,
+      "step": 8598
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5104467749276448e-05,
+      "loss": 0.9842,
+      "step": 8599
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5091089124166372e-05,
+      "loss": 0.9756,
+      "step": 8600
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5077710499056295e-05,
+      "loss": 0.9702,
+      "step": 8601
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5064331873946217e-05,
+      "loss": 0.8326,
+      "step": 8602
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.505095324883614e-05,
+      "loss": 1.0562,
+      "step": 8603
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5037574623726064e-05,
+      "loss": 0.9138,
+      "step": 8604
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5024195998615987e-05,
+      "loss": 0.9092,
+      "step": 8605
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.501081737350591e-05,
+      "loss": 0.8913,
+      "step": 8606
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4997438748395834e-05,
+      "loss": 0.7979,
+      "step": 8607
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4984060123285757e-05,
+      "loss": 0.9911,
+      "step": 8608
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4970681498175682e-05,
+      "loss": 0.9219,
+      "step": 8609
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4957302873065606e-05,
+      "loss": 0.9257,
+      "step": 8610
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4943924247955529e-05,
+      "loss": 1.068,
+      "step": 8611
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4930545622845453e-05,
+      "loss": 0.6279,
+      "step": 8612
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4917166997735376e-05,
+      "loss": 0.7702,
+      "step": 8613
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.49037883726253e-05,
+      "loss": 0.6938,
+      "step": 8614
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4890409747515223e-05,
+      "loss": 0.7169,
+      "step": 8615
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4877031122405146e-05,
+      "loss": 0.7115,
+      "step": 8616
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.486365249729507e-05,
+      "loss": 0.6379,
+      "step": 8617
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4850273872184993e-05,
+      "loss": 0.6533,
+      "step": 8618
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4836895247074916e-05,
+      "loss": 0.6867,
+      "step": 8619
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.482351662196484e-05,
+      "loss": 0.674,
+      "step": 8620
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4810137996854763e-05,
+      "loss": 0.6002,
+      "step": 8621
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4796759371744686e-05,
+      "loss": 0.656,
+      "step": 8622
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.478338074663461e-05,
+      "loss": 0.5932,
+      "step": 8623
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4770002121524533e-05,
+      "loss": 0.5057,
+      "step": 8624
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4756623496414458e-05,
+      "loss": 0.6757,
+      "step": 8625
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4743244871304382e-05,
+      "loss": 0.5777,
+      "step": 8626
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4729866246194305e-05,
+      "loss": 0.4736,
+      "step": 8627
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4716487621084228e-05,
+      "loss": 0.4983,
+      "step": 8628
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.470310899597415e-05,
+      "loss": 0.5027,
+      "step": 8629
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4689730370864074e-05,
+      "loss": 0.5555,
+      "step": 8630
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4676351745753997e-05,
+      "loss": 1.3254,
+      "step": 8631
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.466297312064392e-05,
+      "loss": 1.3215,
+      "step": 8632
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4649594495533844e-05,
+      "loss": 1.1211,
+      "step": 8633
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4636215870423767e-05,
+      "loss": 1.1064,
+      "step": 8634
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.462283724531369e-05,
+      "loss": 1.0268,
+      "step": 8635
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4609458620203614e-05,
+      "loss": 1.0091,
+      "step": 8636
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4596079995093537e-05,
+      "loss": 1.1716,
+      "step": 8637
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.458270136998346e-05,
+      "loss": 1.0217,
+      "step": 8638
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4569322744873384e-05,
+      "loss": 0.9986,
+      "step": 8639
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4555944119763307e-05,
+      "loss": 1.0729,
+      "step": 8640
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4542565494653233e-05,
+      "loss": 1.2474,
+      "step": 8641
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4529186869543156e-05,
+      "loss": 1.041,
+      "step": 8642
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.451580824443308e-05,
+      "loss": 1.0498,
+      "step": 8643
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4502429619323003e-05,
+      "loss": 0.8664,
+      "step": 8644
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4489050994212926e-05,
+      "loss": 1.0998,
+      "step": 8645
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.447567236910285e-05,
+      "loss": 1.1825,
+      "step": 8646
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4462293743992773e-05,
+      "loss": 0.9566,
+      "step": 8647
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4448915118882696e-05,
+      "loss": 0.9924,
+      "step": 8648
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.443553649377262e-05,
+      "loss": 0.8778,
+      "step": 8649
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4422157868662543e-05,
+      "loss": 1.0033,
+      "step": 8650
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4408779243552466e-05,
+      "loss": 1.0275,
+      "step": 8651
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.439540061844239e-05,
+      "loss": 0.8168,
+      "step": 8652
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4382021993332313e-05,
+      "loss": 0.8185,
+      "step": 8653
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4368643368222237e-05,
+      "loss": 0.9611,
+      "step": 8654
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4355264743112158e-05,
+      "loss": 1.0267,
+      "step": 8655
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4341886118002082e-05,
+      "loss": 0.9893,
+      "step": 8656
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4328507492892005e-05,
+      "loss": 0.849,
+      "step": 8657
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.431512886778193e-05,
+      "loss": 0.774,
+      "step": 8658
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4301750242671854e-05,
+      "loss": 0.8761,
+      "step": 8659
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4288371617561777e-05,
+      "loss": 0.8838,
+      "step": 8660
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.42749929924517e-05,
+      "loss": 0.8223,
+      "step": 8661
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4261614367341624e-05,
+      "loss": 0.7164,
+      "step": 8662
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4248235742231547e-05,
+      "loss": 0.7901,
+      "step": 8663
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.423485711712147e-05,
+      "loss": 0.7677,
+      "step": 8664
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4221478492011394e-05,
+      "loss": 0.8244,
+      "step": 8665
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4208099866901317e-05,
+      "loss": 0.7818,
+      "step": 8666
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.419472124179124e-05,
+      "loss": 0.6478,
+      "step": 8667
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4181342616681164e-05,
+      "loss": 0.6649,
+      "step": 8668
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4167963991571087e-05,
+      "loss": 0.7252,
+      "step": 8669
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4154585366461011e-05,
+      "loss": 0.7254,
+      "step": 8670
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4141206741350934e-05,
+      "loss": 0.5109,
+      "step": 8671
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4127828116240858e-05,
+      "loss": 0.7524,
+      "step": 8672
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4114449491130781e-05,
+      "loss": 0.6431,
+      "step": 8673
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4101070866020706e-05,
+      "loss": 0.6703,
+      "step": 8674
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.408769224091063e-05,
+      "loss": 0.5684,
+      "step": 8675
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4074313615800553e-05,
+      "loss": 0.4999,
+      "step": 8676
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4060934990690476e-05,
+      "loss": 0.5227,
+      "step": 8677
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.40475563655804e-05,
+      "loss": 0.4725,
+      "step": 8678
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4034177740470323e-05,
+      "loss": 0.5186,
+      "step": 8679
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4020799115360246e-05,
+      "loss": 0.6301,
+      "step": 8680
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.400742049025017e-05,
+      "loss": 1.3825,
+      "step": 8681
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3994041865140092e-05,
+      "loss": 1.4368,
+      "step": 8682
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3980663240030015e-05,
+      "loss": 1.2381,
+      "step": 8683
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3967284614919938e-05,
+      "loss": 1.2041,
+      "step": 8684
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3953905989809862e-05,
+      "loss": 1.0633,
+      "step": 8685
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3940527364699785e-05,
+      "loss": 1.0566,
+      "step": 8686
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3927148739589708e-05,
+      "loss": 0.9515,
+      "step": 8687
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3913770114479632e-05,
+      "loss": 1.0993,
+      "step": 8688
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3900391489369555e-05,
+      "loss": 1.12,
+      "step": 8689
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.388701286425948e-05,
+      "loss": 0.991,
+      "step": 8690
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3873634239149404e-05,
+      "loss": 0.9008,
+      "step": 8691
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3860255614039327e-05,
+      "loss": 1.077,
+      "step": 8692
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.384687698892925e-05,
+      "loss": 0.9926,
+      "step": 8693
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3833498363819174e-05,
+      "loss": 1.0028,
+      "step": 8694
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3820119738709097e-05,
+      "loss": 1.1002,
+      "step": 8695
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.380674111359902e-05,
+      "loss": 0.8357,
+      "step": 8696
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3793362488488944e-05,
+      "loss": 1.0453,
+      "step": 8697
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3779983863378867e-05,
+      "loss": 1.0354,
+      "step": 8698
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3766605238268791e-05,
+      "loss": 0.9096,
+      "step": 8699
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3753226613158714e-05,
+      "loss": 1.0014,
+      "step": 8700
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3739847988048638e-05,
+      "loss": 1.07,
+      "step": 8701
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3726469362938561e-05,
+      "loss": 1.0748,
+      "step": 8702
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3713090737828484e-05,
+      "loss": 0.9144,
+      "step": 8703
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3699712112718408e-05,
+      "loss": 1.064,
+      "step": 8704
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3686333487608331e-05,
+      "loss": 0.9574,
+      "step": 8705
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3672954862498255e-05,
+      "loss": 0.8693,
+      "step": 8706
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.365957623738818e-05,
+      "loss": 0.9202,
+      "step": 8707
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3646197612278103e-05,
+      "loss": 0.7506,
+      "step": 8708
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3632818987168025e-05,
+      "loss": 0.8331,
+      "step": 8709
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3619440362057948e-05,
+      "loss": 0.8177,
+      "step": 8710
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3606061736947872e-05,
+      "loss": 0.7095,
+      "step": 8711
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3592683111837795e-05,
+      "loss": 0.7961,
+      "step": 8712
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3579304486727718e-05,
+      "loss": 0.7623,
+      "step": 8713
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3565925861617642e-05,
+      "loss": 0.8185,
+      "step": 8714
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3552547236507565e-05,
+      "loss": 0.7849,
+      "step": 8715
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3539168611397488e-05,
+      "loss": 0.6551,
+      "step": 8716
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3525789986287412e-05,
+      "loss": 0.6666,
+      "step": 8717
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3512411361177335e-05,
+      "loss": 0.6032,
+      "step": 8718
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3499032736067259e-05,
+      "loss": 0.657,
+      "step": 8719
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3485654110957182e-05,
+      "loss": 0.6704,
+      "step": 8720
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3472275485847105e-05,
+      "loss": 0.6741,
+      "step": 8721
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3458896860737029e-05,
+      "loss": 0.6432,
+      "step": 8722
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3445518235626954e-05,
+      "loss": 0.6303,
+      "step": 8723
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3432139610516877e-05,
+      "loss": 0.5445,
+      "step": 8724
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.34187609854068e-05,
+      "loss": 0.624,
+      "step": 8725
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3405382360296724e-05,
+      "loss": 0.5777,
+      "step": 8726
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3392003735186647e-05,
+      "loss": 0.5064,
+      "step": 8727
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3378625110076571e-05,
+      "loss": 0.5206,
+      "step": 8728
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3365246484966494e-05,
+      "loss": 0.5228,
+      "step": 8729
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3351867859856418e-05,
+      "loss": 0.6471,
+      "step": 8730
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3338489234746341e-05,
+      "loss": 1.3698,
+      "step": 8731
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.3325110609636264e-05,
+      "loss": 1.2194,
+      "step": 8732
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.3311731984526188e-05,
+      "loss": 1.2988,
+      "step": 8733
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.3298353359416111e-05,
+      "loss": 1.1996,
+      "step": 8734
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.3284974734306033e-05,
+      "loss": 1.2027,
+      "step": 8735
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.3271596109195956e-05,
+      "loss": 0.9923,
+      "step": 8736
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.325821748408588e-05,
+      "loss": 1.0943,
+      "step": 8737
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.3244838858975803e-05,
+      "loss": 0.9867,
+      "step": 8738
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.3231460233865728e-05,
+      "loss": 1.2417,
+      "step": 8739
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.3218081608755652e-05,
+      "loss": 1.1227,
+      "step": 8740
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.3204702983645575e-05,
+      "loss": 1.0044,
+      "step": 8741
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.3191324358535498e-05,
+      "loss": 1.1909,
+      "step": 8742
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.3177945733425422e-05,
+      "loss": 1.0723,
+      "step": 8743
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.3164567108315345e-05,
+      "loss": 0.9338,
+      "step": 8744
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.3151188483205268e-05,
+      "loss": 0.8722,
+      "step": 8745
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.3137809858095192e-05,
+      "loss": 1.1632,
+      "step": 8746
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.3124431232985115e-05,
+      "loss": 0.9014,
+      "step": 8747
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.3111052607875039e-05,
+      "loss": 0.8824,
+      "step": 8748
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.3097673982764962e-05,
+      "loss": 1.007,
+      "step": 8749
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.3084295357654885e-05,
+      "loss": 0.9594,
+      "step": 8750
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.3070916732544809e-05,
+      "loss": 0.884,
+      "step": 8751
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.3057538107434732e-05,
+      "loss": 0.8127,
+      "step": 8752
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.3044159482324656e-05,
+      "loss": 0.8978,
+      "step": 8753
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.3030780857214579e-05,
+      "loss": 0.9009,
+      "step": 8754
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.3017402232104502e-05,
+      "loss": 0.903,
+      "step": 8755
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.3004023606994427e-05,
+      "loss": 1.0421,
+      "step": 8756
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2990644981884351e-05,
+      "loss": 1.0953,
+      "step": 8757
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2977266356774274e-05,
+      "loss": 0.9625,
+      "step": 8758
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2963887731664198e-05,
+      "loss": 0.9055,
+      "step": 8759
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2950509106554121e-05,
+      "loss": 0.8562,
+      "step": 8760
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2937130481444044e-05,
+      "loss": 0.7562,
+      "step": 8761
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2923751856333966e-05,
+      "loss": 0.7984,
+      "step": 8762
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.291037323122389e-05,
+      "loss": 0.8009,
+      "step": 8763
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2896994606113813e-05,
+      "loss": 0.7612,
+      "step": 8764
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2883615981003736e-05,
+      "loss": 0.7619,
+      "step": 8765
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.287023735589366e-05,
+      "loss": 0.7413,
+      "step": 8766
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2856858730783583e-05,
+      "loss": 0.739,
+      "step": 8767
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2843480105673506e-05,
+      "loss": 0.633,
+      "step": 8768
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.283010148056343e-05,
+      "loss": 0.5059,
+      "step": 8769
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2816722855453353e-05,
+      "loss": 0.8026,
+      "step": 8770
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2803344230343277e-05,
+      "loss": 0.5841,
+      "step": 8771
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2789965605233202e-05,
+      "loss": 0.7607,
+      "step": 8772
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2776586980123125e-05,
+      "loss": 0.5575,
+      "step": 8773
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2763208355013049e-05,
+      "loss": 0.648,
+      "step": 8774
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2749829729902972e-05,
+      "loss": 0.5728,
+      "step": 8775
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2736451104792895e-05,
+      "loss": 0.6196,
+      "step": 8776
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2723072479682819e-05,
+      "loss": 0.4439,
+      "step": 8777
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2709693854572742e-05,
+      "loss": 0.5642,
+      "step": 8778
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2696315229462665e-05,
+      "loss": 0.5713,
+      "step": 8779
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2682936604352589e-05,
+      "loss": 0.546,
+      "step": 8780
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2669557979242512e-05,
+      "loss": 1.5102,
+      "step": 8781
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2656179354132436e-05,
+      "loss": 1.3157,
+      "step": 8782
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2642800729022359e-05,
+      "loss": 1.2232,
+      "step": 8783
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2629422103912282e-05,
+      "loss": 1.0862,
+      "step": 8784
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2616043478802206e-05,
+      "loss": 1.2067,
+      "step": 8785
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.260266485369213e-05,
+      "loss": 1.1688,
+      "step": 8786
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2589286228582053e-05,
+      "loss": 0.9585,
+      "step": 8787
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2575907603471974e-05,
+      "loss": 1.1835,
+      "step": 8788
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.25625289783619e-05,
+      "loss": 1.1614,
+      "step": 8789
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2549150353251823e-05,
+      "loss": 1.0056,
+      "step": 8790
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2535771728141746e-05,
+      "loss": 0.9466,
+      "step": 8791
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.252239310303167e-05,
+      "loss": 1.034,
+      "step": 8792
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2509014477921593e-05,
+      "loss": 0.8654,
+      "step": 8793
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2495635852811516e-05,
+      "loss": 1.0166,
+      "step": 8794
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.248225722770144e-05,
+      "loss": 1.0786,
+      "step": 8795
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.2468878602591363e-05,
+      "loss": 1.0786,
+      "step": 8796
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.2455499977481286e-05,
+      "loss": 0.9696,
+      "step": 8797
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.244212135237121e-05,
+      "loss": 0.9565,
+      "step": 8798
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.2428742727261133e-05,
+      "loss": 0.8754,
+      "step": 8799
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.2415364102151057e-05,
+      "loss": 0.9407,
+      "step": 8800
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.240198547704098e-05,
+      "loss": 0.9022,
+      "step": 8801
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.2388606851930903e-05,
+      "loss": 0.9116,
+      "step": 8802
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.2375228226820827e-05,
+      "loss": 0.7722,
+      "step": 8803
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.236184960171075e-05,
+      "loss": 0.7976,
+      "step": 8804
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.2348470976600675e-05,
+      "loss": 0.9696,
+      "step": 8805
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.2335092351490599e-05,
+      "loss": 0.8553,
+      "step": 8806
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.2321713726380522e-05,
+      "loss": 0.8848,
+      "step": 8807
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.2308335101270445e-05,
+      "loss": 0.7232,
+      "step": 8808
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.2294956476160369e-05,
+      "loss": 0.6822,
+      "step": 8809
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.2281577851050292e-05,
+      "loss": 0.8793,
+      "step": 8810
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.2268199225940216e-05,
+      "loss": 0.881,
+      "step": 8811
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.2254820600830139e-05,
+      "loss": 0.9418,
+      "step": 8812
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.2241441975720062e-05,
+      "loss": 0.8817,
+      "step": 8813
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.2228063350609984e-05,
+      "loss": 0.7054,
+      "step": 8814
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.2214684725499908e-05,
+      "loss": 0.904,
+      "step": 8815
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.2201306100389831e-05,
+      "loss": 0.7248,
+      "step": 8816
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.2187927475279754e-05,
+      "loss": 0.5779,
+      "step": 8817
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.2174548850169678e-05,
+      "loss": 0.6716,
+      "step": 8818
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.2161170225059601e-05,
+      "loss": 0.7048,
+      "step": 8819
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.2147791599949524e-05,
+      "loss": 0.5127,
+      "step": 8820
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.213441297483945e-05,
+      "loss": 0.6851,
+      "step": 8821
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.2121034349729373e-05,
+      "loss": 0.7613,
+      "step": 8822
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.2107655724619296e-05,
+      "loss": 0.5002,
+      "step": 8823
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.209427709950922e-05,
+      "loss": 0.617,
+      "step": 8824
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.2080898474399143e-05,
+      "loss": 0.7265,
+      "step": 8825
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.2067519849289066e-05,
+      "loss": 0.5002,
+      "step": 8826
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.205414122417899e-05,
+      "loss": 0.4089,
+      "step": 8827
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.2040762599068913e-05,
+      "loss": 0.5289,
+      "step": 8828
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.2027383973958837e-05,
+      "loss": 0.605,
+      "step": 8829
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.201400534884876e-05,
+      "loss": 0.4683,
+      "step": 8830
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.2000626723738683e-05,
+      "loss": 1.5986,
+      "step": 8831
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.1987248098628607e-05,
+      "loss": 1.4041,
+      "step": 8832
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.197386947351853e-05,
+      "loss": 1.2688,
+      "step": 8833
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.1960490848408454e-05,
+      "loss": 0.9548,
+      "step": 8834
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.1947112223298377e-05,
+      "loss": 1.0186,
+      "step": 8835
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.19337335981883e-05,
+      "loss": 0.9477,
+      "step": 8836
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.1920354973078224e-05,
+      "loss": 1.1712,
+      "step": 8837
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.1906976347968149e-05,
+      "loss": 1.0721,
+      "step": 8838
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.1893597722858072e-05,
+      "loss": 0.9499,
+      "step": 8839
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.1880219097747996e-05,
+      "loss": 1.2151,
+      "step": 8840
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.1866840472637917e-05,
+      "loss": 1.114,
+      "step": 8841
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.185346184752784e-05,
+      "loss": 1.1057,
+      "step": 8842
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.1840083222417764e-05,
+      "loss": 0.9799,
+      "step": 8843
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.1826704597307688e-05,
+      "loss": 0.9041,
+      "step": 8844
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.1813325972197611e-05,
+      "loss": 1.0809,
+      "step": 8845
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.1799947347087534e-05,
+      "loss": 1.0279,
+      "step": 8846
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.1786568721977458e-05,
+      "loss": 0.9377,
+      "step": 8847
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.1773190096867381e-05,
+      "loss": 1.022,
+      "step": 8848
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.1759811471757304e-05,
+      "loss": 0.9002,
+      "step": 8849
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.1746432846647228e-05,
+      "loss": 1.0422,
+      "step": 8850
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.1733054221537151e-05,
+      "loss": 0.9713,
+      "step": 8851
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.1719675596427075e-05,
+      "loss": 0.8212,
+      "step": 8852
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.1706296971316998e-05,
+      "loss": 1.0012,
+      "step": 8853
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.1692918346206923e-05,
+      "loss": 0.9809,
+      "step": 8854
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.1679539721096846e-05,
+      "loss": 0.7417,
+      "step": 8855
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.166616109598677e-05,
+      "loss": 0.8263,
+      "step": 8856
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.1652782470876693e-05,
+      "loss": 0.8692,
+      "step": 8857
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.1639403845766617e-05,
+      "loss": 0.9974,
+      "step": 8858
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.162602522065654e-05,
+      "loss": 0.8262,
+      "step": 8859
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.1612646595546463e-05,
+      "loss": 0.811,
+      "step": 8860
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.1599267970436387e-05,
+      "loss": 0.8567,
+      "step": 8861
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.158588934532631e-05,
+      "loss": 0.8572,
+      "step": 8862
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.1572510720216234e-05,
+      "loss": 0.9331,
+      "step": 8863
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.1559132095106157e-05,
+      "loss": 0.6819,
+      "step": 8864
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.154575346999608e-05,
+      "loss": 0.8439,
+      "step": 8865
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.1532374844886004e-05,
+      "loss": 0.803,
+      "step": 8866
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.1518996219775925e-05,
+      "loss": 0.7622,
+      "step": 8867
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.1505617594665849e-05,
+      "loss": 0.7674,
+      "step": 8868
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.1492238969555772e-05,
+      "loss": 0.6295,
+      "step": 8869
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.1478860344445697e-05,
+      "loss": 0.5767,
+      "step": 8870
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.146548171933562e-05,
+      "loss": 0.6559,
+      "step": 8871
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.1452103094225544e-05,
+      "loss": 0.5666,
+      "step": 8872
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.1438724469115468e-05,
+      "loss": 0.5901,
+      "step": 8873
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.1425345844005391e-05,
+      "loss": 0.5741,
+      "step": 8874
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.1411967218895314e-05,
+      "loss": 0.4439,
+      "step": 8875
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.1398588593785238e-05,
+      "loss": 0.5724,
+      "step": 8876
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.1385209968675161e-05,
+      "loss": 0.5535,
+      "step": 8877
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.1371831343565084e-05,
+      "loss": 0.4144,
+      "step": 8878
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.1358452718455008e-05,
+      "loss": 0.5135,
+      "step": 8879
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.1345074093344931e-05,
+      "loss": 0.5016,
+      "step": 8880
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.1331695468234855e-05,
+      "loss": 1.4911,
+      "step": 8881
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.1318316843124778e-05,
+      "loss": 1.3381,
+      "step": 8882
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.1304938218014701e-05,
+      "loss": 1.1749,
+      "step": 8883
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.1291559592904625e-05,
+      "loss": 1.1259,
+      "step": 8884
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.1278180967794548e-05,
+      "loss": 1.155,
+      "step": 8885
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.1264802342684472e-05,
+      "loss": 1.0587,
+      "step": 8886
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.1251423717574397e-05,
+      "loss": 1.2168,
+      "step": 8887
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.123804509246432e-05,
+      "loss": 1.0892,
+      "step": 8888
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.1224666467354243e-05,
+      "loss": 1.0268,
+      "step": 8889
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.1211287842244167e-05,
+      "loss": 0.9207,
+      "step": 8890
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.119790921713409e-05,
+      "loss": 0.9331,
+      "step": 8891
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.1184530592024014e-05,
+      "loss": 1.1158,
+      "step": 8892
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.1171151966913937e-05,
+      "loss": 1.0122,
+      "step": 8893
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1157773341803859e-05,
+      "loss": 1.0589,
+      "step": 8894
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1144394716693782e-05,
+      "loss": 1.1209,
+      "step": 8895
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1131016091583705e-05,
+      "loss": 1.1425,
+      "step": 8896
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1117637466473629e-05,
+      "loss": 1.0852,
+      "step": 8897
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1104258841363552e-05,
+      "loss": 1.053,
+      "step": 8898
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1090880216253476e-05,
+      "loss": 1.0032,
+      "step": 8899
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1077501591143399e-05,
+      "loss": 1.0804,
+      "step": 8900
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1064122966033322e-05,
+      "loss": 0.9631,
+      "step": 8901
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1050744340923246e-05,
+      "loss": 0.9947,
+      "step": 8902
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1037365715813171e-05,
+      "loss": 0.78,
+      "step": 8903
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1023987090703094e-05,
+      "loss": 1.0233,
+      "step": 8904
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.1010608465593018e-05,
+      "loss": 0.873,
+      "step": 8905
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.0997229840482941e-05,
+      "loss": 1.0262,
+      "step": 8906
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.0983851215372864e-05,
+      "loss": 0.963,
+      "step": 8907
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.0970472590262788e-05,
+      "loss": 0.7826,
+      "step": 8908
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.0957093965152711e-05,
+      "loss": 0.9447,
+      "step": 8909
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.0943715340042635e-05,
+      "loss": 0.8211,
+      "step": 8910
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.0930336714932558e-05,
+      "loss": 0.8164,
+      "step": 8911
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.0916958089822481e-05,
+      "loss": 0.7702,
+      "step": 8912
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.0903579464712405e-05,
+      "loss": 0.6934,
+      "step": 8913
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.0890200839602328e-05,
+      "loss": 0.6868,
+      "step": 8914
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.0876822214492252e-05,
+      "loss": 0.77,
+      "step": 8915
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.0863443589382175e-05,
+      "loss": 0.8499,
+      "step": 8916
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.0850064964272098e-05,
+      "loss": 0.6125,
+      "step": 8917
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.0836686339162022e-05,
+      "loss": 0.7259,
+      "step": 8918
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.0823307714051947e-05,
+      "loss": 0.7535,
+      "step": 8919
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.080992908894187e-05,
+      "loss": 0.6638,
+      "step": 8920
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.0796550463831792e-05,
+      "loss": 0.5336,
+      "step": 8921
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.0783171838721715e-05,
+      "loss": 0.6104,
+      "step": 8922
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.0769793213611639e-05,
+      "loss": 0.5593,
+      "step": 8923
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.0756414588501562e-05,
+      "loss": 0.5202,
+      "step": 8924
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.0743035963391486e-05,
+      "loss": 0.4963,
+      "step": 8925
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.0729657338281409e-05,
+      "loss": 0.5194,
+      "step": 8926
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.0716278713171332e-05,
+      "loss": 0.4963,
+      "step": 8927
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.0702900088061256e-05,
+      "loss": 0.5723,
+      "step": 8928
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.0689521462951179e-05,
+      "loss": 0.4489,
+      "step": 8929
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.0676142837841102e-05,
+      "loss": 0.5179,
+      "step": 8930
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.0662764212731026e-05,
+      "loss": 1.4987,
+      "step": 8931
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.064938558762095e-05,
+      "loss": 1.2797,
+      "step": 8932
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.0636006962510873e-05,
+      "loss": 1.235,
+      "step": 8933
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.0622628337400796e-05,
+      "loss": 1.0294,
+      "step": 8934
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.060924971229072e-05,
+      "loss": 1.1159,
+      "step": 8935
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.0595871087180644e-05,
+      "loss": 1.1278,
+      "step": 8936
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.0582492462070568e-05,
+      "loss": 1.0077,
+      "step": 8937
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.0569113836960491e-05,
+      "loss": 1.0839,
+      "step": 8938
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.0555735211850415e-05,
+      "loss": 0.8558,
+      "step": 8939
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.0542356586740338e-05,
+      "loss": 1.0913,
+      "step": 8940
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.0528977961630261e-05,
+      "loss": 1.0879,
+      "step": 8941
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.0515599336520185e-05,
+      "loss": 0.969,
+      "step": 8942
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.0502220711410108e-05,
+      "loss": 1.0312,
+      "step": 8943
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.0488842086300032e-05,
+      "loss": 1.0353,
+      "step": 8944
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.0475463461189955e-05,
+      "loss": 0.9928,
+      "step": 8945
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.0462084836079878e-05,
+      "loss": 1.0136,
+      "step": 8946
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.04487062109698e-05,
+      "loss": 0.8975,
+      "step": 8947
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.0435327585859723e-05,
+      "loss": 1.0114,
+      "step": 8948
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.0421948960749647e-05,
+      "loss": 0.8295,
+      "step": 8949
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.040857033563957e-05,
+      "loss": 0.9239,
+      "step": 8950
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.0395191710529494e-05,
+      "loss": 0.9062,
+      "step": 8951
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.0381813085419419e-05,
+      "loss": 0.9273,
+      "step": 8952
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.0368434460309342e-05,
+      "loss": 0.921,
+      "step": 8953
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.0355055835199266e-05,
+      "loss": 0.8247,
+      "step": 8954
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.0341677210089189e-05,
+      "loss": 0.896,
+      "step": 8955
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.0328298584979112e-05,
+      "loss": 0.9014,
+      "step": 8956
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.0314919959869036e-05,
+      "loss": 0.9501,
+      "step": 8957
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.0301541334758959e-05,
+      "loss": 0.848,
+      "step": 8958
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.0288162709648882e-05,
+      "loss": 0.851,
+      "step": 8959
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.0274784084538806e-05,
+      "loss": 0.8766,
+      "step": 8960
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.026140545942873e-05,
+      "loss": 0.7941,
+      "step": 8961
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.0248026834318653e-05,
+      "loss": 0.9241,
+      "step": 8962
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.0234648209208576e-05,
+      "loss": 1.0805,
+      "step": 8963
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.02212695840985e-05,
+      "loss": 0.7361,
+      "step": 8964
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.0207890958988423e-05,
+      "loss": 0.7376,
+      "step": 8965
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.0194512333878346e-05,
+      "loss": 0.73,
+      "step": 8966
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.018113370876827e-05,
+      "loss": 0.8277,
+      "step": 8967
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.0167755083658195e-05,
+      "loss": 0.7244,
+      "step": 8968
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.0154376458548118e-05,
+      "loss": 0.6372,
+      "step": 8969
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.0140997833438041e-05,
+      "loss": 0.6766,
+      "step": 8970
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.0127619208327965e-05,
+      "loss": 0.6968,
+      "step": 8971
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.0114240583217888e-05,
+      "loss": 0.6922,
+      "step": 8972
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.010086195810781e-05,
+      "loss": 0.6429,
+      "step": 8973
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.0087483332997733e-05,
+      "loss": 0.6018,
+      "step": 8974
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.0074104707887657e-05,
+      "loss": 0.4326,
+      "step": 8975
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.006072608277758e-05,
+      "loss": 0.5196,
+      "step": 8976
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.0047347457667503e-05,
+      "loss": 0.5291,
+      "step": 8977
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.0033968832557427e-05,
+      "loss": 0.6146,
+      "step": 8978
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.002059020744735e-05,
+      "loss": 0.4015,
+      "step": 8979
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.0007211582337274e-05,
+      "loss": 0.5272,
+      "step": 8980
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.993832957227197e-06,
+      "loss": 1.319,
+      "step": 8981
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.98045433211712e-06,
+      "loss": 1.2415,
+      "step": 8982
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.967075707007044e-06,
+      "loss": 1.3598,
+      "step": 8983
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.953697081896967e-06,
+      "loss": 1.1866,
+      "step": 8984
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.940318456786892e-06,
+      "loss": 1.2456,
+      "step": 8985
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.926939831676816e-06,
+      "loss": 0.9064,
+      "step": 8986
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.913561206566739e-06,
+      "loss": 1.3214,
+      "step": 8987
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.900182581456662e-06,
+      "loss": 0.8784,
+      "step": 8988
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.886803956346586e-06,
+      "loss": 1.0821,
+      "step": 8989
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.87342533123651e-06,
+      "loss": 0.8237,
+      "step": 8990
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.860046706126433e-06,
+      "loss": 1.066,
+      "step": 8991
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.846668081016356e-06,
+      "loss": 1.0576,
+      "step": 8992
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.83328945590628e-06,
+      "loss": 1.0039,
+      "step": 8993
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.819910830796203e-06,
+      "loss": 0.9692,
+      "step": 8994
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.806532205686126e-06,
+      "loss": 1.1192,
+      "step": 8995
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.79315358057605e-06,
+      "loss": 0.9501,
+      "step": 8996
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.779774955465973e-06,
+      "loss": 0.8108,
+      "step": 8997
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.766396330355896e-06,
+      "loss": 1.064,
+      "step": 8998
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.75301770524582e-06,
+      "loss": 0.9186,
+      "step": 8999
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.739639080135741e-06,
+      "loss": 0.9735,
+      "step": 9000
+    },
+    {
+      "epoch": 2.78,
+      "eval_bleu": 5.1478661845701255e-14,
+      "eval_loss": 1.6609143018722534,
+      "eval_runtime": 2612.2734,
+      "eval_samples_per_second": 5.65,
+      "eval_steps_per_second": 0.706,
+      "step": 9000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.726260455025667e-06,
+      "loss": 0.8773,
+      "step": 9001
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.71288182991559e-06,
+      "loss": 0.7978,
+      "step": 9002
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.699503204805513e-06,
+      "loss": 1.0432,
+      "step": 9003
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.686124579695437e-06,
+      "loss": 0.8124,
+      "step": 9004
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.67274595458536e-06,
+      "loss": 0.9717,
+      "step": 9005
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.659367329475283e-06,
+      "loss": 0.7882,
+      "step": 9006
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.645988704365207e-06,
+      "loss": 0.7777,
+      "step": 9007
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.63261007925513e-06,
+      "loss": 0.6951,
+      "step": 9008
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.619231454145054e-06,
+      "loss": 0.9025,
+      "step": 9009
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.605852829034977e-06,
+      "loss": 0.7763,
+      "step": 9010
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.5924742039249e-06,
+      "loss": 0.8751,
+      "step": 9011
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.579095578814824e-06,
+      "loss": 0.7569,
+      "step": 9012
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.565716953704747e-06,
+      "loss": 0.9118,
+      "step": 9013
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.55233832859467e-06,
+      "loss": 0.7789,
+      "step": 9014
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.538959703484594e-06,
+      "loss": 0.676,
+      "step": 9015
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.525581078374517e-06,
+      "loss": 0.7341,
+      "step": 9016
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.512202453264442e-06,
+      "loss": 0.6816,
+      "step": 9017
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.498823828154366e-06,
+      "loss": 0.6098,
+      "step": 9018
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.48544520304429e-06,
+      "loss": 0.5318,
+      "step": 9019
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.472066577934213e-06,
+      "loss": 0.7089,
+      "step": 9020
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.458687952824136e-06,
+      "loss": 0.6537,
+      "step": 9021
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.44530932771406e-06,
+      "loss": 0.5897,
+      "step": 9022
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.431930702603983e-06,
+      "loss": 0.5425,
+      "step": 9023
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.418552077493906e-06,
+      "loss": 0.5228,
+      "step": 9024
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.40517345238383e-06,
+      "loss": 0.5926,
+      "step": 9025
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.391794827273751e-06,
+      "loss": 0.4943,
+      "step": 9026
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.378416202163675e-06,
+      "loss": 0.5876,
+      "step": 9027
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.365037577053598e-06,
+      "loss": 0.4491,
+      "step": 9028
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.351658951943521e-06,
+      "loss": 0.5131,
+      "step": 9029
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.338280326833445e-06,
+      "loss": 0.6544,
+      "step": 9030
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.324901701723368e-06,
+      "loss": 1.4727,
+      "step": 9031
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.311523076613292e-06,
+      "loss": 1.266,
+      "step": 9032
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.298144451503215e-06,
+      "loss": 1.1106,
+      "step": 9033
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.28476582639314e-06,
+      "loss": 1.1855,
+      "step": 9034
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.271387201283064e-06,
+      "loss": 1.1367,
+      "step": 9035
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.258008576172987e-06,
+      "loss": 1.1089,
+      "step": 9036
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.24462995106291e-06,
+      "loss": 1.1534,
+      "step": 9037
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.231251325952834e-06,
+      "loss": 1.2454,
+      "step": 9038
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.217872700842757e-06,
+      "loss": 0.9415,
+      "step": 9039
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.20449407573268e-06,
+      "loss": 0.9656,
+      "step": 9040
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.191115450622604e-06,
+      "loss": 1.1224,
+      "step": 9041
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.177736825512527e-06,
+      "loss": 1.0096,
+      "step": 9042
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.16435820040245e-06,
+      "loss": 1.059,
+      "step": 9043
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.150979575292374e-06,
+      "loss": 0.9918,
+      "step": 9044
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.137600950182297e-06,
+      "loss": 1.1359,
+      "step": 9045
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.12422232507222e-06,
+      "loss": 1.0559,
+      "step": 9046
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.110843699962144e-06,
+      "loss": 1.1192,
+      "step": 9047
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.097465074852068e-06,
+      "loss": 0.8447,
+      "step": 9048
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.084086449741991e-06,
+      "loss": 0.9379,
+      "step": 9049
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.070707824631916e-06,
+      "loss": 0.9851,
+      "step": 9050
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.05732919952184e-06,
+      "loss": 0.8688,
+      "step": 9051
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.043950574411763e-06,
+      "loss": 0.7835,
+      "step": 9052
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.030571949301685e-06,
+      "loss": 0.9276,
+      "step": 9053
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.017193324191608e-06,
+      "loss": 0.9167,
+      "step": 9054
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.003814699081531e-06,
+      "loss": 0.9053,
+      "step": 9055
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.990436073971455e-06,
+      "loss": 0.9289,
+      "step": 9056
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.977057448861378e-06,
+      "loss": 0.9199,
+      "step": 9057
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.963678823751301e-06,
+      "loss": 0.8212,
+      "step": 9058
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.950300198641225e-06,
+      "loss": 0.9617,
+      "step": 9059
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.936921573531148e-06,
+      "loss": 0.8073,
+      "step": 9060
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.923542948421072e-06,
+      "loss": 0.7178,
+      "step": 9061
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.910164323310995e-06,
+      "loss": 0.8376,
+      "step": 9062
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.896785698200918e-06,
+      "loss": 0.7824,
+      "step": 9063
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.883407073090842e-06,
+      "loss": 0.5636,
+      "step": 9064
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.870028447980765e-06,
+      "loss": 0.798,
+      "step": 9065
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.85664982287069e-06,
+      "loss": 0.685,
+      "step": 9066
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.843271197760614e-06,
+      "loss": 0.7426,
+      "step": 9067
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.829892572650537e-06,
+      "loss": 0.6268,
+      "step": 9068
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.81651394754046e-06,
+      "loss": 0.4848,
+      "step": 9069
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.803135322430384e-06,
+      "loss": 0.5911,
+      "step": 9070
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.789756697320307e-06,
+      "loss": 0.5931,
+      "step": 9071
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.77637807221023e-06,
+      "loss": 0.6147,
+      "step": 9072
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.762999447100154e-06,
+      "loss": 0.5587,
+      "step": 9073
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.749620821990077e-06,
+      "loss": 0.5861,
+      "step": 9074
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.73624219688e-06,
+      "loss": 0.6663,
+      "step": 9075
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.722863571769924e-06,
+      "loss": 0.6621,
+      "step": 9076
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.709484946659848e-06,
+      "loss": 0.5538,
+      "step": 9077
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.696106321549771e-06,
+      "loss": 0.4878,
+      "step": 9078
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.682727696439693e-06,
+      "loss": 0.4542,
+      "step": 9079
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.669349071329616e-06,
+      "loss": 0.5787,
+      "step": 9080
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.65597044621954e-06,
+      "loss": 1.5125,
+      "step": 9081
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.642591821109463e-06,
+      "loss": 1.2632,
+      "step": 9082
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.629213195999388e-06,
+      "loss": 1.1224,
+      "step": 9083
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.615834570889311e-06,
+      "loss": 1.1016,
+      "step": 9084
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.602455945779235e-06,
+      "loss": 1.1041,
+      "step": 9085
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.589077320669158e-06,
+      "loss": 1.3018,
+      "step": 9086
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.575698695559081e-06,
+      "loss": 1.0053,
+      "step": 9087
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.562320070449005e-06,
+      "loss": 1.0663,
+      "step": 9088
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.548941445338928e-06,
+      "loss": 1.1411,
+      "step": 9089
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.535562820228852e-06,
+      "loss": 0.8447,
+      "step": 9090
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.522184195118775e-06,
+      "loss": 0.9435,
+      "step": 9091
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.508805570008698e-06,
+      "loss": 1.0904,
+      "step": 9092
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.495426944898622e-06,
+      "loss": 0.997,
+      "step": 9093
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.482048319788545e-06,
+      "loss": 0.9159,
+      "step": 9094
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.468669694678469e-06,
+      "loss": 1.1771,
+      "step": 9095
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.455291069568392e-06,
+      "loss": 1.1562,
+      "step": 9096
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.441912444458315e-06,
+      "loss": 1.0605,
+      "step": 9097
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.428533819348239e-06,
+      "loss": 1.2046,
+      "step": 9098
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.415155194238164e-06,
+      "loss": 1.058,
+      "step": 9099
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.401776569128087e-06,
+      "loss": 0.8943,
+      "step": 9100
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.38839794401801e-06,
+      "loss": 0.8206,
+      "step": 9101
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.375019318907934e-06,
+      "loss": 0.9711,
+      "step": 9102
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.361640693797857e-06,
+      "loss": 0.9461,
+      "step": 9103
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.34826206868778e-06,
+      "loss": 0.9246,
+      "step": 9104
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.334883443577704e-06,
+      "loss": 0.8954,
+      "step": 9105
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.321504818467626e-06,
+      "loss": 1.0303,
+      "step": 9106
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.30812619335755e-06,
+      "loss": 0.7717,
+      "step": 9107
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.294747568247473e-06,
+      "loss": 0.8358,
+      "step": 9108
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.281368943137396e-06,
+      "loss": 0.8922,
+      "step": 9109
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.26799031802732e-06,
+      "loss": 0.8701,
+      "step": 9110
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.254611692917243e-06,
+      "loss": 0.8358,
+      "step": 9111
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.241233067807166e-06,
+      "loss": 0.8785,
+      "step": 9112
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.22785444269709e-06,
+      "loss": 0.713,
+      "step": 9113
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.214475817587013e-06,
+      "loss": 0.7167,
+      "step": 9114
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.201097192476938e-06,
+      "loss": 0.7395,
+      "step": 9115
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.187718567366861e-06,
+      "loss": 0.7396,
+      "step": 9116
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.174339942256785e-06,
+      "loss": 0.705,
+      "step": 9117
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.160961317146708e-06,
+      "loss": 0.6603,
+      "step": 9118
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.147582692036632e-06,
+      "loss": 0.5748,
+      "step": 9119
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.134204066926555e-06,
+      "loss": 0.6187,
+      "step": 9120
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.120825441816478e-06,
+      "loss": 0.7915,
+      "step": 9121
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.107446816706402e-06,
+      "loss": 0.5595,
+      "step": 9122
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.094068191596325e-06,
+      "loss": 0.6335,
+      "step": 9123
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.080689566486249e-06,
+      "loss": 0.5047,
+      "step": 9124
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.067310941376172e-06,
+      "loss": 0.5858,
+      "step": 9125
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.053932316266095e-06,
+      "loss": 0.5731,
+      "step": 9126
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.040553691156019e-06,
+      "loss": 0.5735,
+      "step": 9127
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.027175066045942e-06,
+      "loss": 0.4551,
+      "step": 9128
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.013796440935866e-06,
+      "loss": 0.4203,
+      "step": 9129
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.000417815825789e-06,
+      "loss": 0.5173,
+      "step": 9130
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.987039190715712e-06,
+      "loss": 1.6029,
+      "step": 9131
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.973660565605637e-06,
+      "loss": 1.332,
+      "step": 9132
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.960281940495559e-06,
+      "loss": 1.082,
+      "step": 9133
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.946903315385483e-06,
+      "loss": 1.0363,
+      "step": 9134
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.933524690275406e-06,
+      "loss": 1.0102,
+      "step": 9135
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.92014606516533e-06,
+      "loss": 1.0412,
+      "step": 9136
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.906767440055253e-06,
+      "loss": 1.1423,
+      "step": 9137
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.893388814945176e-06,
+      "loss": 1.0563,
+      "step": 9138
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.8800101898351e-06,
+      "loss": 0.9562,
+      "step": 9139
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.866631564725023e-06,
+      "loss": 1.0409,
+      "step": 9140
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.853252939614946e-06,
+      "loss": 1.0358,
+      "step": 9141
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.83987431450487e-06,
+      "loss": 1.1443,
+      "step": 9142
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.826495689394793e-06,
+      "loss": 0.939,
+      "step": 9143
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.813117064284716e-06,
+      "loss": 1.0359,
+      "step": 9144
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.79973843917464e-06,
+      "loss": 1.1472,
+      "step": 9145
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.786359814064563e-06,
+      "loss": 0.923,
+      "step": 9146
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.772981188954487e-06,
+      "loss": 0.9179,
+      "step": 9147
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.759602563844412e-06,
+      "loss": 0.898,
+      "step": 9148
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.746223938734335e-06,
+      "loss": 0.9351,
+      "step": 9149
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.732845313624258e-06,
+      "loss": 0.9105,
+      "step": 9150
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.719466688514182e-06,
+      "loss": 0.9846,
+      "step": 9151
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.706088063404105e-06,
+      "loss": 0.9663,
+      "step": 9152
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.692709438294029e-06,
+      "loss": 0.9726,
+      "step": 9153
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.679330813183952e-06,
+      "loss": 0.9219,
+      "step": 9154
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.665952188073875e-06,
+      "loss": 0.9877,
+      "step": 9155
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.652573562963799e-06,
+      "loss": 0.9217,
+      "step": 9156
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.639194937853722e-06,
+      "loss": 1.0211,
+      "step": 9157
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.625816312743645e-06,
+      "loss": 0.8181,
+      "step": 9158
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.612437687633568e-06,
+      "loss": 0.9538,
+      "step": 9159
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.599059062523492e-06,
+      "loss": 0.747,
+      "step": 9160
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.585680437413416e-06,
+      "loss": 0.84,
+      "step": 9161
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.572301812303339e-06,
+      "loss": 0.7301,
+      "step": 9162
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.5589231871932625e-06,
+      "loss": 0.7328,
+      "step": 9163
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.545544562083186e-06,
+      "loss": 0.7929,
+      "step": 9164
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.5321659369731085e-06,
+      "loss": 0.7553,
+      "step": 9165
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.518787311863032e-06,
+      "loss": 0.7668,
+      "step": 9166
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.505408686752955e-06,
+      "loss": 0.7184,
+      "step": 9167
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.492030061642879e-06,
+      "loss": 0.7206,
+      "step": 9168
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.478651436532803e-06,
+      "loss": 0.5305,
+      "step": 9169
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.465272811422726e-06,
+      "loss": 0.7207,
+      "step": 9170
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.45189418631265e-06,
+      "loss": 0.776,
+      "step": 9171
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.438515561202573e-06,
+      "loss": 0.6496,
+      "step": 9172
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.4251369360924964e-06,
+      "loss": 0.6333,
+      "step": 9173
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.41175831098242e-06,
+      "loss": 0.4496,
+      "step": 9174
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.398379685872343e-06,
+      "loss": 0.556,
+      "step": 9175
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.385001060762267e-06,
+      "loss": 0.7446,
+      "step": 9176
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.371622435652191e-06,
+      "loss": 0.4359,
+      "step": 9177
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.358243810542114e-06,
+      "loss": 0.4709,
+      "step": 9178
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.344865185432037e-06,
+      "loss": 0.5096,
+      "step": 9179
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.33148656032196e-06,
+      "loss": 0.5121,
+      "step": 9180
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.3181079352118836e-06,
+      "loss": 1.4951,
+      "step": 9181
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.304729310101807e-06,
+      "loss": 1.2771,
+      "step": 9182
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.29135068499173e-06,
+      "loss": 1.1838,
+      "step": 9183
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.277972059881654e-06,
+      "loss": 1.0196,
+      "step": 9184
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.264593434771578e-06,
+      "loss": 1.1577,
+      "step": 9185
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.251214809661501e-06,
+      "loss": 0.985,
+      "step": 9186
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.237836184551425e-06,
+      "loss": 1.2737,
+      "step": 9187
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.224457559441348e-06,
+      "loss": 0.9861,
+      "step": 9188
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.2110789343312715e-06,
+      "loss": 1.198,
+      "step": 9189
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.197700309221195e-06,
+      "loss": 1.038,
+      "step": 9190
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.184321684111118e-06,
+      "loss": 0.907,
+      "step": 9191
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.170943059001041e-06,
+      "loss": 1.0913,
+      "step": 9192
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.157564433890965e-06,
+      "loss": 1.0367,
+      "step": 9193
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.1441858087808885e-06,
+      "loss": 1.0395,
+      "step": 9194
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.130807183670812e-06,
+      "loss": 1.0355,
+      "step": 9195
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.117428558560735e-06,
+      "loss": 0.9488,
+      "step": 9196
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.104049933450659e-06,
+      "loss": 0.874,
+      "step": 9197
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.090671308340582e-06,
+      "loss": 0.9265,
+      "step": 9198
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.0772926832305054e-06,
+      "loss": 0.8019,
+      "step": 9199
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.063914058120429e-06,
+      "loss": 1.0477,
+      "step": 9200
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.050535433010353e-06,
+      "loss": 0.8588,
+      "step": 9201
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.0371568079002764e-06,
+      "loss": 0.7452,
+      "step": 9202
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.0237781827902e-06,
+      "loss": 0.8386,
+      "step": 9203
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.010399557680123e-06,
+      "loss": 0.875,
+      "step": 9204
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.997020932570046e-06,
+      "loss": 0.8799,
+      "step": 9205
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.983642307459969e-06,
+      "loss": 0.8683,
+      "step": 9206
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.9702636823498925e-06,
+      "loss": 0.8897,
+      "step": 9207
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.956885057239816e-06,
+      "loss": 0.8024,
+      "step": 9208
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.94350643212974e-06,
+      "loss": 0.8326,
+      "step": 9209
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.9301278070196636e-06,
+      "loss": 0.7475,
+      "step": 9210
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.916749181909587e-06,
+      "loss": 0.7995,
+      "step": 9211
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.90337055679951e-06,
+      "loss": 0.8366,
+      "step": 9212
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.889991931689434e-06,
+      "loss": 0.8885,
+      "step": 9213
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.876613306579357e-06,
+      "loss": 0.6983,
+      "step": 9214
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.8632346814692805e-06,
+      "loss": 0.7596,
+      "step": 9215
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.849856056359204e-06,
+      "loss": 0.7266,
+      "step": 9216
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.836477431249127e-06,
+      "loss": 0.5742,
+      "step": 9217
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.8230988061390515e-06,
+      "loss": 0.6389,
+      "step": 9218
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.809720181028974e-06,
+      "loss": 0.7051,
+      "step": 9219
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.7963415559188975e-06,
+      "loss": 0.7187,
+      "step": 9220
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.782962930808821e-06,
+      "loss": 0.6155,
+      "step": 9221
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.769584305698744e-06,
+      "loss": 0.7036,
+      "step": 9222
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.756205680588668e-06,
+      "loss": 0.5566,
+      "step": 9223
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.742827055478591e-06,
+      "loss": 0.5448,
+      "step": 9224
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.729448430368514e-06,
+      "loss": 0.6427,
+      "step": 9225
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.716069805258439e-06,
+      "loss": 0.5116,
+      "step": 9226
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.702691180148362e-06,
+      "loss": 0.4325,
+      "step": 9227
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.6893125550382854e-06,
+      "loss": 0.5346,
+      "step": 9228
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.675933929928209e-06,
+      "loss": 0.4995,
+      "step": 9229
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.662555304818132e-06,
+      "loss": 0.4798,
+      "step": 9230
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.649176679708056e-06,
+      "loss": 1.6098,
+      "step": 9231
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.635798054597978e-06,
+      "loss": 1.2835,
+      "step": 9232
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.6224194294879015e-06,
+      "loss": 1.2861,
+      "step": 9233
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.609040804377826e-06,
+      "loss": 1.1651,
+      "step": 9234
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.595662179267749e-06,
+      "loss": 1.0939,
+      "step": 9235
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.5822835541576726e-06,
+      "loss": 1.0911,
+      "step": 9236
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.568904929047596e-06,
+      "loss": 1.1565,
+      "step": 9237
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.555526303937519e-06,
+      "loss": 1.0615,
+      "step": 9238
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.542147678827443e-06,
+      "loss": 1.2047,
+      "step": 9239
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.528769053717366e-06,
+      "loss": 0.9187,
+      "step": 9240
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.5153904286072895e-06,
+      "loss": 1.3402,
+      "step": 9241
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.502011803497214e-06,
+      "loss": 0.9037,
+      "step": 9242
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.488633178387137e-06,
+      "loss": 1.005,
+      "step": 9243
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.4752545532770605e-06,
+      "loss": 0.9642,
+      "step": 9244
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.461875928166983e-06,
+      "loss": 1.0621,
+      "step": 9245
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.4484973030569065e-06,
+      "loss": 0.8998,
+      "step": 9246
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.43511867794683e-06,
+      "loss": 1.0109,
+      "step": 9247
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.421740052836753e-06,
+      "loss": 0.8638,
+      "step": 9248
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.408361427726677e-06,
+      "loss": 0.8969,
+      "step": 9249
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.394982802616601e-06,
+      "loss": 1.0272,
+      "step": 9250
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.381604177506524e-06,
+      "loss": 0.9433,
+      "step": 9251
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.368225552396448e-06,
+      "loss": 0.8717,
+      "step": 9252
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.354846927286371e-06,
+      "loss": 0.968,
+      "step": 9253
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.3414683021762944e-06,
+      "loss": 0.914,
+      "step": 9254
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.328089677066218e-06,
+      "loss": 0.8831,
+      "step": 9255
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.314711051956141e-06,
+      "loss": 0.7706,
+      "step": 9256
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.301332426846065e-06,
+      "loss": 0.7139,
+      "step": 9257
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.287953801735987e-06,
+      "loss": 0.8413,
+      "step": 9258
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.274575176625911e-06,
+      "loss": 0.9358,
+      "step": 9259
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.261196551515835e-06,
+      "loss": 0.7958,
+      "step": 9260
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.247817926405758e-06,
+      "loss": 0.8002,
+      "step": 9261
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.2344393012956815e-06,
+      "loss": 0.8709,
+      "step": 9262
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.221060676185605e-06,
+      "loss": 0.854,
+      "step": 9263
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.207682051075528e-06,
+      "loss": 0.767,
+      "step": 9264
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.194303425965452e-06,
+      "loss": 0.7607,
+      "step": 9265
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.180924800855375e-06,
+      "loss": 0.6134,
+      "step": 9266
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.167546175745299e-06,
+      "loss": 0.6539,
+      "step": 9267
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.154167550635223e-06,
+      "loss": 0.5933,
+      "step": 9268
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.140788925525146e-06,
+      "loss": 0.6469,
+      "step": 9269
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.1274103004150695e-06,
+      "loss": 0.6,
+      "step": 9270
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.114031675304992e-06,
+      "loss": 0.5606,
+      "step": 9271
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.1006530501949154e-06,
+      "loss": 0.5034,
+      "step": 9272
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.087274425084839e-06,
+      "loss": 0.5428,
+      "step": 9273
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.073895799974762e-06,
+      "loss": 0.4957,
+      "step": 9274
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.0605171748646865e-06,
+      "loss": 0.5123,
+      "step": 9275
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.04713854975461e-06,
+      "loss": 0.5607,
+      "step": 9276
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.033759924644533e-06,
+      "loss": 0.606,
+      "step": 9277
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.020381299534457e-06,
+      "loss": 0.4146,
+      "step": 9278
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.00700267442438e-06,
+      "loss": 0.6548,
+      "step": 9279
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.993624049314303e-06,
+      "loss": 0.5216,
+      "step": 9280
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.980245424204227e-06,
+      "loss": 1.3113,
+      "step": 9281
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.96686679909415e-06,
+      "loss": 1.2847,
+      "step": 9282
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.9534881739840744e-06,
+      "loss": 1.2046,
+      "step": 9283
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.940109548873998e-06,
+      "loss": 1.1712,
+      "step": 9284
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.92673092376392e-06,
+      "loss": 0.9835,
+      "step": 9285
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.913352298653844e-06,
+      "loss": 1.0292,
+      "step": 9286
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.899973673543767e-06,
+      "loss": 1.0711,
+      "step": 9287
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.8865950484336905e-06,
+      "loss": 1.136,
+      "step": 9288
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.873216423323614e-06,
+      "loss": 1.1729,
+      "step": 9289
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.859837798213537e-06,
+      "loss": 1.0657,
+      "step": 9290
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.8464591731034616e-06,
+      "loss": 0.9173,
+      "step": 9291
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.833080547993385e-06,
+      "loss": 1.1876,
+      "step": 9292
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.819701922883308e-06,
+      "loss": 1.0713,
+      "step": 9293
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.806323297773232e-06,
+      "loss": 1.0495,
+      "step": 9294
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.792944672663155e-06,
+      "loss": 0.9296,
+      "step": 9295
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.7795660475530785e-06,
+      "loss": 0.9611,
+      "step": 9296
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.766187422443002e-06,
+      "loss": 0.878,
+      "step": 9297
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.7528087973329244e-06,
+      "loss": 0.9794,
+      "step": 9298
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.739430172222849e-06,
+      "loss": 0.9486,
+      "step": 9299
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.726051547112772e-06,
+      "loss": 0.9055,
+      "step": 9300
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.7126729220026955e-06,
+      "loss": 1.0354,
+      "step": 9301
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.699294296892619e-06,
+      "loss": 0.8779,
+      "step": 9302
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.685915671782542e-06,
+      "loss": 0.9539,
+      "step": 9303
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.672537046672466e-06,
+      "loss": 0.888,
+      "step": 9304
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.659158421562389e-06,
+      "loss": 0.7465,
+      "step": 9305
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.645779796452312e-06,
+      "loss": 0.9254,
+      "step": 9306
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.632401171342236e-06,
+      "loss": 0.9288,
+      "step": 9307
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.61902254623216e-06,
+      "loss": 0.8853,
+      "step": 9308
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.6056439211220834e-06,
+      "loss": 0.6962,
+      "step": 9309
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.592265296012007e-06,
+      "loss": 0.9565,
+      "step": 9310
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.578886670901929e-06,
+      "loss": 0.7137,
+      "step": 9311
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.565508045791853e-06,
+      "loss": 0.8508,
+      "step": 9312
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.552129420681776e-06,
+      "loss": 0.7373,
+      "step": 9313
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.5387507955716995e-06,
+      "loss": 0.7359,
+      "step": 9314
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.525372170461623e-06,
+      "loss": 0.69,
+      "step": 9315
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.511993545351547e-06,
+      "loss": 0.5871,
+      "step": 9316
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.4986149202414705e-06,
+      "loss": 0.7761,
+      "step": 9317
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.485236295131394e-06,
+      "loss": 0.6932,
+      "step": 9318
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.471857670021317e-06,
+      "loss": 0.6308,
+      "step": 9319
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.458479044911241e-06,
+      "loss": 0.555,
+      "step": 9320
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.445100419801164e-06,
+      "loss": 0.6436,
+      "step": 9321
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.4317217946910875e-06,
+      "loss": 0.6678,
+      "step": 9322
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.418343169581011e-06,
+      "loss": 0.6071,
+      "step": 9323
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.404964544470935e-06,
+      "loss": 0.5579,
+      "step": 9324
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.391585919360858e-06,
+      "loss": 0.7333,
+      "step": 9325
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.378207294250781e-06,
+      "loss": 0.4798,
+      "step": 9326
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.3648286691407044e-06,
+      "loss": 0.4433,
+      "step": 9327
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.351450044030628e-06,
+      "loss": 0.5827,
+      "step": 9328
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.338071418920551e-06,
+      "loss": 0.5636,
+      "step": 9329
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.324692793810475e-06,
+      "loss": 0.5397,
+      "step": 9330
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.311314168700398e-06,
+      "loss": 1.4495,
+      "step": 9331
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.297935543590322e-06,
+      "loss": 1.2786,
+      "step": 9332
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.284556918480246e-06,
+      "loss": 1.2234,
+      "step": 9333
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.271178293370169e-06,
+      "loss": 1.1248,
+      "step": 9334
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.257799668260092e-06,
+      "loss": 1.0982,
+      "step": 9335
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.244421043150016e-06,
+      "loss": 1.2485,
+      "step": 9336
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.231042418039939e-06,
+      "loss": 1.0394,
+      "step": 9337
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.217663792929862e-06,
+      "loss": 1.0077,
+      "step": 9338
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.204285167819785e-06,
+      "loss": 1.093,
+      "step": 9339
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.190906542709709e-06,
+      "loss": 0.9967,
+      "step": 9340
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.177527917599633e-06,
+      "loss": 1.0751,
+      "step": 9341
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.164149292489556e-06,
+      "loss": 1.0522,
+      "step": 9342
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.1507706673794795e-06,
+      "loss": 0.9479,
+      "step": 9343
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.137392042269403e-06,
+      "loss": 1.0033,
+      "step": 9344
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.124013417159326e-06,
+      "loss": 0.994,
+      "step": 9345
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.11063479204925e-06,
+      "loss": 0.9137,
+      "step": 9346
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.097256166939173e-06,
+      "loss": 0.9937,
+      "step": 9347
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.083877541829097e-06,
+      "loss": 1.0115,
+      "step": 9348
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.070498916719021e-06,
+      "loss": 0.9073,
+      "step": 9349
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.057120291608944e-06,
+      "loss": 1.011,
+      "step": 9350
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.043741666498867e-06,
+      "loss": 0.9649,
+      "step": 9351
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.03036304138879e-06,
+      "loss": 1.0726,
+      "step": 9352
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.0169844162787134e-06,
+      "loss": 1.0997,
+      "step": 9353
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.003605791168637e-06,
+      "loss": 0.8407,
+      "step": 9354
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.99022716605856e-06,
+      "loss": 0.9332,
+      "step": 9355
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.976848540948484e-06,
+      "loss": 0.8343,
+      "step": 9356
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.963469915838408e-06,
+      "loss": 0.9521,
+      "step": 9357
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.950091290728331e-06,
+      "loss": 0.8055,
+      "step": 9358
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.936712665618255e-06,
+      "loss": 0.774,
+      "step": 9359
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.923334040508178e-06,
+      "loss": 0.7177,
+      "step": 9360
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.909955415398101e-06,
+      "loss": 0.8715,
+      "step": 9361
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.896576790288025e-06,
+      "loss": 0.7971,
+      "step": 9362
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.883198165177948e-06,
+      "loss": 0.7326,
+      "step": 9363
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.869819540067871e-06,
+      "loss": 0.8641,
+      "step": 9364
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.856440914957795e-06,
+      "loss": 0.7888,
+      "step": 9365
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.843062289847718e-06,
+      "loss": 0.6987,
+      "step": 9366
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.829683664737642e-06,
+      "loss": 0.6713,
+      "step": 9367
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.816305039627565e-06,
+      "loss": 0.6068,
+      "step": 9368
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.8029264145174885e-06,
+      "loss": 0.6935,
+      "step": 9369
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.789547789407412e-06,
+      "loss": 0.6419,
+      "step": 9370
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.776169164297335e-06,
+      "loss": 0.5993,
+      "step": 9371
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.762790539187259e-06,
+      "loss": 0.4675,
+      "step": 9372
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.749411914077183e-06,
+      "loss": 0.5209,
+      "step": 9373
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.736033288967106e-06,
+      "loss": 0.5404,
+      "step": 9374
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.72265466385703e-06,
+      "loss": 0.4849,
+      "step": 9375
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.709276038746953e-06,
+      "loss": 0.6688,
+      "step": 9376
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.695897413636876e-06,
+      "loss": 0.4752,
+      "step": 9377
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.682518788526799e-06,
+      "loss": 0.4361,
+      "step": 9378
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.6691401634167224e-06,
+      "loss": 0.5092,
+      "step": 9379
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.655761538306646e-06,
+      "loss": 0.5452,
+      "step": 9380
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.64238291319657e-06,
+      "loss": 1.3908,
+      "step": 9381
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.6290042880864934e-06,
+      "loss": 1.2827,
+      "step": 9382
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.615625662976417e-06,
+      "loss": 1.204,
+      "step": 9383
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.60224703786634e-06,
+      "loss": 1.1699,
+      "step": 9384
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.588868412756264e-06,
+      "loss": 1.1393,
+      "step": 9385
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.575489787646187e-06,
+      "loss": 1.2102,
+      "step": 9386
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.56211116253611e-06,
+      "loss": 1.1346,
+      "step": 9387
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.548732537426034e-06,
+      "loss": 0.9462,
+      "step": 9388
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.535353912315958e-06,
+      "loss": 1.1098,
+      "step": 9389
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.521975287205881e-06,
+      "loss": 1.1075,
+      "step": 9390
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.508596662095804e-06,
+      "loss": 0.9542,
+      "step": 9391
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.495218036985727e-06,
+      "loss": 0.8769,
+      "step": 9392
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.481839411875651e-06,
+      "loss": 0.9921,
+      "step": 9393
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.468460786765574e-06,
+      "loss": 0.9514,
+      "step": 9394
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.4550821616554975e-06,
+      "loss": 0.827,
+      "step": 9395
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.441703536545421e-06,
+      "loss": 1.1136,
+      "step": 9396
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.428324911435345e-06,
+      "loss": 1.2153,
+      "step": 9397
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.4149462863252685e-06,
+      "loss": 0.9511,
+      "step": 9398
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.401567661215192e-06,
+      "loss": 1.1205,
+      "step": 9399
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.388189036105115e-06,
+      "loss": 0.9215,
+      "step": 9400
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.374810410995039e-06,
+      "loss": 0.8443,
+      "step": 9401
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.361431785884962e-06,
+      "loss": 0.9641,
+      "step": 9402
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.3480531607748855e-06,
+      "loss": 0.8661,
+      "step": 9403
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.334674535664808e-06,
+      "loss": 0.8445,
+      "step": 9404
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.321295910554731e-06,
+      "loss": 0.8812,
+      "step": 9405
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.307917285444656e-06,
+      "loss": 0.8217,
+      "step": 9406
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.294538660334579e-06,
+      "loss": 0.8723,
+      "step": 9407
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.2811600352245024e-06,
+      "loss": 0.8809,
+      "step": 9408
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.267781410114426e-06,
+      "loss": 0.8589,
+      "step": 9409
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.254402785004349e-06,
+      "loss": 0.8301,
+      "step": 9410
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.241024159894273e-06,
+      "loss": 1.0007,
+      "step": 9411
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.227645534784196e-06,
+      "loss": 0.7481,
+      "step": 9412
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.214266909674119e-06,
+      "loss": 0.6726,
+      "step": 9413
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.200888284564044e-06,
+      "loss": 0.6833,
+      "step": 9414
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.187509659453967e-06,
+      "loss": 0.6738,
+      "step": 9415
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.17413103434389e-06,
+      "loss": 0.665,
+      "step": 9416
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.160752409233813e-06,
+      "loss": 0.6978,
+      "step": 9417
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.147373784123736e-06,
+      "loss": 0.6515,
+      "step": 9418
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.13399515901366e-06,
+      "loss": 0.7053,
+      "step": 9419
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.120616533903583e-06,
+      "loss": 0.7635,
+      "step": 9420
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.1072379087935065e-06,
+      "loss": 0.6796,
+      "step": 9421
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.093859283683431e-06,
+      "loss": 0.5162,
+      "step": 9422
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.080480658573354e-06,
+      "loss": 0.7313,
+      "step": 9423
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.0671020334632775e-06,
+      "loss": 0.6632,
+      "step": 9424
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.053723408353201e-06,
+      "loss": 0.522,
+      "step": 9425
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.040344783243124e-06,
+      "loss": 0.5149,
+      "step": 9426
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.026966158133048e-06,
+      "loss": 0.5097,
+      "step": 9427
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.013587533022971e-06,
+      "loss": 0.5138,
+      "step": 9428
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.0002089079128945e-06,
+      "loss": 0.4937,
+      "step": 9429
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.986830282802819e-06,
+      "loss": 0.599,
+      "step": 9430
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.973451657692741e-06,
+      "loss": 1.4611,
+      "step": 9431
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.960073032582665e-06,
+      "loss": 1.195,
+      "step": 9432
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.946694407472588e-06,
+      "loss": 1.1474,
+      "step": 9433
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.9333157823625114e-06,
+      "loss": 1.1244,
+      "step": 9434
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.919937157252435e-06,
+      "loss": 1.1866,
+      "step": 9435
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.906558532142358e-06,
+      "loss": 1.0639,
+      "step": 9436
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.893179907032282e-06,
+      "loss": 1.1705,
+      "step": 9437
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.879801281922206e-06,
+      "loss": 0.9972,
+      "step": 9438
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.866422656812129e-06,
+      "loss": 0.933,
+      "step": 9439
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.853044031702053e-06,
+      "loss": 1.0089,
+      "step": 9440
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.839665406591976e-06,
+      "loss": 1.1227,
+      "step": 9441
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.826286781481899e-06,
+      "loss": 1.1519,
+      "step": 9442
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.8129081563718224e-06,
+      "loss": 0.9629,
+      "step": 9443
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.799529531261746e-06,
+      "loss": 0.9732,
+      "step": 9444
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.7861509061516696e-06,
+      "loss": 0.9677,
+      "step": 9445
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.772772281041593e-06,
+      "loss": 0.9659,
+      "step": 9446
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.759393655931516e-06,
+      "loss": 1.0411,
+      "step": 9447
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.7460150308214393e-06,
+      "loss": 0.927,
+      "step": 9448
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.732636405711363e-06,
+      "loss": 1.0798,
+      "step": 9449
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.7192577806012865e-06,
+      "loss": 0.917,
+      "step": 9450
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.70587915549121e-06,
+      "loss": 1.1298,
+      "step": 9451
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.6925005303811333e-06,
+      "loss": 0.9049,
+      "step": 9452
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.679121905271057e-06,
+      "loss": 0.9914,
+      "step": 9453
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.66574328016098e-06,
+      "loss": 1.0441,
+      "step": 9454
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.6523646550509035e-06,
+      "loss": 0.9244,
+      "step": 9455
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.638986029940827e-06,
+      "loss": 0.7304,
+      "step": 9456
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.6256074048307507e-06,
+      "loss": 0.7718,
+      "step": 9457
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.612228779720674e-06,
+      "loss": 0.8532,
+      "step": 9458
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5988501546105975e-06,
+      "loss": 0.8993,
+      "step": 9459
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5854715295005204e-06,
+      "loss": 0.9313,
+      "step": 9460
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5720929043904442e-06,
+      "loss": 0.9468,
+      "step": 9461
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5587142792803676e-06,
+      "loss": 0.8136,
+      "step": 9462
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.545335654170291e-06,
+      "loss": 0.859,
+      "step": 9463
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5319570290602144e-06,
+      "loss": 0.7754,
+      "step": 9464
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5185784039501382e-06,
+      "loss": 0.6979,
+      "step": 9465
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5051997788400616e-06,
+      "loss": 0.6938,
+      "step": 9466
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.4918211537299846e-06,
+      "loss": 0.7342,
+      "step": 9467
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.478442528619908e-06,
+      "loss": 0.6396,
+      "step": 9468
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.4650639035098318e-06,
+      "loss": 0.7131,
+      "step": 9469
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.451685278399755e-06,
+      "loss": 0.6687,
+      "step": 9470
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.4383066532896786e-06,
+      "loss": 0.5647,
+      "step": 9471
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.424928028179602e-06,
+      "loss": 0.5727,
+      "step": 9472
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.4115494030695258e-06,
+      "loss": 0.5795,
+      "step": 9473
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.3981707779594487e-06,
+      "loss": 0.6494,
+      "step": 9474
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.384792152849372e-06,
+      "loss": 0.5316,
+      "step": 9475
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.3714135277392955e-06,
+      "loss": 0.6008,
+      "step": 9476
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.3580349026292193e-06,
+      "loss": 0.449,
+      "step": 9477
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.3446562775191427e-06,
+      "loss": 0.499,
+      "step": 9478
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.331277652409066e-06,
+      "loss": 0.4935,
+      "step": 9479
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.317899027298989e-06,
+      "loss": 0.4342,
+      "step": 9480
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.304520402188913e-06,
+      "loss": 1.3358,
+      "step": 9481
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.2911417770788363e-06,
+      "loss": 1.4824,
+      "step": 9482
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.2777631519687597e-06,
+      "loss": 1.23,
+      "step": 9483
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.264384526858683e-06,
+      "loss": 1.0919,
+      "step": 9484
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.251005901748607e-06,
+      "loss": 1.0084,
+      "step": 9485
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.2376272766385303e-06,
+      "loss": 1.1028,
+      "step": 9486
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.2242486515284532e-06,
+      "loss": 1.1441,
+      "step": 9487
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.2108700264183766e-06,
+      "loss": 1.1281,
+      "step": 9488
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1974914013083004e-06,
+      "loss": 1.0217,
+      "step": 9489
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.184112776198224e-06,
+      "loss": 1.0701,
+      "step": 9490
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1707341510881472e-06,
+      "loss": 1.0513,
+      "step": 9491
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1573555259780706e-06,
+      "loss": 0.9638,
+      "step": 9492
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1439769008679936e-06,
+      "loss": 0.7851,
+      "step": 9493
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1305982757579174e-06,
+      "loss": 0.9784,
+      "step": 9494
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1172196506478408e-06,
+      "loss": 0.8187,
+      "step": 9495
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.103841025537764e-06,
+      "loss": 0.8308,
+      "step": 9496
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.0904624004276876e-06,
+      "loss": 0.9546,
+      "step": 9497
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.0770837753176114e-06,
+      "loss": 0.7907,
+      "step": 9498
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.0637051502075348e-06,
+      "loss": 1.0205,
+      "step": 9499
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.0503265250974577e-06,
+      "loss": 0.908,
+      "step": 9500
+    },
+    {
+      "epoch": 2.93,
+      "eval_bleu": 2.691896075803751e-15,
+      "eval_loss": 1.642454743385315,
+      "eval_runtime": 2632.4174,
+      "eval_samples_per_second": 5.607,
+      "eval_steps_per_second": 0.701,
+      "step": 9500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.036947899987381e-06,
+      "loss": 0.9419,
+      "step": 9501
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.023569274877305e-06,
+      "loss": 0.8354,
+      "step": 9502
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.0101906497672283e-06,
+      "loss": 0.9972,
+      "step": 9503
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9968120246571517e-06,
+      "loss": 0.9667,
+      "step": 9504
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.983433399547075e-06,
+      "loss": 1.0181,
+      "step": 9505
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.970054774436999e-06,
+      "loss": 0.8289,
+      "step": 9506
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.956676149326922e-06,
+      "loss": 0.7405,
+      "step": 9507
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9432975242168453e-06,
+      "loss": 0.791,
+      "step": 9508
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9299188991067687e-06,
+      "loss": 0.6459,
+      "step": 9509
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.9165402739966925e-06,
+      "loss": 0.8508,
+      "step": 9510
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.903161648886616e-06,
+      "loss": 0.7412,
+      "step": 9511
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.8897830237765393e-06,
+      "loss": 0.7792,
+      "step": 9512
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.8764043986664622e-06,
+      "loss": 0.7224,
+      "step": 9513
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.863025773556386e-06,
+      "loss": 0.6916,
+      "step": 9514
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.8496471484463094e-06,
+      "loss": 0.7001,
+      "step": 9515
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.836268523336233e-06,
+      "loss": 0.8281,
+      "step": 9516
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.822889898226156e-06,
+      "loss": 0.7466,
+      "step": 9517
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.80951127311608e-06,
+      "loss": 0.8398,
+      "step": 9518
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.7961326480060034e-06,
+      "loss": 0.6768,
+      "step": 9519
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.7827540228959264e-06,
+      "loss": 0.5267,
+      "step": 9520
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.7693753977858498e-06,
+      "loss": 0.567,
+      "step": 9521
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.7559967726757736e-06,
+      "loss": 0.7549,
+      "step": 9522
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.742618147565697e-06,
+      "loss": 0.6478,
+      "step": 9523
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.7292395224556204e-06,
+      "loss": 0.5793,
+      "step": 9524
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.7158608973455437e-06,
+      "loss": 0.6202,
+      "step": 9525
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.7024822722354676e-06,
+      "loss": 0.5262,
+      "step": 9526
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.6891036471253905e-06,
+      "loss": 0.5202,
+      "step": 9527
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.675725022015314e-06,
+      "loss": 0.5803,
+      "step": 9528
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.6623463969052373e-06,
+      "loss": 0.5051,
+      "step": 9529
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.648967771795161e-06,
+      "loss": 0.5374,
+      "step": 9530
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.6355891466850845e-06,
+      "loss": 1.519,
+      "step": 9531
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.622210521575008e-06,
+      "loss": 1.4184,
+      "step": 9532
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.608831896464931e-06,
+      "loss": 1.1781,
+      "step": 9533
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.5954532713548547e-06,
+      "loss": 1.2827,
+      "step": 9534
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.582074646244778e-06,
+      "loss": 1.0881,
+      "step": 9535
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.5686960211347015e-06,
+      "loss": 1.0587,
+      "step": 9536
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.555317396024625e-06,
+      "loss": 1.0132,
+      "step": 9537
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.5419387709145487e-06,
+      "loss": 1.0915,
+      "step": 9538
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.528560145804472e-06,
+      "loss": 0.848,
+      "step": 9539
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.515181520694395e-06,
+      "loss": 0.9918,
+      "step": 9540
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.5018028955843184e-06,
+      "loss": 1.122,
+      "step": 9541
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.488424270474242e-06,
+      "loss": 1.1158,
+      "step": 9542
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.4750456453641656e-06,
+      "loss": 0.9253,
+      "step": 9543
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.461667020254089e-06,
+      "loss": 0.9115,
+      "step": 9544
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.4482883951440124e-06,
+      "loss": 0.9191,
+      "step": 9545
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.4349097700339354e-06,
+      "loss": 0.9861,
+      "step": 9546
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.421531144923859e-06,
+      "loss": 0.8225,
+      "step": 9547
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.4081525198137826e-06,
+      "loss": 0.8938,
+      "step": 9548
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.394773894703706e-06,
+      "loss": 0.9004,
+      "step": 9549
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.3813952695936293e-06,
+      "loss": 0.8472,
+      "step": 9550
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.368016644483553e-06,
+      "loss": 0.8712,
+      "step": 9551
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.3546380193734766e-06,
+      "loss": 0.9511,
+      "step": 9552
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.3412593942633995e-06,
+      "loss": 0.9569,
+      "step": 9553
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.327880769153323e-06,
+      "loss": 0.9591,
+      "step": 9554
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.3145021440432467e-06,
+      "loss": 0.9423,
+      "step": 9555
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.30112351893317e-06,
+      "loss": 0.64,
+      "step": 9556
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.2877448938230935e-06,
+      "loss": 1.0455,
+      "step": 9557
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.274366268713017e-06,
+      "loss": 0.827,
+      "step": 9558
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.2609876436029407e-06,
+      "loss": 0.8116,
+      "step": 9559
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.2476090184928637e-06,
+      "loss": 0.7428,
+      "step": 9560
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.234230393382787e-06,
+      "loss": 0.6055,
+      "step": 9561
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.2208517682727105e-06,
+      "loss": 0.9007,
+      "step": 9562
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.2074731431626343e-06,
+      "loss": 0.7565,
+      "step": 9563
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.1940945180525577e-06,
+      "loss": 0.6953,
+      "step": 9564
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.180715892942481e-06,
+      "loss": 0.8225,
+      "step": 9565
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.167337267832404e-06,
+      "loss": 0.6215,
+      "step": 9566
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.153958642722328e-06,
+      "loss": 0.6019,
+      "step": 9567
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.1405800176122512e-06,
+      "loss": 0.4749,
+      "step": 9568
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.1272013925021746e-06,
+      "loss": 0.727,
+      "step": 9569
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.113822767392098e-06,
+      "loss": 0.5533,
+      "step": 9570
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.100444142282022e-06,
+      "loss": 0.6097,
+      "step": 9571
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.087065517171945e-06,
+      "loss": 0.6077,
+      "step": 9572
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.073686892061868e-06,
+      "loss": 0.5721,
+      "step": 9573
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.0603082669517916e-06,
+      "loss": 0.5106,
+      "step": 9574
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.0469296418417154e-06,
+      "loss": 0.4599,
+      "step": 9575
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.0335510167316388e-06,
+      "loss": 0.7191,
+      "step": 9576
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.020172391621562e-06,
+      "loss": 0.5368,
+      "step": 9577
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.0067937665114855e-06,
+      "loss": 0.6343,
+      "step": 9578
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9934151414014094e-06,
+      "loss": 0.4098,
+      "step": 9579
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9800365162913323e-06,
+      "loss": 0.5768,
+      "step": 9580
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9666578911812557e-06,
+      "loss": 1.4989,
+      "step": 9581
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.953279266071179e-06,
+      "loss": 1.225,
+      "step": 9582
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.939900640961103e-06,
+      "loss": 1.0279,
+      "step": 9583
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9265220158510263e-06,
+      "loss": 0.9913,
+      "step": 9584
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9131433907409497e-06,
+      "loss": 1.1361,
+      "step": 9585
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.899764765630873e-06,
+      "loss": 1.1686,
+      "step": 9586
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8863861405207965e-06,
+      "loss": 1.2323,
+      "step": 9587
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8730075154107197e-06,
+      "loss": 1.0978,
+      "step": 9588
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8596288903006433e-06,
+      "loss": 0.972,
+      "step": 9589
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8462502651905667e-06,
+      "loss": 1.0542,
+      "step": 9590
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.83287164008049e-06,
+      "loss": 1.1187,
+      "step": 9591
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8194930149704134e-06,
+      "loss": 0.9245,
+      "step": 9592
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.806114389860337e-06,
+      "loss": 0.9522,
+      "step": 9593
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.7927357647502602e-06,
+      "loss": 1.0926,
+      "step": 9594
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.7793571396401838e-06,
+      "loss": 1.1008,
+      "step": 9595
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.7659785145301072e-06,
+      "loss": 1.0556,
+      "step": 9596
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.7525998894200308e-06,
+      "loss": 1.0946,
+      "step": 9597
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.739221264309954e-06,
+      "loss": 0.9136,
+      "step": 9598
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.7258426391998776e-06,
+      "loss": 0.8757,
+      "step": 9599
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.712464014089801e-06,
+      "loss": 1.0656,
+      "step": 9600
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.6990853889797244e-06,
+      "loss": 0.8046,
+      "step": 9601
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.6857067638696478e-06,
+      "loss": 0.8747,
+      "step": 9602
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.6723281387595714e-06,
+      "loss": 1.0734,
+      "step": 9603
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.6589495136494945e-06,
+      "loss": 0.8563,
+      "step": 9604
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.6455708885394181e-06,
+      "loss": 0.7425,
+      "step": 9605
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.6321922634293415e-06,
+      "loss": 0.8824,
+      "step": 9606
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.6188136383192651e-06,
+      "loss": 0.9796,
+      "step": 9607
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.6054350132091883e-06,
+      "loss": 0.9165,
+      "step": 9608
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.592056388099112e-06,
+      "loss": 0.7714,
+      "step": 9609
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.5786777629890353e-06,
+      "loss": 0.7861,
+      "step": 9610
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.5652991378789587e-06,
+      "loss": 0.7675,
+      "step": 9611
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.551920512768882e-06,
+      "loss": 0.7867,
+      "step": 9612
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.5385418876588057e-06,
+      "loss": 0.8248,
+      "step": 9613
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.5251632625487289e-06,
+      "loss": 0.7516,
+      "step": 9614
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.5117846374386525e-06,
+      "loss": 0.7175,
+      "step": 9615
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4984060123285759e-06,
+      "loss": 0.5529,
+      "step": 9616
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4850273872184995e-06,
+      "loss": 0.8753,
+      "step": 9617
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4716487621084226e-06,
+      "loss": 0.6222,
+      "step": 9618
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4582701369983462e-06,
+      "loss": 0.637,
+      "step": 9619
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4448915118882696e-06,
+      "loss": 0.5395,
+      "step": 9620
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.431512886778193e-06,
+      "loss": 0.6587,
+      "step": 9621
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4181342616681164e-06,
+      "loss": 0.6536,
+      "step": 9622
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.40475563655804e-06,
+      "loss": 0.5169,
+      "step": 9623
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.3913770114479632e-06,
+      "loss": 0.607,
+      "step": 9624
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.3779983863378868e-06,
+      "loss": 0.5533,
+      "step": 9625
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.3646197612278102e-06,
+      "loss": 0.4895,
+      "step": 9626
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.3512411361177338e-06,
+      "loss": 0.4477,
+      "step": 9627
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.337862511007657e-06,
+      "loss": 0.5016,
+      "step": 9628
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.3244838858975806e-06,
+      "loss": 0.4173,
+      "step": 9629
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.311105260787504e-06,
+      "loss": 0.4342,
+      "step": 9630
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.2977266356774273e-06,
+      "loss": 1.5676,
+      "step": 9631
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.2843480105673507e-06,
+      "loss": 1.2351,
+      "step": 9632
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.2709693854572743e-06,
+      "loss": 1.2043,
+      "step": 9633
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.2575907603471975e-06,
+      "loss": 1.2333,
+      "step": 9634
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.244212135237121e-06,
+      "loss": 1.3693,
+      "step": 9635
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.2308335101270445e-06,
+      "loss": 0.9607,
+      "step": 9636
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.2174548850169677e-06,
+      "loss": 1.107,
+      "step": 9637
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.2040762599068913e-06,
+      "loss": 0.9648,
+      "step": 9638
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.1906976347968147e-06,
+      "loss": 1.0693,
+      "step": 9639
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.1773190096867383e-06,
+      "loss": 0.9978,
+      "step": 9640
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.1639403845766615e-06,
+      "loss": 1.0144,
+      "step": 9641
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.150561759466585e-06,
+      "loss": 0.99,
+      "step": 9642
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.1371831343565084e-06,
+      "loss": 0.9204,
+      "step": 9643
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.1238045092464318e-06,
+      "loss": 0.9571,
+      "step": 9644
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.1104258841363552e-06,
+      "loss": 0.921,
+      "step": 9645
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.0970472590262788e-06,
+      "loss": 0.8917,
+      "step": 9646
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.083668633916202e-06,
+      "loss": 0.8802,
+      "step": 9647
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.0702900088061256e-06,
+      "loss": 1.0523,
+      "step": 9648
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.056911383696049e-06,
+      "loss": 0.8858,
+      "step": 9649
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.0435327585859726e-06,
+      "loss": 0.9236,
+      "step": 9650
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.0301541334758958e-06,
+      "loss": 0.8921,
+      "step": 9651
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.0167755083658194e-06,
+      "loss": 1.0013,
+      "step": 9652
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.0033968832557428e-06,
+      "loss": 0.7711,
+      "step": 9653
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.900182581456662e-07,
+      "loss": 0.9429,
+      "step": 9654
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.766396330355896e-07,
+      "loss": 0.7856,
+      "step": 9655
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.632610079255132e-07,
+      "loss": 0.8945,
+      "step": 9656
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.498823828154365e-07,
+      "loss": 0.7041,
+      "step": 9657
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.365037577053598e-07,
+      "loss": 0.8907,
+      "step": 9658
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.231251325952833e-07,
+      "loss": 0.8547,
+      "step": 9659
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.097465074852067e-07,
+      "loss": 0.8312,
+      "step": 9660
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.963678823751301e-07,
+      "loss": 0.9402,
+      "step": 9661
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.829892572650536e-07,
+      "loss": 0.8204,
+      "step": 9662
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.69610632154977e-07,
+      "loss": 0.621,
+      "step": 9663
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.562320070449005e-07,
+      "loss": 0.7596,
+      "step": 9664
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.428533819348239e-07,
+      "loss": 0.7414,
+      "step": 9665
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.294747568247473e-07,
+      "loss": 0.8141,
+      "step": 9666
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.160961317146708e-07,
+      "loss": 0.7397,
+      "step": 9667
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.027175066045942e-07,
+      "loss": 0.735,
+      "step": 9668
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.893388814945177e-07,
+      "loss": 0.7381,
+      "step": 9669
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.75960256384441e-07,
+      "loss": 0.5865,
+      "step": 9670
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.625816312743644e-07,
+      "loss": 0.5425,
+      "step": 9671
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.492030061642879e-07,
+      "loss": 0.5735,
+      "step": 9672
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.358243810542113e-07,
+      "loss": 0.5775,
+      "step": 9673
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.224457559441348e-07,
+      "loss": 0.5167,
+      "step": 9674
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.090671308340582e-07,
+      "loss": 0.5635,
+      "step": 9675
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.956885057239816e-07,
+      "loss": 0.4392,
+      "step": 9676
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.823098806139051e-07,
+      "loss": 0.4631,
+      "step": 9677
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.689312555038285e-07,
+      "loss": 0.4944,
+      "step": 9678
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.55552630393752e-07,
+      "loss": 0.4798,
+      "step": 9679
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.421740052836754e-07,
+      "loss": 0.4669,
+      "step": 9680
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.287953801735988e-07,
+      "loss": 1.5484,
+      "step": 9681
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.154167550635223e-07,
+      "loss": 1.2068,
+      "step": 9682
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.020381299534456e-07,
+      "loss": 1.1248,
+      "step": 9683
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.886595048433691e-07,
+      "loss": 1.1409,
+      "step": 9684
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.752808797332925e-07,
+      "loss": 1.2358,
+      "step": 9685
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.619022546232159e-07,
+      "loss": 0.9598,
+      "step": 9686
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.485236295131394e-07,
+      "loss": 1.0289,
+      "step": 9687
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.351450044030628e-07,
+      "loss": 1.0486,
+      "step": 9688
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.217663792929863e-07,
+      "loss": 1.0945,
+      "step": 9689
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.083877541829097e-07,
+      "loss": 1.1211,
+      "step": 9690
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.950091290728331e-07,
+      "loss": 1.0167,
+      "step": 9691
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.816305039627566e-07,
+      "loss": 0.8957,
+      "step": 9692
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.682518788526799e-07,
+      "loss": 0.7859,
+      "step": 9693
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.5487325374260336e-07,
+      "loss": 0.9702,
+      "step": 9694
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.414946286325268e-07,
+      "loss": 1.0147,
+      "step": 9695
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.2811600352245024e-07,
+      "loss": 0.9213,
+      "step": 9696
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.1473737841237363e-07,
+      "loss": 0.9927,
+      "step": 9697
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.013587533022971e-07,
+      "loss": 0.8984,
+      "step": 9698
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.879801281922205e-07,
+      "loss": 0.8921,
+      "step": 9699
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.7460150308214396e-07,
+      "loss": 0.8909,
+      "step": 9700
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.612228779720674e-07,
+      "loss": 0.8453,
+      "step": 9701
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.478442528619908e-07,
+      "loss": 1.01,
+      "step": 9702
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.3446562775191424e-07,
+      "loss": 0.8194,
+      "step": 9703
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.210870026418377e-07,
+      "loss": 0.8076,
+      "step": 9704
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.077083775317611e-07,
+      "loss": 0.812,
+      "step": 9705
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.9432975242168457e-07,
+      "loss": 0.8257,
+      "step": 9706
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.8095112731160796e-07,
+      "loss": 0.6693,
+      "step": 9707
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.675725022015314e-07,
+      "loss": 0.7341,
+      "step": 9708
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.5419387709145485e-07,
+      "loss": 0.7275,
+      "step": 9709
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.408152519813783e-07,
+      "loss": 0.7139,
+      "step": 9710
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.2743662687130168e-07,
+      "loss": 0.5855,
+      "step": 9711
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.1405800176122512e-07,
+      "loss": 0.6136,
+      "step": 9712
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.0067937665114854e-07,
+      "loss": 0.5685,
+      "step": 9713
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8730075154107198e-07,
+      "loss": 0.5891,
+      "step": 9714
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.739221264309954e-07,
+      "loss": 0.4881,
+      "step": 9715
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.6054350132091884e-07,
+      "loss": 0.5283,
+      "step": 9716
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4716487621084228e-07,
+      "loss": 0.5305,
+      "step": 9717
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.337862511007657e-07,
+      "loss": 0.4957,
+      "step": 9718
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.2040762599068914e-07,
+      "loss": 0.4841,
+      "step": 9719
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.0702900088061256e-07,
+      "loss": 0.4724,
+      "step": 9720
+    },
+    {
+      "epoch": 3.0,
+      "step": 9720,
+      "total_flos": 0.0,
+      "train_loss": 1.7970677693255643,
+      "train_runtime": 95217.2299,
+      "train_samples_per_second": 6.534,
+      "train_steps_per_second": 0.102
+    }
+  ],
+  "max_steps": 9720,
+  "num_train_epochs": 3,
+  "total_flos": 0.0,
+  "trial_name": null,
+  "trial_params": null
+}