diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,18133 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.998758226747796,
+  "global_step": 3018,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.197802197802198e-07,
+      "loss": 1.857,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.395604395604396e-07,
+      "loss": 1.9076,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.593406593406594e-07,
+      "loss": 1.5819,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.791208791208792e-07,
+      "loss": 1.5108,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.098901098901099e-06,
+      "loss": 1.5007,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3186813186813187e-06,
+      "loss": 1.5128,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5384615384615387e-06,
+      "loss": 1.5127,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7582417582417585e-06,
+      "loss": 1.5531,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9780219780219782e-06,
+      "loss": 1.5101,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.197802197802198e-06,
+      "loss": 1.5169,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4175824175824177e-06,
+      "loss": 1.5145,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6373626373626375e-06,
+      "loss": 1.465,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 1.5203,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 1.593,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2967032967032968e-06,
+      "loss": 1.5754,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.516483516483517e-06,
+      "loss": 1.4178,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7362637362637367e-06,
+      "loss": 1.5023,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9560439560439565e-06,
+      "loss": 1.4339,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.175824175824177e-06,
+      "loss": 1.3713,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.395604395604396e-06,
+      "loss": 1.4162,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 1.3654,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8351648351648355e-06,
+      "loss": 1.368,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.054945054945055e-06,
+      "loss": 1.4825,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.274725274725275e-06,
+      "loss": 1.3857,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.494505494505495e-06,
+      "loss": 1.3848,
+      "step": 25
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 1.4064,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.934065934065935e-06,
+      "loss": 1.4408,
+      "step": 27
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.153846153846155e-06,
+      "loss": 1.4052,
+      "step": 28
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.373626373626373e-06,
+      "loss": 1.4642,
+      "step": 29
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.5934065934065935e-06,
+      "loss": 1.3062,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.813186813186814e-06,
+      "loss": 1.3899,
+      "step": 31
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.032967032967034e-06,
+      "loss": 1.3341,
+      "step": 32
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.252747252747253e-06,
+      "loss": 1.2997,
+      "step": 33
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.472527472527473e-06,
+      "loss": 1.3662,
+      "step": 34
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 1.3904,
+      "step": 35
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.912087912087913e-06,
+      "loss": 1.2869,
+      "step": 36
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.131868131868132e-06,
+      "loss": 1.3415,
+      "step": 37
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.351648351648353e-06,
+      "loss": 1.3412,
+      "step": 38
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 1.369,
+      "step": 39
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.791208791208792e-06,
+      "loss": 1.3108,
+      "step": 40
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.010989010989011e-06,
+      "loss": 1.317,
+      "step": 41
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 1.3668,
+      "step": 42
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.450549450549452e-06,
+      "loss": 1.2865,
+      "step": 43
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.670329670329671e-06,
+      "loss": 1.3577,
+      "step": 44
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.890109890109892e-06,
+      "loss": 1.277,
+      "step": 45
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.010989010989011e-05,
+      "loss": 1.345,
+      "step": 46
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0329670329670332e-05,
+      "loss": 1.3789,
+      "step": 47
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.054945054945055e-05,
+      "loss": 1.3709,
+      "step": 48
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.076923076923077e-05,
+      "loss": 1.2436,
+      "step": 49
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.098901098901099e-05,
+      "loss": 1.2523,
+      "step": 50
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.120879120879121e-05,
+      "loss": 1.2135,
+      "step": 51
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 1.278,
+      "step": 52
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.164835164835165e-05,
+      "loss": 1.2677,
+      "step": 53
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.186813186813187e-05,
+      "loss": 1.365,
+      "step": 54
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2087912087912089e-05,
+      "loss": 1.3791,
+      "step": 55
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.230769230769231e-05,
+      "loss": 1.2029,
+      "step": 56
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2527472527472529e-05,
+      "loss": 1.5197,
+      "step": 57
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2747252747252747e-05,
+      "loss": 1.3553,
+      "step": 58
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.296703296703297e-05,
+      "loss": 1.3865,
+      "step": 59
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3186813186813187e-05,
+      "loss": 1.1873,
+      "step": 60
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3406593406593406e-05,
+      "loss": 1.2704,
+      "step": 61
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3626373626373627e-05,
+      "loss": 1.2286,
+      "step": 62
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 1.2615,
+      "step": 63
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4065934065934068e-05,
+      "loss": 1.4381,
+      "step": 64
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4285714285714287e-05,
+      "loss": 1.2671,
+      "step": 65
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4505494505494506e-05,
+      "loss": 1.3389,
+      "step": 66
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4725274725274727e-05,
+      "loss": 1.284,
+      "step": 67
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4945054945054947e-05,
+      "loss": 1.2456,
+      "step": 68
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5164835164835166e-05,
+      "loss": 1.2233,
+      "step": 69
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 1.256,
+      "step": 70
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5604395604395605e-05,
+      "loss": 1.2236,
+      "step": 71
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5824175824175826e-05,
+      "loss": 1.3757,
+      "step": 72
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6043956043956047e-05,
+      "loss": 1.1794,
+      "step": 73
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6263736263736265e-05,
+      "loss": 1.1644,
+      "step": 74
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6483516483516486e-05,
+      "loss": 1.2276,
+      "step": 75
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6703296703296707e-05,
+      "loss": 1.2681,
+      "step": 76
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6923076923076924e-05,
+      "loss": 1.3798,
+      "step": 77
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 1.26,
+      "step": 78
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7362637362637363e-05,
+      "loss": 1.3156,
+      "step": 79
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7582417582417584e-05,
+      "loss": 1.2937,
+      "step": 80
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.78021978021978e-05,
+      "loss": 1.3784,
+      "step": 81
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8021978021978023e-05,
+      "loss": 1.2026,
+      "step": 82
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8241758241758244e-05,
+      "loss": 1.1973,
+      "step": 83
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8461538461538465e-05,
+      "loss": 1.2322,
+      "step": 84
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8681318681318682e-05,
+      "loss": 1.3839,
+      "step": 85
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8901098901098903e-05,
+      "loss": 1.2404,
+      "step": 86
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9120879120879124e-05,
+      "loss": 1.217,
+      "step": 87
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9340659340659342e-05,
+      "loss": 1.2182,
+      "step": 88
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9560439560439563e-05,
+      "loss": 1.2798,
+      "step": 89
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780219780219784e-05,
+      "loss": 1.1804,
+      "step": 90
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 1.1991,
+      "step": 91
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999994239976396e-05,
+      "loss": 1.2834,
+      "step": 92
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999976959912216e-05,
+      "loss": 1.2375,
+      "step": 93
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999948159827375e-05,
+      "loss": 1.3955,
+      "step": 94
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999990783975504e-05,
+      "loss": 1.2823,
+      "step": 95
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999855999741666e-05,
+      "loss": 1.2728,
+      "step": 96
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999979263984697e-05,
+      "loss": 1.2174,
+      "step": 97
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999971776014395e-05,
+      "loss": 1.245,
+      "step": 98
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999631360718856e-05,
+      "loss": 1.3856,
+      "step": 99
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999953344167123e-05,
+      "loss": 1.1501,
+      "step": 100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999424003113875e-05,
+      "loss": 1.1947,
+      "step": 101
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999303045172858e-05,
+      "loss": 1.1054,
+      "step": 102
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999917056798753e-05,
+      "loss": 1.352,
+      "step": 103
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.99990265717105e-05,
+      "loss": 1.318,
+      "step": 104
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998871056507658e-05,
+      "loss": 1.2017,
+      "step": 105
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999870402255815e-05,
+      "loss": 1.1707,
+      "step": 106
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999852547005441e-05,
+      "loss": 1.1385,
+      "step": 107
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998335399202126e-05,
+      "loss": 1.2419,
+      "step": 108
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998133810220256e-05,
+      "loss": 1.1817,
+      "step": 109
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997920703341043e-05,
+      "loss": 1.2494,
+      "step": 110
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997696078809977e-05,
+      "loss": 1.2272,
+      "step": 111
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999745993688583e-05,
+      "loss": 1.1943,
+      "step": 112
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999721227784064e-05,
+      "loss": 1.1794,
+      "step": 113
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996953101959707e-05,
+      "loss": 1.3727,
+      "step": 114
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996682409541604e-05,
+      "loss": 1.203,
+      "step": 115
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996400200898177e-05,
+      "loss": 1.2031,
+      "step": 116
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996106476354527e-05,
+      "loss": 1.192,
+      "step": 117
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999580123624902e-05,
+      "loss": 1.182,
+      "step": 118
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995484480933305e-05,
+      "loss": 1.2069,
+      "step": 119
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995156210772278e-05,
+      "loss": 1.2995,
+      "step": 120
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999481642614411e-05,
+      "loss": 1.2198,
+      "step": 121
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999446512744023e-05,
+      "loss": 1.1378,
+      "step": 122
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994102315065347e-05,
+      "loss": 1.2325,
+      "step": 123
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993727989437412e-05,
+      "loss": 1.2717,
+      "step": 124
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993342150987657e-05,
+      "loss": 1.2272,
+      "step": 125
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992944800160563e-05,
+      "loss": 1.1739,
+      "step": 126
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992535937413885e-05,
+      "loss": 1.2907,
+      "step": 127
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992115563218633e-05,
+      "loss": 1.3453,
+      "step": 128
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999168367805908e-05,
+      "loss": 1.2386,
+      "step": 129
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999124028243276e-05,
+      "loss": 1.1527,
+      "step": 130
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990785376850468e-05,
+      "loss": 1.1981,
+      "step": 131
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990318961836256e-05,
+      "loss": 1.184,
+      "step": 132
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989841037927437e-05,
+      "loss": 1.2159,
+      "step": 133
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989351605674576e-05,
+      "loss": 1.2839,
+      "step": 134
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988850665641513e-05,
+      "loss": 1.1864,
+      "step": 135
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9988338218405325e-05,
+      "loss": 1.1975,
+      "step": 136
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987814264556352e-05,
+      "loss": 1.1894,
+      "step": 137
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987278804698197e-05,
+      "loss": 1.2011,
+      "step": 138
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998673183944771e-05,
+      "loss": 1.3174,
+      "step": 139
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986173369434997e-05,
+      "loss": 1.2157,
+      "step": 140
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998560339530342e-05,
+      "loss": 1.1173,
+      "step": 141
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998502191770959e-05,
+      "loss": 1.2038,
+      "step": 142
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998442893732337e-05,
+      "loss": 1.1725,
+      "step": 143
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983824454827882e-05,
+      "loss": 1.1229,
+      "step": 144
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983208470919487e-05,
+      "loss": 1.2739,
+      "step": 145
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9982580986307806e-05,
+      "loss": 1.2375,
+      "step": 146
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99819420017157e-05,
+      "loss": 1.2171,
+      "step": 147
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981291517879285e-05,
+      "loss": 1.3107,
+      "step": 148
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998062953554792e-05,
+      "loss": 1.2166,
+      "step": 149
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979956055484214e-05,
+      "loss": 1.1476,
+      "step": 150
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979271078464014e-05,
+      "loss": 1.1539,
+      "step": 151
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978574605276423e-05,
+      "loss": 1.2386,
+      "step": 152
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977866636723778e-05,
+      "loss": 1.2683,
+      "step": 153
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977147173621663e-05,
+      "loss": 1.1482,
+      "step": 154
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99764162167989e-05,
+      "loss": 1.1144,
+      "step": 155
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9975673767097562e-05,
+      "loss": 1.2814,
+      "step": 156
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997491982537295e-05,
+      "loss": 1.11,
+      "step": 157
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974154392493604e-05,
+      "loss": 1.1794,
+      "step": 158
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973377469341317e-05,
+      "loss": 1.2871,
+      "step": 159
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99725890568111e-05,
+      "loss": 1.1077,
+      "step": 160
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971789155811203e-05,
+      "loss": 1.1633,
+      "step": 161
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970977767263128e-05,
+      "loss": 1.2092,
+      "step": 162
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970154892101593e-05,
+      "loss": 1.2191,
+      "step": 163
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969320531274555e-05,
+      "loss": 1.1391,
+      "step": 164
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99684746857432e-05,
+      "loss": 1.2027,
+      "step": 165
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996761735648195e-05,
+      "loss": 1.1539,
+      "step": 166
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996674854447844e-05,
+      "loss": 1.0734,
+      "step": 167
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996586825073356e-05,
+      "loss": 1.116,
+      "step": 168
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996497647626141e-05,
+      "loss": 1.1846,
+      "step": 169
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996407322208931e-05,
+      "loss": 1.1519,
+      "step": 170
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963158489257822e-05,
+      "loss": 1.1231,
+      "step": 171
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996223227882072e-05,
+      "loss": 1.1637,
+      "step": 172
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961294591845e-05,
+      "loss": 1.1765,
+      "step": 173
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960345429410887e-05,
+      "loss": 1.2632,
+      "step": 174
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959384792611812e-05,
+      "loss": 1.3243,
+      "step": 175
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995841268255444e-05,
+      "loss": 1.094,
+      "step": 176
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957429100358648e-05,
+      "loss": 1.0604,
+      "step": 177
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956434047157523e-05,
+      "loss": 1.2406,
+      "step": 178
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995542752409737e-05,
+      "loss": 1.1936,
+      "step": 179
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995440953233771e-05,
+      "loss": 1.2459,
+      "step": 180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953380073051282e-05,
+      "loss": 1.1015,
+      "step": 181
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952339147424013e-05,
+      "loss": 1.0389,
+      "step": 182
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951286756655065e-05,
+      "loss": 1.1056,
+      "step": 183
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9950222901956794e-05,
+      "loss": 1.2455,
+      "step": 184
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949147584554768e-05,
+      "loss": 1.1464,
+      "step": 185
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948060805687755e-05,
+      "loss": 1.2003,
+      "step": 186
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946962566607725e-05,
+      "loss": 1.1246,
+      "step": 187
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9945852868579863e-05,
+      "loss": 1.1303,
+      "step": 188
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944731712882546e-05,
+      "loss": 1.2277,
+      "step": 189
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943599100807343e-05,
+      "loss": 1.1725,
+      "step": 190
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942455033659034e-05,
+      "loss": 1.2074,
+      "step": 191
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994129951275559e-05,
+      "loss": 1.1627,
+      "step": 192
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940132539428175e-05,
+      "loss": 1.2714,
+      "step": 193
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938954115021146e-05,
+      "loss": 1.1761,
+      "step": 194
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937764240892056e-05,
+      "loss": 1.1329,
+      "step": 195
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993656291841165e-05,
+      "loss": 1.1667,
+      "step": 196
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9935350148963848e-05,
+      "loss": 1.1947,
+      "step": 197
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993412593394577e-05,
+      "loss": 1.17,
+      "step": 198
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993289027476772e-05,
+      "loss": 1.2174,
+      "step": 199
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993164317285318e-05,
+      "loss": 1.1878,
+      "step": 200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993038462963882e-05,
+      "loss": 1.188,
+      "step": 201
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9929114646574483e-05,
+      "loss": 1.1251,
+      "step": 202
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.99278332251232e-05,
+      "loss": 1.1455,
+      "step": 203
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926540366761176e-05,
+      "loss": 1.1472,
+      "step": 204
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9925236072977786e-05,
+      "loss": 1.0667,
+      "step": 205
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992392034527558e-05,
+      "loss": 1.1779,
+      "step": 206
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992259318517029e-05,
+      "loss": 1.1302,
+      "step": 207
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921254594190806e-05,
+      "loss": 1.1847,
+      "step": 208
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991990457387919e-05,
+      "loss": 1.1763,
+      "step": 209
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9918543125790678e-05,
+      "loss": 1.1252,
+      "step": 210
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917170251493658e-05,
+      "loss": 1.1956,
+      "step": 211
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9915785952569694e-05,
+      "loss": 1.1405,
+      "step": 212
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914390230613495e-05,
+      "loss": 1.1472,
+      "step": 213
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9912983087232953e-05,
+      "loss": 1.1955,
+      "step": 214
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991156452404909e-05,
+      "loss": 1.1236,
+      "step": 215
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9910134542696104e-05,
+      "loss": 1.086,
+      "step": 216
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9908693144821342e-05,
+      "loss": 1.1708,
+      "step": 217
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9907240332085295e-05,
+      "loss": 1.1823,
+      "step": 218
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9905776106161617e-05,
+      "loss": 1.1018,
+      "step": 219
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9904300468737105e-05,
+      "loss": 1.2135,
+      "step": 220
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990281342151169e-05,
+      "loss": 1.1453,
+      "step": 221
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9901314966198466e-05,
+      "loss": 1.0699,
+      "step": 222
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9899805104523655e-05,
+      "loss": 1.1232,
+      "step": 223
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989828383822663e-05,
+      "loss": 1.2512,
+      "step": 224
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989675116905989e-05,
+      "loss": 1.2023,
+      "step": 225
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9895207098789088e-05,
+      "loss": 1.1602,
+      "step": 226
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989365162919299e-05,
+      "loss": 1.1019,
+      "step": 227
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989208476206351e-05,
+      "loss": 1.1236,
+      "step": 228
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9890506499205683e-05,
+      "loss": 1.0919,
+      "step": 229
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888916842437677e-05,
+      "loss": 1.2479,
+      "step": 230
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9887315793590782e-05,
+      "loss": 1.0864,
+      "step": 231
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9885703354509418e-05,
+      "loss": 1.2032,
+      "step": 232
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9884079527051117e-05,
+      "loss": 1.2054,
+      "step": 233
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9882444313086544e-05,
+      "loss": 1.1226,
+      "step": 234
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9880797714499467e-05,
+      "loss": 1.1817,
+      "step": 235
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9879139733186773e-05,
+      "loss": 1.1162,
+      "step": 236
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987747037105847e-05,
+      "loss": 1.2191,
+      "step": 237
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987578963003767e-05,
+      "loss": 1.1444,
+      "step": 238
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9874097512060594e-05,
+      "loss": 1.1144,
+      "step": 239
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872394019076568e-05,
+      "loss": 1.2105,
+      "step": 240
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9870679153048024e-05,
+      "loss": 1.141,
+      "step": 241
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9868952915950497e-05,
+      "loss": 1.1884,
+      "step": 242
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986721530977262e-05,
+      "loss": 1.0902,
+      "step": 243
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986546633651613e-05,
+      "loss": 1.1493,
+      "step": 244
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9863705998195837e-05,
+      "loss": 1.1195,
+      "step": 245
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9861934296839674e-05,
+      "loss": 1.1179,
+      "step": 246
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9860151234488643e-05,
+      "loss": 1.1495,
+      "step": 247
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9858356813196843e-05,
+      "loss": 1.1065,
+      "step": 248
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9856551035031448e-05,
+      "loss": 1.0966,
+      "step": 249
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9854733902072733e-05,
+      "loss": 1.0773,
+      "step": 250
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985290541641404e-05,
+      "loss": 1.0843,
+      "step": 251
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985106558016179e-05,
+      "loss": 1.1145,
+      "step": 252
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9849214395435486e-05,
+      "loss": 1.0365,
+      "step": 253
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9847351864367702e-05,
+      "loss": 1.0851,
+      "step": 254
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984547798910408e-05,
+      "loss": 1.1416,
+      "step": 255
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9843592771803342e-05,
+      "loss": 1.0686,
+      "step": 256
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9841696214637254e-05,
+      "loss": 1.2181,
+      "step": 257
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983978831979067e-05,
+      "loss": 1.159,
+      "step": 258
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9837869089461482e-05,
+      "loss": 1.1045,
+      "step": 259
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983593852586066e-05,
+      "loss": 1.2073,
+      "step": 260
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9833996631212224e-05,
+      "loss": 1.108,
+      "step": 261
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983204340775324e-05,
+      "loss": 1.0463,
+      "step": 262
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983007885773384e-05,
+      "loss": 1.2549,
+      "step": 263
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9828102983417188e-05,
+      "loss": 1.1577,
+      "step": 264
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.98261157870795e-05,
+      "loss": 1.0691,
+      "step": 265
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9824117271010033e-05,
+      "loss": 1.2491,
+      "step": 266
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9822107437511095e-05,
+      "loss": 1.1905,
+      "step": 267
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.982008628889802e-05,
+      "loss": 1.1256,
+      "step": 268
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9818053827499182e-05,
+      "loss": 1.0609,
+      "step": 269
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9816010055655982e-05,
+      "loss": 1.0189,
+      "step": 270
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9813954975722857e-05,
+      "loss": 1.0823,
+      "step": 271
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981188859006727e-05,
+      "loss": 1.0477,
+      "step": 272
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9809810901069708e-05,
+      "loss": 1.0465,
+      "step": 273
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9807721911123672e-05,
+      "loss": 1.0858,
+      "step": 274
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9805621622635698e-05,
+      "loss": 1.137,
+      "step": 275
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980351003802532e-05,
+      "loss": 1.0943,
+      "step": 276
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9801387159725095e-05,
+      "loss": 1.1346,
+      "step": 277
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9799252990180594e-05,
+      "loss": 1.2057,
+      "step": 278
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9797107531850388e-05,
+      "loss": 1.1946,
+      "step": 279
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979495078720605e-05,
+      "loss": 1.1656,
+      "step": 280
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9792782758732167e-05,
+      "loss": 1.0259,
+      "step": 281
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9790603448926315e-05,
+      "loss": 1.0535,
+      "step": 282
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9788412860299067e-05,
+      "loss": 1.1863,
+      "step": 283
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9786210995373998e-05,
+      "loss": 1.1006,
+      "step": 284
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9783997856687657e-05,
+      "loss": 1.1358,
+      "step": 285
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.97817734467896e-05,
+      "loss": 1.1176,
+      "step": 286
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9779537768242353e-05,
+      "loss": 1.1386,
+      "step": 287
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9777290823621428e-05,
+      "loss": 1.0769,
+      "step": 288
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9775032615515313e-05,
+      "loss": 1.1898,
+      "step": 289
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9772763146525482e-05,
+      "loss": 1.1079,
+      "step": 290
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9770482419266368e-05,
+      "loss": 1.023,
+      "step": 291
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976819043636538e-05,
+      "loss": 1.2183,
+      "step": 292
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9765887200462898e-05,
+      "loss": 1.1046,
+      "step": 293
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9763572714212257e-05,
+      "loss": 1.0485,
+      "step": 294
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9761246980279757e-05,
+      "loss": 1.1207,
+      "step": 295
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975891000134465e-05,
+      "loss": 1.185,
+      "step": 296
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9756561780099153e-05,
+      "loss": 1.0975,
+      "step": 297
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9754202319248422e-05,
+      "loss": 1.0578,
+      "step": 298
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9751831621510575e-05,
+      "loss": 1.0833,
+      "step": 299
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974944968961666e-05,
+      "loss": 1.1906,
+      "step": 300
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9747056526310677e-05,
+      "loss": 1.0568,
+      "step": 301
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974465213434956e-05,
+      "loss": 1.1479,
+      "step": 302
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9742236516503182e-05,
+      "loss": 1.0679,
+      "step": 303
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9739809675554344e-05,
+      "loss": 1.1251,
+      "step": 304
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973737161429878e-05,
+      "loss": 1.0486,
+      "step": 305
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9734922335545146e-05,
+      "loss": 1.0871,
+      "step": 306
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9732461842115024e-05,
+      "loss": 1.1149,
+      "step": 307
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9729990136842914e-05,
+      "loss": 1.0299,
+      "step": 308
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9727507222576233e-05,
+      "loss": 0.9818,
+      "step": 309
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9725013102175305e-05,
+      "loss": 1.1297,
+      "step": 310
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9722507778513374e-05,
+      "loss": 1.0917,
+      "step": 311
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9719991254476585e-05,
+      "loss": 1.0904,
+      "step": 312
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971746353296398e-05,
+      "loss": 1.1798,
+      "step": 313
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9714924616887515e-05,
+      "loss": 1.1828,
+      "step": 314
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9712374509172023e-05,
+      "loss": 1.083,
+      "step": 315
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.970981321275525e-05,
+      "loss": 1.2174,
+      "step": 316
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9707240730587814e-05,
+      "loss": 1.1087,
+      "step": 317
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970465706563323e-05,
+      "loss": 1.0329,
+      "step": 318
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9702062220867888e-05,
+      "loss": 1.0912,
+      "step": 319
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969945619928107e-05,
+      "loss": 1.0217,
+      "step": 320
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9696839003874916e-05,
+      "loss": 1.0584,
+      "step": 321
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9694210637664457e-05,
+      "loss": 1.1692,
+      "step": 322
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9691571103677575e-05,
+      "loss": 1.0735,
+      "step": 323
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9688920404955026e-05,
+      "loss": 1.0372,
+      "step": 324
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968625854455043e-05,
+      "loss": 1.0784,
+      "step": 325
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968358552553027e-05,
+      "loss": 1.2163,
+      "step": 326
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968090135097386e-05,
+      "loss": 1.0093,
+      "step": 327
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9678206023973398e-05,
+      "loss": 1.0259,
+      "step": 328
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9675499547633904e-05,
+      "loss": 1.0726,
+      "step": 329
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967278192507325e-05,
+      "loss": 1.0878,
+      "step": 330
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967005315942216e-05,
+      "loss": 1.0033,
+      "step": 331
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9667313253824173e-05,
+      "loss": 1.0328,
+      "step": 332
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9664562211435684e-05,
+      "loss": 1.0887,
+      "step": 333
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9661800035425898e-05,
+      "loss": 1.1236,
+      "step": 334
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9659026728976863e-05,
+      "loss": 1.0661,
+      "step": 335
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9656242295283436e-05,
+      "loss": 1.1136,
+      "step": 336
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9653446737553298e-05,
+      "loss": 1.0232,
+      "step": 337
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.965064005900694e-05,
+      "loss": 1.1146,
+      "step": 338
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9647822262877683e-05,
+      "loss": 1.0263,
+      "step": 339
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964499335241163e-05,
+      "loss": 1.0896,
+      "step": 340
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9642153330867696e-05,
+      "loss": 1.0844,
+      "step": 341
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963930220151761e-05,
+      "loss": 1.0391,
+      "step": 342
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9636439967645883e-05,
+      "loss": 1.0262,
+      "step": 343
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9633566632549817e-05,
+      "loss": 1.0168,
+      "step": 344
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9630682199539515e-05,
+      "loss": 1.1468,
+      "step": 345
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962778667193785e-05,
+      "loss": 1.017,
+      "step": 346
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962488005308049e-05,
+      "loss": 1.0289,
+      "step": 347
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9621962346315867e-05,
+      "loss": 1.1109,
+      "step": 348
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.96190335550052e-05,
+      "loss": 1.0799,
+      "step": 349
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9616093682522463e-05,
+      "loss": 0.9757,
+      "step": 350
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9613142732254408e-05,
+      "loss": 1.0068,
+      "step": 351
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9610180707600546e-05,
+      "loss": 1.1785,
+      "step": 352
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9607207611973136e-05,
+      "loss": 1.0131,
+      "step": 353
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9604223448797206e-05,
+      "loss": 1.0776,
+      "step": 354
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960122822151052e-05,
+      "loss": 1.0684,
+      "step": 355
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9598221933563596e-05,
+      "loss": 1.0223,
+      "step": 356
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9595204588419693e-05,
+      "loss": 1.0459,
+      "step": 357
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9592176189554806e-05,
+      "loss": 0.9484,
+      "step": 358
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9589136740457666e-05,
+      "loss": 1.0816,
+      "step": 359
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958608624462973e-05,
+      "loss": 1.0486,
+      "step": 360
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9583024705585185e-05,
+      "loss": 1.1275,
+      "step": 361
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957995212685094e-05,
+      "loss": 1.1745,
+      "step": 362
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9576868511966617e-05,
+      "loss": 1.061,
+      "step": 363
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957377386448456e-05,
+      "loss": 1.0522,
+      "step": 364
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957066818796981e-05,
+      "loss": 1.0947,
+      "step": 365
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9567551486000127e-05,
+      "loss": 0.9368,
+      "step": 366
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9564423762165963e-05,
+      "loss": 1.1541,
+      "step": 367
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9561285020070474e-05,
+      "loss": 1.0413,
+      "step": 368
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9558135263329503e-05,
+      "loss": 1.0287,
+      "step": 369
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9554974495571585e-05,
+      "loss": 1.0976,
+      "step": 370
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955180272043794e-05,
+      "loss": 1.1347,
+      "step": 371
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954861994158247e-05,
+      "loss": 0.9774,
+      "step": 372
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9545426162671748e-05,
+      "loss": 1.0629,
+      "step": 373
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9542221387385024e-05,
+      "loss": 1.0006,
+      "step": 374
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9539005619414216e-05,
+      "loss": 1.0564,
+      "step": 375
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9535778862463895e-05,
+      "loss": 1.0578,
+      "step": 376
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953254112025131e-05,
+      "loss": 1.1175,
+      "step": 377
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9529292396506356e-05,
+      "loss": 1.0151,
+      "step": 378
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952603269497157e-05,
+      "loss": 1.1013,
+      "step": 379
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952276201940215e-05,
+      "loss": 1.1345,
+      "step": 380
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9519480373565925e-05,
+      "loss": 1.014,
+      "step": 381
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951618776124337e-05,
+      "loss": 1.0081,
+      "step": 382
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951288418622759e-05,
+      "loss": 1.0605,
+      "step": 383
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9509569652324317e-05,
+      "loss": 1.0862,
+      "step": 384
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9506244163351913e-05,
+      "loss": 1.0538,
+      "step": 385
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950290772314135e-05,
+      "loss": 1.0704,
+      "step": 386
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.949956033553623e-05,
+      "loss": 0.9881,
+      "step": 387
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949620200439276e-05,
+      "loss": 1.0917,
+      "step": 388
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9492832733579747e-05,
+      "loss": 1.0544,
+      "step": 389
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948945252697861e-05,
+      "loss": 1.0591,
+      "step": 390
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9486061388483363e-05,
+      "loss": 1.0396,
+      "step": 391
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9482659322000615e-05,
+      "loss": 1.0908,
+      "step": 392
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947924633144956e-05,
+      "loss": 1.0952,
+      "step": 393
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947582242076198e-05,
+      "loss": 1.0547,
+      "step": 394
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9472387593882238e-05,
+      "loss": 1.0388,
+      "step": 395
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946894185476727e-05,
+      "loss": 1.067,
+      "step": 396
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9465485207386582e-05,
+      "loss": 1.1939,
+      "step": 397
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.946201765572225e-05,
+      "loss": 1.1885,
+      "step": 398
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945853920376891e-05,
+      "loss": 1.1151,
+      "step": 399
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9455049855533753e-05,
+      "loss": 1.0552,
+      "step": 400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9451549615036524e-05,
+      "loss": 1.0787,
+      "step": 401
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944803848630952e-05,
+      "loss": 1.0536,
+      "step": 402
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944451647339758e-05,
+      "loss": 1.0354,
+      "step": 403
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9440983580358074e-05,
+      "loss": 1.1789,
+      "step": 404
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943743981126091e-05,
+      "loss": 1.0837,
+      "step": 405
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9433885170188532e-05,
+      "loss": 1.036,
+      "step": 406
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9430319661235906e-05,
+      "loss": 1.056,
+      "step": 407
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9426743288510504e-05,
+      "loss": 1.1119,
+      "step": 408
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942315605613233e-05,
+      "loss": 1.12,
+      "step": 409
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.94195579682339e-05,
+      "loss": 1.2406,
+      "step": 410
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9415949028960218e-05,
+      "loss": 1.0051,
+      "step": 411
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9412329242468805e-05,
+      "loss": 1.0974,
+      "step": 412
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9408698612929667e-05,
+      "loss": 1.1008,
+      "step": 413
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9405057144525312e-05,
+      "loss": 1.1607,
+      "step": 414
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9401404841450727e-05,
+      "loss": 1.0294,
+      "step": 415
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9397741707913377e-05,
+      "loss": 1.0452,
+      "step": 416
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9394067748133215e-05,
+      "loss": 1.0351,
+      "step": 417
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.939038296634266e-05,
+      "loss": 1.015,
+      "step": 418
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9386687366786598e-05,
+      "loss": 1.0397,
+      "step": 419
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9382980953722374e-05,
+      "loss": 1.1448,
+      "step": 420
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9379263731419792e-05,
+      "loss": 1.0542,
+      "step": 421
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937553570416112e-05,
+      "loss": 1.1629,
+      "step": 422
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9371796876241048e-05,
+      "loss": 1.0762,
+      "step": 423
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9368047251966735e-05,
+      "loss": 0.9923,
+      "step": 424
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936428683565776e-05,
+      "loss": 1.1327,
+      "step": 425
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9360515631646143e-05,
+      "loss": 1.0868,
+      "step": 426
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.935673364427633e-05,
+      "loss": 0.9833,
+      "step": 427
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9352940877905186e-05,
+      "loss": 1.0613,
+      "step": 428
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9349137336901993e-05,
+      "loss": 1.0324,
+      "step": 429
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9345323025648455e-05,
+      "loss": 1.1215,
+      "step": 430
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934149794853867e-05,
+      "loss": 0.9763,
+      "step": 431
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9337662109979153e-05,
+      "loss": 1.0216,
+      "step": 432
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.93338155143888e-05,
+      "loss": 1.0124,
+      "step": 433
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932995816619891e-05,
+      "loss": 1.051,
+      "step": 434
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9326090069853168e-05,
+      "loss": 1.0046,
+      "step": 435
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9322211229807634e-05,
+      "loss": 1.0407,
+      "step": 436
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9318321650530758e-05,
+      "loss": 0.9982,
+      "step": 437
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9314421336503347e-05,
+      "loss": 0.9923,
+      "step": 438
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9310510292218586e-05,
+      "loss": 1.0008,
+      "step": 439
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9306588522182012e-05,
+      "loss": 1.0143,
+      "step": 440
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9302656030911526e-05,
+      "loss": 0.9197,
+      "step": 441
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929871282293738e-05,
+      "loss": 1.0706,
+      "step": 442
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9294758902802154e-05,
+      "loss": 1.0828,
+      "step": 443
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.92907942750608e-05,
+      "loss": 0.9735,
+      "step": 444
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9286818944280575e-05,
+      "loss": 1.0318,
+      "step": 445
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928283291504109e-05,
+      "loss": 1.0768,
+      "step": 446
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.927883619193426e-05,
+      "loss": 1.038,
+      "step": 447
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.927482877956433e-05,
+      "loss": 1.0843,
+      "step": 448
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9270810682547862e-05,
+      "loss": 1.0814,
+      "step": 449
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9266781905513723e-05,
+      "loss": 1.0507,
+      "step": 450
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926274245310308e-05,
+      "loss": 1.0202,
+      "step": 451
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9258692329969403e-05,
+      "loss": 1.1054,
+      "step": 452
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925463154077845e-05,
+      "loss": 0.9717,
+      "step": 453
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9250560090208276e-05,
+      "loss": 0.9705,
+      "step": 454
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9246477982949202e-05,
+      "loss": 1.054,
+      "step": 455
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9242385223703844e-05,
+      "loss": 1.1278,
+      "step": 456
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9238281817187074e-05,
+      "loss": 1.0761,
+      "step": 457
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9234167768126034e-05,
+      "loss": 1.0515,
+      "step": 458
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9230043081260138e-05,
+      "loss": 1.0214,
+      "step": 459
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9225907761341034e-05,
+      "loss": 1.0151,
+      "step": 460
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9221761813132635e-05,
+      "loss": 0.9832,
+      "step": 461
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9217605241411092e-05,
+      "loss": 1.0644,
+      "step": 462
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9213438050964795e-05,
+      "loss": 1.107,
+      "step": 463
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920926024659437e-05,
+      "loss": 0.966,
+      "step": 464
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920507183311266e-05,
+      "loss": 0.9919,
+      "step": 465
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9200872815344748e-05,
+      "loss": 1.0665,
+      "step": 466
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9196663198127916e-05,
+      "loss": 1.0623,
+      "step": 467
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9192442986311657e-05,
+      "loss": 1.0055,
+      "step": 468
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9188212184757682e-05,
+      "loss": 1.0901,
+      "step": 469
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.91839707983399e-05,
+      "loss": 1.1639,
+      "step": 470
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9179718831944395e-05,
+      "loss": 1.0335,
+      "step": 471
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9175456290469464e-05,
+      "loss": 1.0705,
+      "step": 472
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9171183178825566e-05,
+      "loss": 0.9857,
+      "step": 473
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916689950193535e-05,
+      "loss": 1.015,
+      "step": 474
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916260526473363e-05,
+      "loss": 1.0463,
+      "step": 475
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9158300472167396e-05,
+      "loss": 0.9718,
+      "step": 476
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915398512919578e-05,
+      "loss": 0.9486,
+      "step": 477
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9149659240790076e-05,
+      "loss": 0.9973,
+      "step": 478
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9145322811933733e-05,
+      "loss": 1.1265,
+      "step": 479
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914097584762234e-05,
+      "loss": 0.9907,
+      "step": 480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913661835286361e-05,
+      "loss": 1.1122,
+      "step": 481
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913225033267741e-05,
+      "loss": 1.0172,
+      "step": 482
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9127871792095713e-05,
+      "loss": 0.9932,
+      "step": 483
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912348273616262e-05,
+      "loss": 1.0101,
+      "step": 484
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9119083169934343e-05,
+      "loss": 0.9624,
+      "step": 485
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9114673098479203e-05,
+      "loss": 1.1384,
+      "step": 486
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9110252526877625e-05,
+      "loss": 1.0095,
+      "step": 487
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9105821460222128e-05,
+      "loss": 0.9867,
+      "step": 488
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9101379903617322e-05,
+      "loss": 1.0431,
+      "step": 489
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9096927862179896e-05,
+      "loss": 0.9668,
+      "step": 490
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909246534103863e-05,
+      "loss": 1.0046,
+      "step": 491
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908799234533437e-05,
+      "loss": 0.9993,
+      "step": 492
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908350888022002e-05,
+      "loss": 1.1381,
+      "step": 493
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907901495086056e-05,
+      "loss": 1.0079,
+      "step": 494
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907451056243301e-05,
+      "loss": 1.1189,
+      "step": 495
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9069995720126458e-05,
+      "loss": 1.0519,
+      "step": 496
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9065470429142016e-05,
+      "loss": 0.9908,
+      "step": 497
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9060934694692837e-05,
+      "loss": 1.0001,
+      "step": 498
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905638852200412e-05,
+      "loss": 0.9915,
+      "step": 499
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905183191631307e-05,
+      "loss": 0.9617,
+      "step": 500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904726488286892e-05,
+      "loss": 0.9674,
+      "step": 501
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9042687426932914e-05,
+      "loss": 1.0926,
+      "step": 502
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9038099553778303e-05,
+      "loss": 1.0206,
+      "step": 503
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9033501268690333e-05,
+      "loss": 1.0492,
+      "step": 504
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902889257696626e-05,
+      "loss": 1.0244,
+      "step": 505
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9024273483915313e-05,
+      "loss": 0.9454,
+      "step": 506
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9019643994858707e-05,
+      "loss": 1.0138,
+      "step": 507
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9015004115129644e-05,
+      "loss": 1.0547,
+      "step": 508
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9010353850073272e-05,
+      "loss": 0.9428,
+      "step": 509
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9005693205046735e-05,
+      "loss": 0.9906,
+      "step": 510
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900102218541911e-05,
+      "loss": 1.0321,
+      "step": 511
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899634079657143e-05,
+      "loss": 1.0665,
+      "step": 512
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8991649043896687e-05,
+      "loss": 1.0526,
+      "step": 513
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8986946932799795e-05,
+      "loss": 0.9718,
+      "step": 514
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8982234468697606e-05,
+      "loss": 1.105,
+      "step": 515
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897751165701891e-05,
+      "loss": 0.9788,
+      "step": 516
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.89727785032044e-05,
+      "loss": 1.0236,
+      "step": 517
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8968035012706695e-05,
+      "loss": 0.9764,
+      "step": 518
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896328119099032e-05,
+      "loss": 0.9623,
+      "step": 519
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8958517043531698e-05,
+      "loss": 0.9931,
+      "step": 520
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8953742575819146e-05,
+      "loss": 1.039,
+      "step": 521
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8948957793352878e-05,
+      "loss": 1.1259,
+      "step": 522
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8944162701644987e-05,
+      "loss": 1.0577,
+      "step": 523
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893935730621944e-05,
+      "loss": 1.0458,
+      "step": 524
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8934541612612068e-05,
+      "loss": 0.971,
+      "step": 525
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8929715626370585e-05,
+      "loss": 0.9899,
+      "step": 526
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8924879353054542e-05,
+      "loss": 1.0283,
+      "step": 527
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892003279823535e-05,
+      "loss": 1.0215,
+      "step": 528
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8915175967496265e-05,
+      "loss": 0.9652,
+      "step": 529
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8910308866432376e-05,
+      "loss": 0.9995,
+      "step": 530
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890543150065061e-05,
+      "loss": 1.06,
+      "step": 531
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8900543875769713e-05,
+      "loss": 0.9862,
+      "step": 532
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889564599742025e-05,
+      "loss": 1.1507,
+      "step": 533
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8890737871244608e-05,
+      "loss": 1.0905,
+      "step": 534
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8885819502896962e-05,
+      "loss": 1.0479,
+      "step": 535
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.88808908980433e-05,
+      "loss": 0.9327,
+      "step": 536
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8875952062361397e-05,
+      "loss": 0.9893,
+      "step": 537
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8871003001540818e-05,
+      "loss": 0.9594,
+      "step": 538
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.88660437212829e-05,
+      "loss": 0.9717,
+      "step": 539
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.886107422730076e-05,
+      "loss": 0.9663,
+      "step": 540
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8856094525319277e-05,
+      "loss": 1.0168,
+      "step": 541
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885110462107509e-05,
+      "loss": 1.0055,
+      "step": 542
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8846104520316598e-05,
+      "loss": 0.9281,
+      "step": 543
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8841094228803936e-05,
+      "loss": 1.0649,
+      "step": 544
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8836073752308985e-05,
+      "loss": 0.9956,
+      "step": 545
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8831043096615357e-05,
+      "loss": 1.069,
+      "step": 546
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882600226751839e-05,
+      "loss": 1.0199,
+      "step": 547
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8820951270825142e-05,
+      "loss": 0.9491,
+      "step": 548
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881589011235439e-05,
+      "loss": 0.9925,
+      "step": 549
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881081879793661e-05,
+      "loss": 1.074,
+      "step": 550
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8805737333413977e-05,
+      "loss": 1.1344,
+      "step": 551
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8800645724640367e-05,
+      "loss": 1.094,
+      "step": 552
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8795543977481335e-05,
+      "loss": 1.0359,
+      "step": 553
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8790432097814114e-05,
+      "loss": 0.9722,
+      "step": 554
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.878531009152762e-05,
+      "loss": 1.048,
+      "step": 555
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8780177964522428e-05,
+      "loss": 0.9204,
+      "step": 556
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8775035722710772e-05,
+      "loss": 0.984,
+      "step": 557
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8769883372016533e-05,
+      "loss": 0.9901,
+      "step": 558
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8764720918375247e-05,
+      "loss": 1.0125,
+      "step": 559
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875954836773409e-05,
+      "loss": 1.0303,
+      "step": 560
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8754365726051857e-05,
+      "loss": 1.0075,
+      "step": 561
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8749172999298978e-05,
+      "loss": 1.1571,
+      "step": 562
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8743970193457504e-05,
+      "loss": 1.0354,
+      "step": 563
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8738757314521084e-05,
+      "loss": 1.0175,
+      "step": 564
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8733534368494984e-05,
+      "loss": 1.0578,
+      "step": 565
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872830136139606e-05,
+      "loss": 0.9552,
+      "step": 566
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8723058299252765e-05,
+      "loss": 1.0021,
+      "step": 567
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8717805188105126e-05,
+      "loss": 0.9826,
+      "step": 568
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8712542034004757e-05,
+      "loss": 1.0305,
+      "step": 569
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870726884301483e-05,
+      "loss": 0.9648,
+      "step": 570
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870198562121009e-05,
+      "loss": 1.0994,
+      "step": 571
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8696692374676833e-05,
+      "loss": 1.0414,
+      "step": 572
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869138910951291e-05,
+      "loss": 1.0126,
+      "step": 573
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.86860758318277e-05,
+      "loss": 1.0493,
+      "step": 574
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8680752547742117e-05,
+      "loss": 1.0393,
+      "step": 575
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8675419263388625e-05,
+      "loss": 0.9505,
+      "step": 576
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8670075984911182e-05,
+      "loss": 1.0448,
+      "step": 577
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8664722718465277e-05,
+      "loss": 0.9474,
+      "step": 578
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8659359470217893e-05,
+      "loss": 0.9404,
+      "step": 579
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.865398624634752e-05,
+      "loss": 1.7017,
+      "step": 580
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.864860305304414e-05,
+      "loss": 0.9516,
+      "step": 581
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8643209896509208e-05,
+      "loss": 0.899,
+      "step": 582
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8637806782955673e-05,
+      "loss": 1.0109,
+      "step": 583
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863239371860795e-05,
+      "loss": 0.962,
+      "step": 584
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8626970709701907e-05,
+      "loss": 1.0683,
+      "step": 585
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8621537762484875e-05,
+      "loss": 1.0609,
+      "step": 586
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8616094883215643e-05,
+      "loss": 1.0382,
+      "step": 587
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861064207816443e-05,
+      "loss": 0.9871,
+      "step": 588
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8605179353612893e-05,
+      "loss": 0.9817,
+      "step": 589
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8599706715854114e-05,
+      "loss": 1.0026,
+      "step": 590
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8594224171192604e-05,
+      "loss": 0.9565,
+      "step": 591
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8588731725944277e-05,
+      "loss": 1.0414,
+      "step": 592
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.858322938643645e-05,
+      "loss": 1.025,
+      "step": 593
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857771715900785e-05,
+      "loss": 0.9141,
+      "step": 594
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8572195050008593e-05,
+      "loss": 1.0867,
+      "step": 595
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8566663065800164e-05,
+      "loss": 1.0701,
+      "step": 596
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8561121212755444e-05,
+      "loss": 1.0276,
+      "step": 597
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8555569497258668e-05,
+      "loss": 1.0009,
+      "step": 598
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8550007925705445e-05,
+      "loss": 0.9418,
+      "step": 599
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8544436504502727e-05,
+      "loss": 1.1212,
+      "step": 600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8538855240068815e-05,
+      "loss": 1.07,
+      "step": 601
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8533264138833354e-05,
+      "loss": 0.9664,
+      "step": 602
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8527663207237325e-05,
+      "loss": 1.0686,
+      "step": 603
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852205245173302e-05,
+      "loss": 1.0109,
+      "step": 604
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851643187878406e-05,
+      "loss": 1.0227,
+      "step": 605
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851080149486537e-05,
+      "loss": 0.9318,
+      "step": 606
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.850516130646318e-05,
+      "loss": 1.0834,
+      "step": 607
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8499511320075016e-05,
+      "loss": 0.9881,
+      "step": 608
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.849385154220968e-05,
+      "loss": 0.9847,
+      "step": 609
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.848818197938727e-05,
+      "loss": 1.0415,
+      "step": 610
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.848250263813915e-05,
+      "loss": 0.9779,
+      "step": 611
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8476813525007942e-05,
+      "loss": 1.2033,
+      "step": 612
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8471114646547538e-05,
+      "loss": 0.9722,
+      "step": 613
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.846540600932307e-05,
+      "loss": 0.9678,
+      "step": 614
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845968761991091e-05,
+      "loss": 1.1116,
+      "step": 615
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8453959484898682e-05,
+      "loss": 1.0024,
+      "step": 616
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844822161088521e-05,
+      "loss": 0.977,
+      "step": 617
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8442474004480568e-05,
+      "loss": 1.0348,
+      "step": 618
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8436716672306008e-05,
+      "loss": 0.9941,
+      "step": 619
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8430949620994017e-05,
+      "loss": 1.0023,
+      "step": 620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8425172857188257e-05,
+      "loss": 1.0122,
+      "step": 621
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8419386387543598e-05,
+      "loss": 1.1066,
+      "step": 622
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8413590218726064e-05,
+      "loss": 1.0702,
+      "step": 623
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8407784357412884e-05,
+      "loss": 0.9048,
+      "step": 624
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8401968810292433e-05,
+      "loss": 1.0176,
+      "step": 625
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8396143584064243e-05,
+      "loss": 1.0628,
+      "step": 626
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8390308685439008e-05,
+      "loss": 0.8985,
+      "step": 627
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8384464121138557e-05,
+      "loss": 1.0213,
+      "step": 628
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.837860989789586e-05,
+      "loss": 0.917,
+      "step": 629
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8372746022455003e-05,
+      "loss": 1.0345,
+      "step": 630
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8366872501571202e-05,
+      "loss": 0.9538,
+      "step": 631
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8360989342010778e-05,
+      "loss": 0.8883,
+      "step": 632
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8355096550551167e-05,
+      "loss": 1.0016,
+      "step": 633
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834919413398088e-05,
+      "loss": 1.0493,
+      "step": 634
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8343282099099533e-05,
+      "loss": 1.0448,
+      "step": 635
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8337360452717825e-05,
+      "loss": 1.1184,
+      "step": 636
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8331429201657513e-05,
+      "loss": 0.9843,
+      "step": 637
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832548835275143e-05,
+      "loss": 1.1552,
+      "step": 638
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8319537912843458e-05,
+      "loss": 0.9635,
+      "step": 639
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8313577888788535e-05,
+      "loss": 1.0992,
+      "step": 640
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830760828745264e-05,
+      "loss": 0.9714,
+      "step": 641
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830162911571278e-05,
+      "loss": 0.9923,
+      "step": 642
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8295640380456982e-05,
+      "loss": 0.9965,
+      "step": 643
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.82896420885843e-05,
+      "loss": 1.1099,
+      "step": 644
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8283634247004808e-05,
+      "loss": 0.9821,
+      "step": 645
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827761686263955e-05,
+      "loss": 0.9414,
+      "step": 646
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827158994242059e-05,
+      "loss": 1.0226,
+      "step": 647
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8265553493290963e-05,
+      "loss": 0.9414,
+      "step": 648
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8259507522204695e-05,
+      "loss": 1.0478,
+      "step": 649
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8253452036126768e-05,
+      "loss": 0.9623,
+      "step": 650
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824738704203313e-05,
+      "loss": 1.0158,
+      "step": 651
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824131254691069e-05,
+      "loss": 1.0186,
+      "step": 652
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8235228557757282e-05,
+      "loss": 1.0083,
+      "step": 653
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.82291350815817e-05,
+      "loss": 0.9653,
+      "step": 654
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8223032125403657e-05,
+      "loss": 0.9137,
+      "step": 655
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8216919696253785e-05,
+      "loss": 0.9749,
+      "step": 656
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8210797801173632e-05,
+      "loss": 0.9424,
+      "step": 657
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8204666447215648e-05,
+      "loss": 0.9201,
+      "step": 658
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8198525641443184e-05,
+      "loss": 0.9795,
+      "step": 659
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8192375390930476e-05,
+      "loss": 0.9794,
+      "step": 660
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8186215702762643e-05,
+      "loss": 1.0827,
+      "step": 661
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8180046584035676e-05,
+      "loss": 0.9929,
+      "step": 662
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8173868041856422e-05,
+      "loss": 1.0667,
+      "step": 663
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8167680083342598e-05,
+      "loss": 0.9279,
+      "step": 664
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.816148271562276e-05,
+      "loss": 0.9259,
+      "step": 665
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8155275945836305e-05,
+      "loss": 1.0208,
+      "step": 666
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8149059781133455e-05,
+      "loss": 1.0004,
+      "step": 667
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.814283422867527e-05,
+      "loss": 0.943,
+      "step": 668
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8136599295633613e-05,
+      "loss": 1.0082,
+      "step": 669
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8130354989191153e-05,
+      "loss": 0.9858,
+      "step": 670
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8124101316541363e-05,
+      "loss": 0.9797,
+      "step": 671
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8117838284888505e-05,
+      "loss": 0.9133,
+      "step": 672
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8111565901447616e-05,
+      "loss": 0.9443,
+      "step": 673
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8105284173444518e-05,
+      "loss": 1.0253,
+      "step": 674
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8098993108115785e-05,
+      "loss": 0.9247,
+      "step": 675
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809269271270876e-05,
+      "loss": 0.9237,
+      "step": 676
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8086382994481523e-05,
+      "loss": 0.9274,
+      "step": 677
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.80800639607029e-05,
+      "loss": 1.0003,
+      "step": 678
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8073735618652453e-05,
+      "loss": 0.9362,
+      "step": 679
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8067397975620454e-05,
+      "loss": 1.0314,
+      "step": 680
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8061051038907904e-05,
+      "loss": 1.0222,
+      "step": 681
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.80546948158265e-05,
+      "loss": 0.952,
+      "step": 682
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8048329313698643e-05,
+      "loss": 0.9386,
+      "step": 683
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.804195453985742e-05,
+      "loss": 0.9684,
+      "step": 684
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8035570501646605e-05,
+      "loss": 0.9854,
+      "step": 685
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8029177206420636e-05,
+      "loss": 0.9155,
+      "step": 686
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8022774661544616e-05,
+      "loss": 0.9573,
+      "step": 687
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8016362874394316e-05,
+      "loss": 0.9941,
+      "step": 688
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.800994185235614e-05,
+      "loss": 0.9108,
+      "step": 689
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8003511602827133e-05,
+      "loss": 0.9437,
+      "step": 690
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7997072133214977e-05,
+      "loss": 0.968,
+      "step": 691
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799062345093797e-05,
+      "loss": 0.9177,
+      "step": 692
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7984165563425026e-05,
+      "loss": 0.9907,
+      "step": 693
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.797769847811566e-05,
+      "loss": 0.9745,
+      "step": 694
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7971222202459982e-05,
+      "loss": 0.9267,
+      "step": 695
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.79647367439187e-05,
+      "loss": 1.0649,
+      "step": 696
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7958242109963083e-05,
+      "loss": 0.975,
+      "step": 697
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7951738308074987e-05,
+      "loss": 1.0245,
+      "step": 698
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7945225345746822e-05,
+      "loss": 0.9786,
+      "step": 699
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.793870323048155e-05,
+      "loss": 1.1047,
+      "step": 700
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7932171969792678e-05,
+      "loss": 0.9936,
+      "step": 701
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.792563157120425e-05,
+      "loss": 1.0284,
+      "step": 702
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7919082042250834e-05,
+      "loss": 0.9545,
+      "step": 703
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7912523390477525e-05,
+      "loss": 0.9019,
+      "step": 704
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7905955623439906e-05,
+      "loss": 1.0113,
+      "step": 705
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7899378748704097e-05,
+      "loss": 0.9546,
+      "step": 706
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.789279277384667e-05,
+      "loss": 0.9525,
+      "step": 707
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7886197706454708e-05,
+      "loss": 0.9233,
+      "step": 708
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.787959355412576e-05,
+      "loss": 0.9588,
+      "step": 709
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7872980324467837e-05,
+      "loss": 0.9494,
+      "step": 710
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7866358025099413e-05,
+      "loss": 1.0006,
+      "step": 711
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785972666364941e-05,
+      "loss": 1.0174,
+      "step": 712
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785308624775718e-05,
+      "loss": 0.9879,
+      "step": 713
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7846436785072522e-05,
+      "loss": 0.8791,
+      "step": 714
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7839778283255646e-05,
+      "loss": 0.9658,
+      "step": 715
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7833110749977176e-05,
+      "loss": 1.041,
+      "step": 716
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.782643419291814e-05,
+      "loss": 0.9146,
+      "step": 717
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.781974861976997e-05,
+      "loss": 0.9341,
+      "step": 718
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7813054038234476e-05,
+      "loss": 0.918,
+      "step": 719
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.780635045602384e-05,
+      "loss": 0.9403,
+      "step": 720
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7799637880860624e-05,
+      "loss": 0.9997,
+      "step": 721
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7792916320477752e-05,
+      "loss": 0.9782,
+      "step": 722
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7786185782618485e-05,
+      "loss": 1.0377,
+      "step": 723
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7779446275036443e-05,
+      "loss": 0.934,
+      "step": 724
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7772697805495568e-05,
+      "loss": 1.0435,
+      "step": 725
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.776594038177012e-05,
+      "loss": 0.9557,
+      "step": 726
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7759174011644694e-05,
+      "loss": 0.9719,
+      "step": 727
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7752398702914175e-05,
+      "loss": 0.9669,
+      "step": 728
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.774561446338375e-05,
+      "loss": 0.9492,
+      "step": 729
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7738821300868903e-05,
+      "loss": 0.9068,
+      "step": 730
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7732019223195378e-05,
+      "loss": 0.9715,
+      "step": 731
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7725208238199207e-05,
+      "loss": 0.9009,
+      "step": 732
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7718388353726675e-05,
+      "loss": 0.9893,
+      "step": 733
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7711559577634325e-05,
+      "loss": 0.8749,
+      "step": 734
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.770472191778893e-05,
+      "loss": 1.0991,
+      "step": 735
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7697875382067516e-05,
+      "loss": 0.9029,
+      "step": 736
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769101997835732e-05,
+      "loss": 0.9692,
+      "step": 737
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.76841557145558e-05,
+      "loss": 0.8916,
+      "step": 738
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7677282598570623e-05,
+      "loss": 0.9613,
+      "step": 739
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7670400638319648e-05,
+      "loss": 0.9968,
+      "step": 740
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7663509841730923e-05,
+      "loss": 1.0264,
+      "step": 741
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7656610216742684e-05,
+      "loss": 1.0085,
+      "step": 742
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.764970177130333e-05,
+      "loss": 1.0059,
+      "step": 743
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.764278451337142e-05,
+      "loss": 0.9302,
+      "step": 744
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7635858450915672e-05,
+      "loss": 0.9837,
+      "step": 745
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.762892359191494e-05,
+      "loss": 0.9065,
+      "step": 746
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7621979944358213e-05,
+      "loss": 0.9314,
+      "step": 747
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7615027516244606e-05,
+      "loss": 0.9459,
+      "step": 748
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7608066315583353e-05,
+      "loss": 0.957,
+      "step": 749
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.760109635039379e-05,
+      "loss": 0.9872,
+      "step": 750
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.759411762870534e-05,
+      "loss": 1.0395,
+      "step": 751
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7587130158557536e-05,
+      "loss": 0.9856,
+      "step": 752
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7580133947999967e-05,
+      "loss": 0.8596,
+      "step": 753
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.757312900509231e-05,
+      "loss": 1.003,
+      "step": 754
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7566115337904284e-05,
+      "loss": 0.9759,
+      "step": 755
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755909295451567e-05,
+      "loss": 1.0834,
+      "step": 756
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7552061863016287e-05,
+      "loss": 1.0154,
+      "step": 757
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7545022071505985e-05,
+      "loss": 0.882,
+      "step": 758
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.753797358809464e-05,
+      "loss": 0.9412,
+      "step": 759
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7530916420902132e-05,
+      "loss": 0.9598,
+      "step": 760
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7523850578058353e-05,
+      "loss": 0.9989,
+      "step": 761
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7516776067703193e-05,
+      "loss": 1.0091,
+      "step": 762
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7509692897986513e-05,
+      "loss": 0.8847,
+      "step": 763
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7502601077068164e-05,
+      "loss": 0.9423,
+      "step": 764
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7495500613117954e-05,
+      "loss": 0.9065,
+      "step": 765
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7488391514315653e-05,
+      "loss": 0.9821,
+      "step": 766
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748127378885098e-05,
+      "loss": 0.9941,
+      "step": 767
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7474147444923576e-05,
+      "loss": 0.9746,
+      "step": 768
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7467012490743033e-05,
+      "loss": 0.9674,
+      "step": 769
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7459868934528853e-05,
+      "loss": 0.9577,
+      "step": 770
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.745271678451044e-05,
+      "loss": 0.9191,
+      "step": 771
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7445556048927107e-05,
+      "loss": 1.0651,
+      "step": 772
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.743838673602806e-05,
+      "loss": 1.0361,
+      "step": 773
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7431208854072372e-05,
+      "loss": 0.9426,
+      "step": 774
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7424022411329002e-05,
+      "loss": 0.8847,
+      "step": 775
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7416827416076773e-05,
+      "loss": 0.9866,
+      "step": 776
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7409623876604338e-05,
+      "loss": 0.947,
+      "step": 777
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7402411801210218e-05,
+      "loss": 1.135,
+      "step": 778
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.739519119820276e-05,
+      "loss": 1.0417,
+      "step": 779
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7387962075900132e-05,
+      "loss": 0.941,
+      "step": 780
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.738072444263031e-05,
+      "loss": 0.9628,
+      "step": 781
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.737347830673109e-05,
+      "loss": 0.9056,
+      "step": 782
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.736622367655005e-05,
+      "loss": 1.0806,
+      "step": 783
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.735896056044456e-05,
+      "loss": 0.9011,
+      "step": 784
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7351688966781758e-05,
+      "loss": 0.9651,
+      "step": 785
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7344408903938563e-05,
+      "loss": 0.9953,
+      "step": 786
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.733712038030164e-05,
+      "loss": 1.1021,
+      "step": 787
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.73298234042674e-05,
+      "loss": 0.9627,
+      "step": 788
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7322517984241994e-05,
+      "loss": 0.9278,
+      "step": 789
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.73152041286413e-05,
+      "loss": 0.9298,
+      "step": 790
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.730788184589092e-05,
+      "loss": 0.8826,
+      "step": 791
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.730055114442615e-05,
+      "loss": 1.1484,
+      "step": 792
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7293212032691994e-05,
+      "loss": 0.8964,
+      "step": 793
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.728586451914315e-05,
+      "loss": 1.0016,
+      "step": 794
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7278508612243985e-05,
+      "loss": 1.0018,
+      "step": 795
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7271144320468534e-05,
+      "loss": 0.9498,
+      "step": 796
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.72637716523005e-05,
+      "loss": 0.9951,
+      "step": 797
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7256390616233233e-05,
+      "loss": 0.9682,
+      "step": 798
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.724900122076972e-05,
+      "loss": 0.998,
+      "step": 799
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7241603474422577e-05,
+      "loss": 0.9079,
+      "step": 800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7234197385714044e-05,
+      "loss": 0.9323,
+      "step": 801
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.722678296317597e-05,
+      "loss": 0.9355,
+      "step": 802
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7219360215349806e-05,
+      "loss": 0.9716,
+      "step": 803
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7211929150786592e-05,
+      "loss": 1.0378,
+      "step": 804
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.720448977804695e-05,
+      "loss": 0.9857,
+      "step": 805
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7197042105701075e-05,
+      "loss": 0.9592,
+      "step": 806
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7189586142328712e-05,
+      "loss": 1.0238,
+      "step": 807
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7182121896519174e-05,
+      "loss": 0.9913,
+      "step": 808
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7174649376871305e-05,
+      "loss": 1.1265,
+      "step": 809
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.716716859199348e-05,
+      "loss": 1.1108,
+      "step": 810
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7159679550503606e-05,
+      "loss": 1.07,
+      "step": 811
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7152182261029092e-05,
+      "loss": 0.9826,
+      "step": 812
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7144676732206843e-05,
+      "loss": 0.9354,
+      "step": 813
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7137162972683267e-05,
+      "loss": 0.8626,
+      "step": 814
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7129640991114257e-05,
+      "loss": 0.9354,
+      "step": 815
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7122110796165167e-05,
+      "loss": 0.9367,
+      "step": 816
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7114572396510814e-05,
+      "loss": 0.914,
+      "step": 817
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7107025800835478e-05,
+      "loss": 0.9024,
+      "step": 818
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7099471017832864e-05,
+      "loss": 1.0032,
+      "step": 819
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.709190805620612e-05,
+      "loss": 0.955,
+      "step": 820
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.708433692466782e-05,
+      "loss": 1.009,
+      "step": 821
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7076757631939936e-05,
+      "loss": 0.9114,
+      "step": 822
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.706917018675385e-05,
+      "loss": 0.8904,
+      "step": 823
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7061574597850337e-05,
+      "loss": 1.0014,
+      "step": 824
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.705397087397955e-05,
+      "loss": 0.9654,
+      "step": 825
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7046359023901014e-05,
+      "loss": 1.0627,
+      "step": 826
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.703873905638362e-05,
+      "loss": 0.9515,
+      "step": 827
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.70311109802056e-05,
+      "loss": 0.9743,
+      "step": 828
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.702347480415454e-05,
+      "loss": 0.9691,
+      "step": 829
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7015830537027348e-05,
+      "loss": 0.9379,
+      "step": 830
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7008178187630257e-05,
+      "loss": 0.9214,
+      "step": 831
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7000517764778805e-05,
+      "loss": 1.0154,
+      "step": 832
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6992849277297843e-05,
+      "loss": 0.9777,
+      "step": 833
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.69851727340215e-05,
+      "loss": 0.886,
+      "step": 834
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.697748814379319e-05,
+      "loss": 0.9383,
+      "step": 835
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6969795515465598e-05,
+      "loss": 0.9922,
+      "step": 836
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.696209485790067e-05,
+      "loss": 0.9719,
+      "step": 837
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.69543861799696e-05,
+      "loss": 0.9022,
+      "step": 838
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.694666949055282e-05,
+      "loss": 0.9908,
+      "step": 839
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6938944798539988e-05,
+      "loss": 0.9036,
+      "step": 840
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.693121211282999e-05,
+      "loss": 0.9134,
+      "step": 841
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6923471442330917e-05,
+      "loss": 1.0636,
+      "step": 842
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.691572279596006e-05,
+      "loss": 0.9242,
+      "step": 843
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6907966182643893e-05,
+      "loss": 0.928,
+      "step": 844
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.690020161131807e-05,
+      "loss": 0.9327,
+      "step": 845
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6892429090927415e-05,
+      "loss": 0.871,
+      "step": 846
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.688464863042591e-05,
+      "loss": 1.0053,
+      "step": 847
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6876860238776682e-05,
+      "loss": 0.9738,
+      "step": 848
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6869063924951992e-05,
+      "loss": 1.0118,
+      "step": 849
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6861259697933233e-05,
+      "loss": 0.9348,
+      "step": 850
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.685344756671091e-05,
+      "loss": 0.8935,
+      "step": 851
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6845627540284638e-05,
+      "loss": 0.8951,
+      "step": 852
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6837799627663117e-05,
+      "loss": 0.93,
+      "step": 853
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.682996383786415e-05,
+      "loss": 1.1583,
+      "step": 854
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6822120179914593e-05,
+      "loss": 0.9834,
+      "step": 855
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6814268662850385e-05,
+      "loss": 1.0177,
+      "step": 856
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.680640929571651e-05,
+      "loss": 0.9182,
+      "step": 857
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6798542087566988e-05,
+      "loss": 0.8747,
+      "step": 858
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6790667047464892e-05,
+      "loss": 1.0865,
+      "step": 859
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6782784184482296e-05,
+      "loss": 0.9221,
+      "step": 860
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6774893507700302e-05,
+      "loss": 0.9917,
+      "step": 861
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6766995026209e-05,
+      "loss": 0.9184,
+      "step": 862
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6759088749107482e-05,
+      "loss": 0.9543,
+      "step": 863
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6751174685503814e-05,
+      "loss": 0.9679,
+      "step": 864
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6743252844515038e-05,
+      "loss": 0.9219,
+      "step": 865
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.673532323526715e-05,
+      "loss": 0.928,
+      "step": 866
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.67273858668951e-05,
+      "loss": 0.9499,
+      "step": 867
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6719440748542772e-05,
+      "loss": 0.9545,
+      "step": 868
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6711487889362977e-05,
+      "loss": 0.9286,
+      "step": 869
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.670352729851745e-05,
+      "loss": 0.9248,
+      "step": 870
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6695558985176833e-05,
+      "loss": 0.9334,
+      "step": 871
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.668758295852065e-05,
+      "loss": 0.9147,
+      "step": 872
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6679599227737333e-05,
+      "loss": 0.8799,
+      "step": 873
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6671607802024165e-05,
+      "loss": 1.0487,
+      "step": 874
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.666360869058732e-05,
+      "loss": 0.9845,
+      "step": 875
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.66556019026418e-05,
+      "loss": 1.013,
+      "step": 876
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.664758744741147e-05,
+      "loss": 0.8664,
+      "step": 877
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.663956533412902e-05,
+      "loss": 0.9449,
+      "step": 878
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6631535572035956e-05,
+      "loss": 0.9263,
+      "step": 879
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6623498170382607e-05,
+      "loss": 0.903,
+      "step": 880
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6615453138428096e-05,
+      "loss": 0.9722,
+      "step": 881
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6607400485440338e-05,
+      "loss": 1.0379,
+      "step": 882
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6599340220696032e-05,
+      "loss": 0.9198,
+      "step": 883
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6591272353480633e-05,
+      "loss": 0.939,
+      "step": 884
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6583196893088368e-05,
+      "loss": 0.9473,
+      "step": 885
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6575113848822205e-05,
+      "loss": 0.9808,
+      "step": 886
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6567023229993845e-05,
+      "loss": 0.9452,
+      "step": 887
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655892504592372e-05,
+      "loss": 0.9679,
+      "step": 888
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6550819305940985e-05,
+      "loss": 1.0076,
+      "step": 889
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6542706019383478e-05,
+      "loss": 1.0137,
+      "step": 890
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.653458519559775e-05,
+      "loss": 0.9804,
+      "step": 891
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.652645684393903e-05,
+      "loss": 1.0446,
+      "step": 892
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.651832097377121e-05,
+      "loss": 0.9564,
+      "step": 893
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.651017759446686e-05,
+      "loss": 1.0149,
+      "step": 894
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6502026715407185e-05,
+      "loss": 1.0087,
+      "step": 895
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.649386834598204e-05,
+      "loss": 0.9346,
+      "step": 896
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.64857024955899e-05,
+      "loss": 0.928,
+      "step": 897
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.647752917363787e-05,
+      "loss": 1.0785,
+      "step": 898
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.646934838954165e-05,
+      "loss": 0.8844,
+      "step": 899
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6461160152725548e-05,
+      "loss": 1.0216,
+      "step": 900
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6452964472622444e-05,
+      "loss": 0.9011,
+      "step": 901
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6444761358673803e-05,
+      "loss": 0.8639,
+      "step": 902
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6436550820329652e-05,
+      "loss": 0.9988,
+      "step": 903
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6428332867048574e-05,
+      "loss": 0.9779,
+      "step": 904
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.642010750829768e-05,
+      "loss": 0.9467,
+      "step": 905
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.641187475355263e-05,
+      "loss": 0.9013,
+      "step": 906
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6403634612297593e-05,
+      "loss": 0.8378,
+      "step": 907
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.639538709402525e-05,
+      "loss": 0.9286,
+      "step": 908
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6387132208236787e-05,
+      "loss": 0.9592,
+      "step": 909
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6378869964441866e-05,
+      "loss": 0.9537,
+      "step": 910
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6370600372158635e-05,
+      "loss": 0.9803,
+      "step": 911
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6362323440913694e-05,
+      "loss": 0.8937,
+      "step": 912
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6354039180242116e-05,
+      "loss": 1.0277,
+      "step": 913
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6345747599687404e-05,
+      "loss": 1.0162,
+      "step": 914
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.63374487088015e-05,
+      "loss": 0.8832,
+      "step": 915
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6329142517144766e-05,
+      "loss": 0.9322,
+      "step": 916
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6320829034285973e-05,
+      "loss": 0.9009,
+      "step": 917
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.631250826980229e-05,
+      "loss": 0.9176,
+      "step": 918
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.630418023327928e-05,
+      "loss": 0.8861,
+      "step": 919
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.629584493431088e-05,
+      "loss": 0.9715,
+      "step": 920
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6287502382499396e-05,
+      "loss": 0.9689,
+      "step": 921
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6279152587455486e-05,
+      "loss": 0.9189,
+      "step": 922
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.627079555879815e-05,
+      "loss": 0.912,
+      "step": 923
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6262431306154723e-05,
+      "loss": 0.9247,
+      "step": 924
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6254059839160866e-05,
+      "loss": 0.8641,
+      "step": 925
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6245681167460556e-05,
+      "loss": 0.9974,
+      "step": 926
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6237295300706052e-05,
+      "loss": 0.999,
+      "step": 927
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6228902248557917e-05,
+      "loss": 0.9828,
+      "step": 928
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6220502020684982e-05,
+      "loss": 0.9932,
+      "step": 929
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6212094626764357e-05,
+      "loss": 0.9182,
+      "step": 930
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6203680076481397e-05,
+      "loss": 0.9917,
+      "step": 931
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6195258379529696e-05,
+      "loss": 0.9454,
+      "step": 932
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.61868295456111e-05,
+      "loss": 0.996,
+      "step": 933
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6178393584435656e-05,
+      "loss": 0.9562,
+      "step": 934
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6169950505721637e-05,
+      "loss": 1.0739,
+      "step": 935
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6161500319195507e-05,
+      "loss": 1.0211,
+      "step": 936
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.615304303459192e-05,
+      "loss": 0.9453,
+      "step": 937
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6144578661653713e-05,
+      "loss": 0.9662,
+      "step": 938
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6136107210131875e-05,
+      "loss": 0.9863,
+      "step": 939
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6127628689785568e-05,
+      "loss": 0.9892,
+      "step": 940
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6119143110382078e-05,
+      "loss": 0.9303,
+      "step": 941
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.611065048169684e-05,
+      "loss": 0.9258,
+      "step": 942
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6102150813513397e-05,
+      "loss": 0.8784,
+      "step": 943
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6093644115623407e-05,
+      "loss": 1.0043,
+      "step": 944
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.608513039782663e-05,
+      "loss": 0.9513,
+      "step": 945
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.607660966993091e-05,
+      "loss": 1.0225,
+      "step": 946
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.606808194175216e-05,
+      "loss": 0.9624,
+      "step": 947
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.605954722311437e-05,
+      "loss": 0.9738,
+      "step": 948
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.605100552384957e-05,
+      "loss": 1.0697,
+      "step": 949
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6042456853797837e-05,
+      "loss": 0.9399,
+      "step": 950
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.603390122280729e-05,
+      "loss": 1.0356,
+      "step": 951
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6025338640734042e-05,
+      "loss": 0.9143,
+      "step": 952
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.601676911744224e-05,
+      "loss": 0.8887,
+      "step": 953
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6008192662804005e-05,
+      "loss": 1.0212,
+      "step": 954
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5999609286699463e-05,
+      "loss": 0.929,
+      "step": 955
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5991018999016695e-05,
+      "loss": 0.9946,
+      "step": 956
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.598242180965176e-05,
+      "loss": 0.8831,
+      "step": 957
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5973817728508656e-05,
+      "loss": 0.9091,
+      "step": 958
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5965206765499332e-05,
+      "loss": 0.9806,
+      "step": 959
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5956588930543646e-05,
+      "loss": 0.8981,
+      "step": 960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5947964233569395e-05,
+      "loss": 0.9299,
+      "step": 961
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5939332684512264e-05,
+      "loss": 0.9445,
+      "step": 962
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5930694293315844e-05,
+      "loss": 0.9879,
+      "step": 963
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.59220490699316e-05,
+      "loss": 0.9391,
+      "step": 964
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.591339702431887e-05,
+      "loss": 0.9206,
+      "step": 965
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5904738166444848e-05,
+      "loss": 0.9061,
+      "step": 966
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5896072506284588e-05,
+      "loss": 0.866,
+      "step": 967
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.588740005382096e-05,
+      "loss": 0.9391,
+      "step": 968
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.587872081904468e-05,
+      "loss": 0.9185,
+      "step": 969
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5870034811954265e-05,
+      "loss": 0.9173,
+      "step": 970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.586134204255603e-05,
+      "loss": 0.8957,
+      "step": 971
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5852642520864096e-05,
+      "loss": 0.8656,
+      "step": 972
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5843936256900344e-05,
+      "loss": 0.9338,
+      "step": 973
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.583522326069444e-05,
+      "loss": 0.9572,
+      "step": 974
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5826503542283788e-05,
+      "loss": 0.8779,
+      "step": 975
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.581777711171355e-05,
+      "loss": 0.9682,
+      "step": 976
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5809043979036613e-05,
+      "loss": 0.8758,
+      "step": 977
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.580030415431359e-05,
+      "loss": 0.8893,
+      "step": 978
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5791557647612798e-05,
+      "loss": 0.9484,
+      "step": 979
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5782804469010256e-05,
+      "loss": 0.892,
+      "step": 980
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.577404462858966e-05,
+      "loss": 0.9455,
+      "step": 981
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.57652781364424e-05,
+      "loss": 0.9649,
+      "step": 982
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5756505002667504e-05,
+      "loss": 0.9457,
+      "step": 983
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5747725237371673e-05,
+      "loss": 0.9408,
+      "step": 984
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5738938850669233e-05,
+      "loss": 0.9963,
+      "step": 985
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5730145852682142e-05,
+      "loss": 0.9247,
+      "step": 986
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.572134625353998e-05,
+      "loss": 0.9377,
+      "step": 987
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5712540063379923e-05,
+      "loss": 1.0319,
+      "step": 988
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5703727292346744e-05,
+      "loss": 1.0588,
+      "step": 989
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5694907950592798e-05,
+      "loss": 0.9539,
+      "step": 990
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5686082048278008e-05,
+      "loss": 1.0211,
+      "step": 991
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5677249595569855e-05,
+      "loss": 0.9259,
+      "step": 992
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5668410602643364e-05,
+      "loss": 0.9424,
+      "step": 993
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.56595650796811e-05,
+      "loss": 0.9273,
+      "step": 994
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5650713036873148e-05,
+      "loss": 0.8924,
+      "step": 995
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.56418544844171e-05,
+      "loss": 0.9331,
+      "step": 996
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.563298943251805e-05,
+      "loss": 0.9477,
+      "step": 997
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5624117891388578e-05,
+      "loss": 0.9196,
+      "step": 998
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.561523987124875e-05,
+      "loss": 0.906,
+      "step": 999
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.560635538232608e-05,
+      "loss": 0.9515,
+      "step": 1000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5597464434855537e-05,
+      "loss": 0.9332,
+      "step": 1001
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5588567039079547e-05,
+      "loss": 1.097,
+      "step": 1002
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.557966320524794e-05,
+      "loss": 1.0052,
+      "step": 1003
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5570752943617982e-05,
+      "loss": 0.9085,
+      "step": 1004
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5561836264454334e-05,
+      "loss": 0.9641,
+      "step": 1005
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5552913178029053e-05,
+      "loss": 0.8953,
+      "step": 1006
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5543983694621575e-05,
+      "loss": 0.9858,
+      "step": 1007
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5535047824518708e-05,
+      "loss": 0.8242,
+      "step": 1008
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5526105578014616e-05,
+      "loss": 0.8011,
+      "step": 1009
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.551715696541081e-05,
+      "loss": 0.7808,
+      "step": 1010
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5508201997016134e-05,
+      "loss": 0.7877,
+      "step": 1011
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.549924068314675e-05,
+      "loss": 0.7543,
+      "step": 1012
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5490273034126147e-05,
+      "loss": 0.8591,
+      "step": 1013
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.548129906028508e-05,
+      "loss": 0.7862,
+      "step": 1014
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5472318771961627e-05,
+      "loss": 0.7966,
+      "step": 1015
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.546333217950111e-05,
+      "loss": 0.7957,
+      "step": 1016
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.545433929325613e-05,
+      "loss": 0.784,
+      "step": 1017
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5445340123586534e-05,
+      "loss": 0.927,
+      "step": 1018
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.543633468085941e-05,
+      "loss": 0.7964,
+      "step": 1019
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5427322975449072e-05,
+      "loss": 0.8731,
+      "step": 1020
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.541830501773704e-05,
+      "loss": 0.815,
+      "step": 1021
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.540928081811205e-05,
+      "loss": 0.7885,
+      "step": 1022
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5400250386970023e-05,
+      "loss": 0.7668,
+      "step": 1023
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5391213734714052e-05,
+      "loss": 0.7657,
+      "step": 1024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5382170871754408e-05,
+      "loss": 0.7849,
+      "step": 1025
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5373121808508514e-05,
+      "loss": 0.809,
+      "step": 1026
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5364066555400924e-05,
+      "loss": 0.7572,
+      "step": 1027
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5355005122863347e-05,
+      "loss": 0.7866,
+      "step": 1028
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5345937521334586e-05,
+      "loss": 0.8255,
+      "step": 1029
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5336863761260556e-05,
+      "loss": 0.7788,
+      "step": 1030
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5327783853094284e-05,
+      "loss": 0.7747,
+      "step": 1031
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5318697807295865e-05,
+      "loss": 0.8128,
+      "step": 1032
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5309605634332454e-05,
+      "loss": 0.749,
+      "step": 1033
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5300507344678292e-05,
+      "loss": 0.8564,
+      "step": 1034
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5291402948814645e-05,
+      "loss": 0.8477,
+      "step": 1035
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5282292457229817e-05,
+      "loss": 0.8543,
+      "step": 1036
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5273175880419144e-05,
+      "loss": 0.7834,
+      "step": 1037
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5264053228884965e-05,
+      "loss": 0.8454,
+      "step": 1038
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5254924513136611e-05,
+      "loss": 0.7793,
+      "step": 1039
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5245789743690412e-05,
+      "loss": 0.7884,
+      "step": 1040
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5236648931069665e-05,
+      "loss": 0.8028,
+      "step": 1041
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5227502085804622e-05,
+      "loss": 0.8192,
+      "step": 1042
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5218349218432503e-05,
+      "loss": 0.7728,
+      "step": 1043
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5209190339497445e-05,
+      "loss": 0.7996,
+      "step": 1044
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5200025459550528e-05,
+      "loss": 0.7861,
+      "step": 1045
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5190854589149727e-05,
+      "loss": 0.8539,
+      "step": 1046
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5181677738859941e-05,
+      "loss": 0.7612,
+      "step": 1047
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.517249491925293e-05,
+      "loss": 0.7674,
+      "step": 1048
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5163306140907362e-05,
+      "loss": 0.7758,
+      "step": 1049
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5154111414408736e-05,
+      "loss": 0.8913,
+      "step": 1050
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5144910750349432e-05,
+      "loss": 0.7514,
+      "step": 1051
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5135704159328655e-05,
+      "loss": 0.7466,
+      "step": 1052
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5126491651952442e-05,
+      "loss": 0.7823,
+      "step": 1053
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5117273238833641e-05,
+      "loss": 0.8221,
+      "step": 1054
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5108048930591914e-05,
+      "loss": 0.8208,
+      "step": 1055
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5098818737853705e-05,
+      "loss": 0.7759,
+      "step": 1056
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5089582671252234e-05,
+      "loss": 0.8006,
+      "step": 1057
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5080340741427501e-05,
+      "loss": 0.7768,
+      "step": 1058
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.507109295902625e-05,
+      "loss": 0.8704,
+      "step": 1059
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5061839334701971e-05,
+      "loss": 0.7788,
+      "step": 1060
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.505257987911488e-05,
+      "loss": 0.8005,
+      "step": 1061
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5043314602931918e-05,
+      "loss": 0.7736,
+      "step": 1062
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5034043516826722e-05,
+      "loss": 0.7729,
+      "step": 1063
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5024766631479631e-05,
+      "loss": 0.8045,
+      "step": 1064
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.501548395757766e-05,
+      "loss": 0.8201,
+      "step": 1065
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.500619550581449e-05,
+      "loss": 0.7842,
+      "step": 1066
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4996901286890464e-05,
+      "loss": 0.9061,
+      "step": 1067
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4987601311512565e-05,
+      "loss": 0.8193,
+      "step": 1068
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4978295590394409e-05,
+      "loss": 0.7475,
+      "step": 1069
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4968984134256233e-05,
+      "loss": 0.8064,
+      "step": 1070
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4959666953824875e-05,
+      "loss": 0.8776,
+      "step": 1071
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.495034405983377e-05,
+      "loss": 0.7684,
+      "step": 1072
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4941015463022942e-05,
+      "loss": 0.758,
+      "step": 1073
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4931681174138972e-05,
+      "loss": 0.8386,
+      "step": 1074
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4922341203935007e-05,
+      "loss": 0.8214,
+      "step": 1075
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4912995563170738e-05,
+      "loss": 0.8357,
+      "step": 1076
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4903644262612387e-05,
+      "loss": 0.768,
+      "step": 1077
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4894287313032695e-05,
+      "loss": 0.8968,
+      "step": 1078
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4884924725210915e-05,
+      "loss": 0.8008,
+      "step": 1079
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4875556509932784e-05,
+      "loss": 0.7935,
+      "step": 1080
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4866182677990542e-05,
+      "loss": 0.7826,
+      "step": 1081
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4856803240182882e-05,
+      "loss": 0.7944,
+      "step": 1082
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4847418207314958e-05,
+      "loss": 0.8079,
+      "step": 1083
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4838027590198376e-05,
+      "loss": 0.8305,
+      "step": 1084
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4828631399651172e-05,
+      "loss": 0.8384,
+      "step": 1085
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4819229646497797e-05,
+      "loss": 0.7789,
+      "step": 1086
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4809822341569123e-05,
+      "loss": 0.815,
+      "step": 1087
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4800409495702401e-05,
+      "loss": 0.8129,
+      "step": 1088
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4790991119741277e-05,
+      "loss": 0.7796,
+      "step": 1089
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4781567224535767e-05,
+      "loss": 0.7821,
+      "step": 1090
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.477213782094224e-05,
+      "loss": 0.8641,
+      "step": 1091
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4762702919823415e-05,
+      "loss": 0.7986,
+      "step": 1092
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4753262532048343e-05,
+      "loss": 0.7994,
+      "step": 1093
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4743816668492393e-05,
+      "loss": 0.8172,
+      "step": 1094
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4734365340037245e-05,
+      "loss": 0.781,
+      "step": 1095
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4724908557570876e-05,
+      "loss": 0.836,
+      "step": 1096
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4715446331987542e-05,
+      "loss": 0.7836,
+      "step": 1097
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4705978674187773e-05,
+      "loss": 0.816,
+      "step": 1098
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4696505595078355e-05,
+      "loss": 0.8179,
+      "step": 1099
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.468702710557232e-05,
+      "loss": 0.7382,
+      "step": 1100
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4677543216588929e-05,
+      "loss": 0.8557,
+      "step": 1101
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4668053939053674e-05,
+      "loss": 0.7272,
+      "step": 1102
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4658559283898244e-05,
+      "loss": 0.8524,
+      "step": 1103
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4649059262060523e-05,
+      "loss": 0.7371,
+      "step": 1104
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4639553884484588e-05,
+      "loss": 0.8349,
+      "step": 1105
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4630043162120674e-05,
+      "loss": 0.8321,
+      "step": 1106
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4620527105925177e-05,
+      "loss": 0.8068,
+      "step": 1107
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4611005726860642e-05,
+      "loss": 0.7652,
+      "step": 1108
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4601479035895745e-05,
+      "loss": 0.9055,
+      "step": 1109
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.459194704400527e-05,
+      "loss": 0.8887,
+      "step": 1110
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4582409762170127e-05,
+      "loss": 0.7688,
+      "step": 1111
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4572867201377303e-05,
+      "loss": 0.7972,
+      "step": 1112
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4563319372619873e-05,
+      "loss": 0.8606,
+      "step": 1113
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4553766286896984e-05,
+      "loss": 0.8057,
+      "step": 1114
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4544207955213833e-05,
+      "loss": 0.7583,
+      "step": 1115
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4534644388581667e-05,
+      "loss": 0.798,
+      "step": 1116
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4525075598017757e-05,
+      "loss": 0.7709,
+      "step": 1117
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4515501594545398e-05,
+      "loss": 0.7875,
+      "step": 1118
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.450592238919388e-05,
+      "loss": 0.9274,
+      "step": 1119
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4496337992998501e-05,
+      "loss": 0.7818,
+      "step": 1120
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4486748417000526e-05,
+      "loss": 0.7887,
+      "step": 1121
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4477153672247193e-05,
+      "loss": 0.8147,
+      "step": 1122
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4467553769791696e-05,
+      "loss": 0.8181,
+      "step": 1123
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4457948720693162e-05,
+      "loss": 0.818,
+      "step": 1124
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4448338536016658e-05,
+      "loss": 0.8068,
+      "step": 1125
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.443872322683316e-05,
+      "loss": 0.8019,
+      "step": 1126
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4429102804219551e-05,
+      "loss": 0.7187,
+      "step": 1127
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4419477279258597e-05,
+      "loss": 0.7983,
+      "step": 1128
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4409846663038958e-05,
+      "loss": 0.8396,
+      "step": 1129
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4400210966655144e-05,
+      "loss": 0.7824,
+      "step": 1130
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.439057020120752e-05,
+      "loss": 0.7822,
+      "step": 1131
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.43809243778023e-05,
+      "loss": 0.8107,
+      "step": 1132
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4371273507551517e-05,
+      "loss": 0.7449,
+      "step": 1133
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4361617601573012e-05,
+      "loss": 0.821,
+      "step": 1134
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4351956670990443e-05,
+      "loss": 0.8027,
+      "step": 1135
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4342290726933243e-05,
+      "loss": 0.8594,
+      "step": 1136
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4332619780536625e-05,
+      "loss": 0.8563,
+      "step": 1137
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4322943842941565e-05,
+      "loss": 0.7578,
+      "step": 1138
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4313262925294794e-05,
+      "loss": 0.8463,
+      "step": 1139
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4303577038748766e-05,
+      "loss": 0.8202,
+      "step": 1140
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4293886194461675e-05,
+      "loss": 0.9129,
+      "step": 1141
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4284190403597415e-05,
+      "loss": 0.7979,
+      "step": 1142
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4274489677325585e-05,
+      "loss": 0.7294,
+      "step": 1143
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4264784026821468e-05,
+      "loss": 0.7436,
+      "step": 1144
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.425507346326602e-05,
+      "loss": 0.7753,
+      "step": 1145
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4245357997845854e-05,
+      "loss": 0.7815,
+      "step": 1146
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4235637641753228e-05,
+      "loss": 0.7785,
+      "step": 1147
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4225912406186049e-05,
+      "loss": 0.8112,
+      "step": 1148
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4216182302347825e-05,
+      "loss": 0.7697,
+      "step": 1149
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4206447341447684e-05,
+      "loss": 0.807,
+      "step": 1150
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4196707534700348e-05,
+      "loss": 0.7501,
+      "step": 1151
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.418696289332612e-05,
+      "loss": 0.769,
+      "step": 1152
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4177213428550873e-05,
+      "loss": 0.7834,
+      "step": 1153
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4167459151606035e-05,
+      "loss": 0.884,
+      "step": 1154
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.415770007372858e-05,
+      "loss": 0.7766,
+      "step": 1155
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4147936206161013e-05,
+      "loss": 0.7556,
+      "step": 1156
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4138167560151354e-05,
+      "loss": 0.7368,
+      "step": 1157
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4128394146953129e-05,
+      "loss": 0.7977,
+      "step": 1158
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4118615977825356e-05,
+      "loss": 0.7623,
+      "step": 1159
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4108833064032536e-05,
+      "loss": 0.8311,
+      "step": 1160
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4099045416844627e-05,
+      "loss": 0.8017,
+      "step": 1161
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4089253047537046e-05,
+      "loss": 0.8081,
+      "step": 1162
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4079455967390652e-05,
+      "loss": 0.8361,
+      "step": 1163
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4069654187691721e-05,
+      "loss": 0.8334,
+      "step": 1164
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4059847719731956e-05,
+      "loss": 0.8352,
+      "step": 1165
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4050036574808452e-05,
+      "loss": 0.7742,
+      "step": 1166
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.404022076422369e-05,
+      "loss": 0.7896,
+      "step": 1167
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4030400299285538e-05,
+      "loss": 1.0652,
+      "step": 1168
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4020575191307213e-05,
+      "loss": 0.7775,
+      "step": 1169
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4010745451607286e-05,
+      "loss": 0.7973,
+      "step": 1170
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4000911091509665e-05,
+      "loss": 0.7309,
+      "step": 1171
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3991072122343581e-05,
+      "loss": 0.8205,
+      "step": 1172
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3981228555443567e-05,
+      "loss": 0.8875,
+      "step": 1173
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3971380402149464e-05,
+      "loss": 0.7948,
+      "step": 1174
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.396152767380639e-05,
+      "loss": 0.813,
+      "step": 1175
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.395167038176473e-05,
+      "loss": 0.883,
+      "step": 1176
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3941808537380137e-05,
+      "loss": 0.895,
+      "step": 1177
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.39319421520135e-05,
+      "loss": 0.7269,
+      "step": 1178
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3922071237030941e-05,
+      "loss": 0.7787,
+      "step": 1179
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3912195803803803e-05,
+      "loss": 0.8368,
+      "step": 1180
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.390231586370863e-05,
+      "loss": 0.7584,
+      "step": 1181
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3892431428127156e-05,
+      "loss": 0.8304,
+      "step": 1182
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3882542508446303e-05,
+      "loss": 0.7189,
+      "step": 1183
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3872649116058155e-05,
+      "loss": 0.8114,
+      "step": 1184
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3862751262359937e-05,
+      "loss": 0.7886,
+      "step": 1185
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.385284895875403e-05,
+      "loss": 0.8807,
+      "step": 1186
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3842942216647937e-05,
+      "loss": 0.8146,
+      "step": 1187
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3833031047454265e-05,
+      "loss": 0.789,
+      "step": 1188
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3823115462590733e-05,
+      "loss": 0.8081,
+      "step": 1189
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3813195473480137e-05,
+      "loss": 0.7778,
+      "step": 1190
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3803271091550354e-05,
+      "loss": 0.8116,
+      "step": 1191
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3793342328234322e-05,
+      "loss": 0.7696,
+      "step": 1192
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3783409194970019e-05,
+      "loss": 0.7183,
+      "step": 1193
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3773471703200458e-05,
+      "loss": 0.7661,
+      "step": 1194
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3763529864373686e-05,
+      "loss": 0.8659,
+      "step": 1195
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.375358368994274e-05,
+      "loss": 0.8231,
+      "step": 1196
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3743633191365661e-05,
+      "loss": 0.7739,
+      "step": 1197
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3733678380105478e-05,
+      "loss": 0.7992,
+      "step": 1198
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.372371926763017e-05,
+      "loss": 0.734,
+      "step": 1199
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3713755865412686e-05,
+      "loss": 0.8403,
+      "step": 1200
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3703788184930916e-05,
+      "loss": 0.798,
+      "step": 1201
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.369381623766767e-05,
+      "loss": 0.7619,
+      "step": 1202
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.368384003511068e-05,
+      "loss": 0.7955,
+      "step": 1203
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3673859588752579e-05,
+      "loss": 0.7417,
+      "step": 1204
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3663874910090888e-05,
+      "loss": 0.8277,
+      "step": 1205
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3653886010628002e-05,
+      "loss": 0.7632,
+      "step": 1206
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3643892901871185e-05,
+      "loss": 0.7678,
+      "step": 1207
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.363389559533254e-05,
+      "loss": 0.8344,
+      "step": 1208
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3623894102529014e-05,
+      "loss": 0.8154,
+      "step": 1209
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3613888434982375e-05,
+      "loss": 0.7765,
+      "step": 1210
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3603878604219198e-05,
+      "loss": 0.8012,
+      "step": 1211
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3593864621770854e-05,
+      "loss": 0.7896,
+      "step": 1212
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3583846499173502e-05,
+      "loss": 0.7795,
+      "step": 1213
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3573824247968061e-05,
+      "loss": 0.7829,
+      "step": 1214
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3563797879700215e-05,
+      "loss": 0.8223,
+      "step": 1215
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3553767405920388e-05,
+      "loss": 0.8031,
+      "step": 1216
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3543732838183735e-05,
+      "loss": 0.766,
+      "step": 1217
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3533694188050117e-05,
+      "loss": 0.8114,
+      "step": 1218
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3523651467084116e-05,
+      "loss": 0.7405,
+      "step": 1219
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3513604686854992e-05,
+      "loss": 0.7788,
+      "step": 1220
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3503553858936676e-05,
+      "loss": 0.8253,
+      "step": 1221
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.349349899490778e-05,
+      "loss": 0.6976,
+      "step": 1222
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3483440106351546e-05,
+      "loss": 0.8243,
+      "step": 1223
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3473377204855864e-05,
+      "loss": 0.927,
+      "step": 1224
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3463310302013246e-05,
+      "loss": 0.8098,
+      "step": 1225
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3453239409420809e-05,
+      "loss": 0.7945,
+      "step": 1226
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3443164538680273e-05,
+      "loss": 0.7191,
+      "step": 1227
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3433085701397932e-05,
+      "loss": 0.8209,
+      "step": 1228
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3423002909184653e-05,
+      "loss": 0.8511,
+      "step": 1229
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3412916173655866e-05,
+      "loss": 0.7499,
+      "step": 1230
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3402825506431534e-05,
+      "loss": 0.7104,
+      "step": 1231
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3392730919136154e-05,
+      "loss": 0.7851,
+      "step": 1232
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3382632423398736e-05,
+      "loss": 0.8497,
+      "step": 1233
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3372530030852802e-05,
+      "loss": 0.7583,
+      "step": 1234
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3362423753136349e-05,
+      "loss": 0.8031,
+      "step": 1235
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3352313601891856e-05,
+      "loss": 0.7894,
+      "step": 1236
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3342199588766269e-05,
+      "loss": 0.814,
+      "step": 1237
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.333208172541098e-05,
+      "loss": 0.7647,
+      "step": 1238
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3321960023481811e-05,
+      "loss": 0.8307,
+      "step": 1239
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3311834494639011e-05,
+      "loss": 0.7289,
+      "step": 1240
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3301705150547242e-05,
+      "loss": 0.8273,
+      "step": 1241
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3291572002875546e-05,
+      "loss": 0.7897,
+      "step": 1242
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.328143506329737e-05,
+      "loss": 0.8573,
+      "step": 1243
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3271294343490508e-05,
+      "loss": 0.7963,
+      "step": 1244
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3261149855137116e-05,
+      "loss": 0.7431,
+      "step": 1245
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3251001609923697e-05,
+      "loss": 0.8469,
+      "step": 1246
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3240849619541078e-05,
+      "loss": 0.7276,
+      "step": 1247
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3230693895684395e-05,
+      "loss": 0.8491,
+      "step": 1248
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3220534450053092e-05,
+      "loss": 0.8971,
+      "step": 1249
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.32103712943509e-05,
+      "loss": 0.8121,
+      "step": 1250
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.320020444028582e-05,
+      "loss": 0.7987,
+      "step": 1251
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3190033899570118e-05,
+      "loss": 0.7661,
+      "step": 1252
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3179859683920304e-05,
+      "loss": 0.8137,
+      "step": 1253
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3169681805057122e-05,
+      "loss": 0.7845,
+      "step": 1254
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3159500274705536e-05,
+      "loss": 0.7194,
+      "step": 1255
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.314931510459472e-05,
+      "loss": 0.7516,
+      "step": 1256
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3139126306458032e-05,
+      "loss": 0.7389,
+      "step": 1257
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3128933892033022e-05,
+      "loss": 0.7802,
+      "step": 1258
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3118737873061397e-05,
+      "loss": 0.8128,
+      "step": 1259
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3108538261289016e-05,
+      "loss": 0.7768,
+      "step": 1260
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3098335068465889e-05,
+      "loss": 0.7376,
+      "step": 1261
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3088128306346132e-05,
+      "loss": 0.8172,
+      "step": 1262
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3077917986687988e-05,
+      "loss": 0.8282,
+      "step": 1263
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.306770412125379e-05,
+      "loss": 0.8424,
+      "step": 1264
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3057486721809966e-05,
+      "loss": 0.7998,
+      "step": 1265
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3047265800127002e-05,
+      "loss": 0.7673,
+      "step": 1266
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3037041367979454e-05,
+      "loss": 0.7527,
+      "step": 1267
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3026813437145909e-05,
+      "loss": 0.7248,
+      "step": 1268
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3016582019408995e-05,
+      "loss": 0.7418,
+      "step": 1269
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3006347126555355e-05,
+      "loss": 0.7479,
+      "step": 1270
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2996108770375631e-05,
+      "loss": 0.7954,
+      "step": 1271
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.298586696266446e-05,
+      "loss": 0.8594,
+      "step": 1272
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2975621715220452e-05,
+      "loss": 0.865,
+      "step": 1273
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.296537303984618e-05,
+      "loss": 0.7659,
+      "step": 1274
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2955120948348168e-05,
+      "loss": 0.8252,
+      "step": 1275
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2944865452536872e-05,
+      "loss": 0.9229,
+      "step": 1276
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2934606564226672e-05,
+      "loss": 0.7821,
+      "step": 1277
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2924344295235856e-05,
+      "loss": 0.8385,
+      "step": 1278
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2914078657386608e-05,
+      "loss": 0.7941,
+      "step": 1279
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2903809662504986e-05,
+      "loss": 0.7529,
+      "step": 1280
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2893537322420925e-05,
+      "loss": 0.7327,
+      "step": 1281
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2883261648968213e-05,
+      "loss": 0.7778,
+      "step": 1282
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2872982653984466e-05,
+      "loss": 0.7715,
+      "step": 1283
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2862700349311137e-05,
+      "loss": 0.7686,
+      "step": 1284
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2852414746793495e-05,
+      "loss": 0.7639,
+      "step": 1285
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2842125858280596e-05,
+      "loss": 0.7271,
+      "step": 1286
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2831833695625291e-05,
+      "loss": 0.7146,
+      "step": 1287
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2821538270684201e-05,
+      "loss": 0.7805,
+      "step": 1288
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2811239595317703e-05,
+      "loss": 0.7787,
+      "step": 1289
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2800937681389918e-05,
+      "loss": 0.7955,
+      "step": 1290
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2790632540768702e-05,
+      "loss": 0.7788,
+      "step": 1291
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2780324185325626e-05,
+      "loss": 0.7792,
+      "step": 1292
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2770012626935962e-05,
+      "loss": 0.7917,
+      "step": 1293
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2759697877478673e-05,
+      "loss": 0.7585,
+      "step": 1294
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2749379948836402e-05,
+      "loss": 0.8267,
+      "step": 1295
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.273905885289545e-05,
+      "loss": 0.902,
+      "step": 1296
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2728734601545769e-05,
+      "loss": 0.7634,
+      "step": 1297
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2718407206680948e-05,
+      "loss": 0.7723,
+      "step": 1298
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2708076680198188e-05,
+      "loss": 0.8189,
+      "step": 1299
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2697743033998308e-05,
+      "loss": 0.7318,
+      "step": 1300
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2687406279985717e-05,
+      "loss": 0.8117,
+      "step": 1301
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2677066430068405e-05,
+      "loss": 0.824,
+      "step": 1302
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.266672349615793e-05,
+      "loss": 0.8125,
+      "step": 1303
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2656377490169391e-05,
+      "loss": 0.7955,
+      "step": 1304
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.264602842402145e-05,
+      "loss": 0.7696,
+      "step": 1305
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.263567630963627e-05,
+      "loss": 0.845,
+      "step": 1306
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2625321158939536e-05,
+      "loss": 0.8133,
+      "step": 1307
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2614962983860437e-05,
+      "loss": 0.8086,
+      "step": 1308
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2604601796331634e-05,
+      "loss": 0.7501,
+      "step": 1309
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2594237608289266e-05,
+      "loss": 0.711,
+      "step": 1310
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2583870431672924e-05,
+      "loss": 0.7897,
+      "step": 1311
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.257350027842565e-05,
+      "loss": 0.8341,
+      "step": 1312
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2563127160493903e-05,
+      "loss": 0.8119,
+      "step": 1313
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2552751089827568e-05,
+      "loss": 0.7931,
+      "step": 1314
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2542372078379928e-05,
+      "loss": 0.7881,
+      "step": 1315
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2531990138107649e-05,
+      "loss": 0.7451,
+      "step": 1316
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2521605280970777e-05,
+      "loss": 0.8178,
+      "step": 1317
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2511217518932719e-05,
+      "loss": 0.7215,
+      "step": 1318
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2500826863960223e-05,
+      "loss": 0.8124,
+      "step": 1319
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2490433328023372e-05,
+      "loss": 0.789,
+      "step": 1320
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2480036923095573e-05,
+      "loss": 0.772,
+      "step": 1321
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2469637661153528e-05,
+      "loss": 0.797,
+      "step": 1322
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.245923555417724e-05,
+      "loss": 0.7654,
+      "step": 1323
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2448830614149985e-05,
+      "loss": 0.8224,
+      "step": 1324
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2438422853058299e-05,
+      "loss": 0.8148,
+      "step": 1325
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2428012282891977e-05,
+      "loss": 0.7854,
+      "step": 1326
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2417598915644043e-05,
+      "loss": 0.79,
+      "step": 1327
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2407182763310741e-05,
+      "loss": 0.7645,
+      "step": 1328
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2396763837891534e-05,
+      "loss": 0.8892,
+      "step": 1329
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2386342151389072e-05,
+      "loss": 0.8411,
+      "step": 1330
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2375917715809183e-05,
+      "loss": 0.7918,
+      "step": 1331
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2365490543160872e-05,
+      "loss": 0.7503,
+      "step": 1332
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2355060645456287e-05,
+      "loss": 0.7793,
+      "step": 1333
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2344628034710718e-05,
+      "loss": 0.7868,
+      "step": 1334
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2334192722942588e-05,
+      "loss": 0.8092,
+      "step": 1335
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.232375472217342e-05,
+      "loss": 0.7496,
+      "step": 1336
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.231331404442784e-05,
+      "loss": 0.7345,
+      "step": 1337
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2302870701733563e-05,
+      "loss": 0.764,
+      "step": 1338
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2292424706121364e-05,
+      "loss": 0.9009,
+      "step": 1339
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2281976069625081e-05,
+      "loss": 0.7925,
+      "step": 1340
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2271524804281593e-05,
+      "loss": 0.7844,
+      "step": 1341
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2261070922130809e-05,
+      "loss": 0.8338,
+      "step": 1342
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2250614435215646e-05,
+      "loss": 0.9971,
+      "step": 1343
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2240155355582031e-05,
+      "loss": 0.7783,
+      "step": 1344
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2229693695278868e-05,
+      "loss": 0.7522,
+      "step": 1345
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2219229466358042e-05,
+      "loss": 0.7612,
+      "step": 1346
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2208762680874396e-05,
+      "loss": 0.7396,
+      "step": 1347
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2198293350885715e-05,
+      "loss": 0.8801,
+      "step": 1348
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.218782148845271e-05,
+      "loss": 0.7506,
+      "step": 1349
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2177347105639028e-05,
+      "loss": 0.8509,
+      "step": 1350
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2166870214511202e-05,
+      "loss": 0.7783,
+      "step": 1351
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2156390827138653e-05,
+      "loss": 0.7954,
+      "step": 1352
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2145908955593698e-05,
+      "loss": 0.8401,
+      "step": 1353
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2135424611951492e-05,
+      "loss": 0.8511,
+      "step": 1354
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.212493780829005e-05,
+      "loss": 0.743,
+      "step": 1355
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2114448556690224e-05,
+      "loss": 0.7978,
+      "step": 1356
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2103956869235677e-05,
+      "loss": 0.8239,
+      "step": 1357
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2093462758012884e-05,
+      "loss": 0.8522,
+      "step": 1358
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2082966235111111e-05,
+      "loss": 0.797,
+      "step": 1359
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2072467312622402e-05,
+      "loss": 0.8123,
+      "step": 1360
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2061966002641565e-05,
+      "loss": 0.836,
+      "step": 1361
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2051462317266159e-05,
+      "loss": 0.7617,
+      "step": 1362
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.204095626859648e-05,
+      "loss": 0.7999,
+      "step": 1363
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.203044786873554e-05,
+      "loss": 0.7806,
+      "step": 1364
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2019937129789074e-05,
+      "loss": 0.7981,
+      "step": 1365
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2009424063865498e-05,
+      "loss": 0.8473,
+      "step": 1366
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1998908683075912e-05,
+      "loss": 0.8133,
+      "step": 1367
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1988390999534087e-05,
+      "loss": 0.7658,
+      "step": 1368
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1977871025356445e-05,
+      "loss": 0.7834,
+      "step": 1369
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1967348772662043e-05,
+      "loss": 0.7579,
+      "step": 1370
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1956824253572568e-05,
+      "loss": 0.7641,
+      "step": 1371
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1946297480212317e-05,
+      "loss": 0.764,
+      "step": 1372
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1935768464708175e-05,
+      "loss": 0.8152,
+      "step": 1373
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1925237219189626e-05,
+      "loss": 0.8059,
+      "step": 1374
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1914703755788715e-05,
+      "loss": 0.8303,
+      "step": 1375
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1904168086640032e-05,
+      "loss": 0.8295,
+      "step": 1376
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1893630223880728e-05,
+      "loss": 0.8018,
+      "step": 1377
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1883090179650464e-05,
+      "loss": 0.8036,
+      "step": 1378
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1872547966091423e-05,
+      "loss": 0.7716,
+      "step": 1379
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1862003595348289e-05,
+      "loss": 0.8971,
+      "step": 1380
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.185145707956822e-05,
+      "loss": 0.7586,
+      "step": 1381
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1840908430900853e-05,
+      "loss": 0.8566,
+      "step": 1382
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1830357661498285e-05,
+      "loss": 0.7511,
+      "step": 1383
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.181980478351505e-05,
+      "loss": 0.7847,
+      "step": 1384
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1809249809108112e-05,
+      "loss": 0.7721,
+      "step": 1385
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1798692750436853e-05,
+      "loss": 0.7144,
+      "step": 1386
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1788133619663056e-05,
+      "loss": 0.7361,
+      "step": 1387
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1777572428950883e-05,
+      "loss": 0.7756,
+      "step": 1388
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1767009190466885e-05,
+      "loss": 0.7868,
+      "step": 1389
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1756443916379956e-05,
+      "loss": 0.7825,
+      "step": 1390
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1745876618861343e-05,
+      "loss": 0.726,
+      "step": 1391
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1735307310084623e-05,
+      "loss": 0.7514,
+      "step": 1392
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1724736002225689e-05,
+      "loss": 0.7595,
+      "step": 1393
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1714162707462737e-05,
+      "loss": 0.7697,
+      "step": 1394
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1703587437976256e-05,
+      "loss": 0.8328,
+      "step": 1395
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1693010205949004e-05,
+      "loss": 0.7689,
+      "step": 1396
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1682431023566002e-05,
+      "loss": 0.8318,
+      "step": 1397
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1671849903014518e-05,
+      "loss": 0.8565,
+      "step": 1398
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1661266856484053e-05,
+      "loss": 0.7919,
+      "step": 1399
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1650681896166325e-05,
+      "loss": 0.8511,
+      "step": 1400
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1640095034255261e-05,
+      "loss": 0.8164,
+      "step": 1401
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1629506282946976e-05,
+      "loss": 0.7911,
+      "step": 1402
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1618915654439757e-05,
+      "loss": 0.7587,
+      "step": 1403
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1608323160934064e-05,
+      "loss": 0.7746,
+      "step": 1404
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1597728814632498e-05,
+      "loss": 0.8115,
+      "step": 1405
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1587132627739792e-05,
+      "loss": 0.8142,
+      "step": 1406
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1576534612462808e-05,
+      "loss": 0.7805,
+      "step": 1407
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1565934781010509e-05,
+      "loss": 0.7503,
+      "step": 1408
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1555333145593945e-05,
+      "loss": 0.738,
+      "step": 1409
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.154472971842626e-05,
+      "loss": 0.7697,
+      "step": 1410
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1534124511722644e-05,
+      "loss": 0.8493,
+      "step": 1411
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1523517537700349e-05,
+      "loss": 0.7953,
+      "step": 1412
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1512908808578657e-05,
+      "loss": 0.7497,
+      "step": 1413
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1502298336578877e-05,
+      "loss": 0.7589,
+      "step": 1414
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1491686133924319e-05,
+      "loss": 0.7614,
+      "step": 1415
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1481072212840292e-05,
+      "loss": 0.7912,
+      "step": 1416
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1470456585554086e-05,
+      "loss": 0.7459,
+      "step": 1417
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1459839264294949e-05,
+      "loss": 0.7893,
+      "step": 1418
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1449220261294093e-05,
+      "loss": 0.8867,
+      "step": 1419
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.143859958878465e-05,
+      "loss": 0.7601,
+      "step": 1420
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1427977259001688e-05,
+      "loss": 0.7347,
+      "step": 1421
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1417353284182186e-05,
+      "loss": 0.7933,
+      "step": 1422
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1406727676565008e-05,
+      "loss": 0.8296,
+      "step": 1423
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1396100448390903e-05,
+      "loss": 0.7551,
+      "step": 1424
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1385471611902495e-05,
+      "loss": 0.7401,
+      "step": 1425
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1374841179344246e-05,
+      "loss": 0.8485,
+      "step": 1426
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.136420916296247e-05,
+      "loss": 0.7458,
+      "step": 1427
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1353575575005298e-05,
+      "loss": 0.7773,
+      "step": 1428
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1342940427722672e-05,
+      "loss": 0.7886,
+      "step": 1429
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1332303733366331e-05,
+      "loss": 0.7061,
+      "step": 1430
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1321665504189805e-05,
+      "loss": 0.8077,
+      "step": 1431
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1311025752448377e-05,
+      "loss": 0.7716,
+      "step": 1432
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.130038449039909e-05,
+      "loss": 0.7652,
+      "step": 1433
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1289741730300733e-05,
+      "loss": 0.7804,
+      "step": 1434
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1279097484413816e-05,
+      "loss": 0.7286,
+      "step": 1435
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1268451765000556e-05,
+      "loss": 0.7871,
+      "step": 1436
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1257804584324876e-05,
+      "loss": 0.7489,
+      "step": 1437
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1247155954652376e-05,
+      "loss": 0.7593,
+      "step": 1438
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1236505888250326e-05,
+      "loss": 0.8116,
+      "step": 1439
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1225854397387658e-05,
+      "loss": 0.7075,
+      "step": 1440
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.121520149433494e-05,
+      "loss": 0.862,
+      "step": 1441
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.120454719136436e-05,
+      "loss": 0.7623,
+      "step": 1442
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1193891500749729e-05,
+      "loss": 0.7976,
+      "step": 1443
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1183234434766455e-05,
+      "loss": 0.7714,
+      "step": 1444
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1172576005691524e-05,
+      "loss": 0.8265,
+      "step": 1445
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.11619162258035e-05,
+      "loss": 0.6897,
+      "step": 1446
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1151255107382497e-05,
+      "loss": 0.7482,
+      "step": 1447
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1140592662710177e-05,
+      "loss": 0.7887,
+      "step": 1448
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1129928904069725e-05,
+      "loss": 0.8,
+      "step": 1449
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1119263843745839e-05,
+      "loss": 0.7738,
+      "step": 1450
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1108597494024722e-05,
+      "loss": 0.7489,
+      "step": 1451
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.109792986719406e-05,
+      "loss": 0.7829,
+      "step": 1452
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1087260975543006e-05,
+      "loss": 0.7978,
+      "step": 1453
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1076590831362175e-05,
+      "loss": 0.7518,
+      "step": 1454
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1065919446943627e-05,
+      "loss": 0.7782,
+      "step": 1455
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.105524683458084e-05,
+      "loss": 0.8089,
+      "step": 1456
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1044573006568719e-05,
+      "loss": 0.7615,
+      "step": 1457
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1033897975203565e-05,
+      "loss": 0.8589,
+      "step": 1458
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.102322175278306e-05,
+      "loss": 0.8049,
+      "step": 1459
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1012544351606267e-05,
+      "loss": 0.7454,
+      "step": 1460
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.10018657839736e-05,
+      "loss": 0.7653,
+      "step": 1461
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.099118606218682e-05,
+      "loss": 0.7607,
+      "step": 1462
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0980505198549019e-05,
+      "loss": 0.7233,
+      "step": 1463
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0969823205364597e-05,
+      "loss": 0.7274,
+      "step": 1464
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0959140094939265e-05,
+      "loss": 0.7925,
+      "step": 1465
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0948455879580018e-05,
+      "loss": 0.8984,
+      "step": 1466
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.093777057159512e-05,
+      "loss": 0.7826,
+      "step": 1467
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0927084183294091e-05,
+      "loss": 0.8782,
+      "step": 1468
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0916396726987713e-05,
+      "loss": 0.7527,
+      "step": 1469
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0905708214987975e-05,
+      "loss": 0.7629,
+      "step": 1470
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.08950186596081e-05,
+      "loss": 0.7561,
+      "step": 1471
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0884328073162503e-05,
+      "loss": 0.7883,
+      "step": 1472
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0873636467966791e-05,
+      "loss": 0.7443,
+      "step": 1473
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0862943856337741e-05,
+      "loss": 0.8675,
+      "step": 1474
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0852250250593299e-05,
+      "loss": 0.7645,
+      "step": 1475
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0841555663052543e-05,
+      "loss": 0.7684,
+      "step": 1476
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0830860106035688e-05,
+      "loss": 0.7336,
+      "step": 1477
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0820163591864073e-05,
+      "loss": 0.7314,
+      "step": 1478
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0809466132860126e-05,
+      "loss": 0.722,
+      "step": 1479
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.079876774134737e-05,
+      "loss": 0.7625,
+      "step": 1480
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0788068429650407e-05,
+      "loss": 0.7812,
+      "step": 1481
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0777368210094895e-05,
+      "loss": 0.8566,
+      "step": 1482
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.076666709500753e-05,
+      "loss": 0.7533,
+      "step": 1483
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0755965096716058e-05,
+      "loss": 0.6562,
+      "step": 1484
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0745262227549223e-05,
+      "loss": 0.7676,
+      "step": 1485
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0734558499836782e-05,
+      "loss": 0.7332,
+      "step": 1486
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0723853925909484e-05,
+      "loss": 0.7807,
+      "step": 1487
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0713148518099046e-05,
+      "loss": 0.7661,
+      "step": 1488
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0702442288738144e-05,
+      "loss": 0.8056,
+      "step": 1489
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0691735250160414e-05,
+      "loss": 0.8018,
+      "step": 1490
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.068102741470041e-05,
+      "loss": 0.7995,
+      "step": 1491
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0670318794693607e-05,
+      "loss": 0.7328,
+      "step": 1492
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.065960940247639e-05,
+      "loss": 0.8241,
+      "step": 1493
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0648899250386027e-05,
+      "loss": 0.7617,
+      "step": 1494
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0638188350760661e-05,
+      "loss": 0.7449,
+      "step": 1495
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0627476715939307e-05,
+      "loss": 0.7893,
+      "step": 1496
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.061676435826181e-05,
+      "loss": 0.7941,
+      "step": 1497
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0606051290068861e-05,
+      "loss": 0.814,
+      "step": 1498
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0595337523701967e-05,
+      "loss": 0.7866,
+      "step": 1499
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0584623071503435e-05,
+      "loss": 0.683,
+      "step": 1500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0573907945816362e-05,
+      "loss": 0.7132,
+      "step": 1501
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.056319215898463e-05,
+      "loss": 0.8452,
+      "step": 1502
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.055247572335287e-05,
+      "loss": 0.7372,
+      "step": 1503
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0541758651266467e-05,
+      "loss": 0.8631,
+      "step": 1504
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0531040955071541e-05,
+      "loss": 0.785,
+      "step": 1505
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0520322647114928e-05,
+      "loss": 0.7865,
+      "step": 1506
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.050960373974417e-05,
+      "loss": 0.894,
+      "step": 1507
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0498884245307499e-05,
+      "loss": 0.7491,
+      "step": 1508
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.048816417615382e-05,
+      "loss": 0.7655,
+      "step": 1509
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0477443544632707e-05,
+      "loss": 0.7372,
+      "step": 1510
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0466722363094376e-05,
+      "loss": 0.7167,
+      "step": 1511
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0456000643889679e-05,
+      "loss": 0.7412,
+      "step": 1512
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.044527839937009e-05,
+      "loss": 0.7599,
+      "step": 1513
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0434555641887678e-05,
+      "loss": 0.7801,
+      "step": 1514
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0423832383795119e-05,
+      "loss": 0.7517,
+      "step": 1515
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.041310863744565e-05,
+      "loss": 0.7469,
+      "step": 1516
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.040238441519308e-05,
+      "loss": 0.8072,
+      "step": 1517
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.039165972939177e-05,
+      "loss": 0.8415,
+      "step": 1518
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0380934592396597e-05,
+      "loss": 0.7603,
+      "step": 1519
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0370209016562978e-05,
+      "loss": 0.8045,
+      "step": 1520
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0359483014246823e-05,
+      "loss": 0.756,
+      "step": 1521
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0348756597804537e-05,
+      "loss": 0.7698,
+      "step": 1522
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0338029779593007e-05,
+      "loss": 0.7624,
+      "step": 1523
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0327302571969574e-05,
+      "loss": 0.722,
+      "step": 1524
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0316574987292031e-05,
+      "loss": 0.8858,
+      "step": 1525
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.030584703791861e-05,
+      "loss": 0.7744,
+      "step": 1526
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0295118736207958e-05,
+      "loss": 0.7287,
+      "step": 1527
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0284390094519128e-05,
+      "loss": 0.7141,
+      "step": 1528
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0273661125211568e-05,
+      "loss": 0.7968,
+      "step": 1529
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0262931840645097e-05,
+      "loss": 0.7667,
+      "step": 1530
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0252202253179903e-05,
+      "loss": 0.8625,
+      "step": 1531
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0241472375176527e-05,
+      "loss": 0.8149,
+      "step": 1532
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0230742218995834e-05,
+      "loss": 0.7966,
+      "step": 1533
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0220011796999011e-05,
+      "loss": 0.7902,
+      "step": 1534
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0209281121547565e-05,
+      "loss": 0.7191,
+      "step": 1535
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0198550205003277e-05,
+      "loss": 0.7237,
+      "step": 1536
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0187819059728213e-05,
+      "loss": 0.7499,
+      "step": 1537
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.017708769808471e-05,
+      "loss": 0.9083,
+      "step": 1538
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.016635613243534e-05,
+      "loss": 0.9039,
+      "step": 1539
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0155624375142919e-05,
+      "loss": 0.7901,
+      "step": 1540
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0144892438570483e-05,
+      "loss": 0.808,
+      "step": 1541
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0134160335081278e-05,
+      "loss": 0.7638,
+      "step": 1542
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0123428077038728e-05,
+      "loss": 0.891,
+      "step": 1543
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0112695676806457e-05,
+      "loss": 0.7685,
+      "step": 1544
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.010196314674823e-05,
+      "loss": 0.7326,
+      "step": 1545
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0091230499227975e-05,
+      "loss": 0.7107,
+      "step": 1546
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0080497746609759e-05,
+      "loss": 0.894,
+      "step": 1547
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0069764901257757e-05,
+      "loss": 0.771,
+      "step": 1548
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0059031975536258e-05,
+      "loss": 0.7328,
+      "step": 1549
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0048298981809646e-05,
+      "loss": 0.7377,
+      "step": 1550
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0037565932442377e-05,
+      "loss": 0.7405,
+      "step": 1551
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0026832839798977e-05,
+      "loss": 0.7657,
+      "step": 1552
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.001609971624402e-05,
+      "loss": 0.8261,
+      "step": 1553
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0005366574142114e-05,
+      "loss": 0.7788,
+      "step": 1554
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.994633425857887e-06,
+      "loss": 0.7646,
+      "step": 1555
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.983900283755983e-06,
+      "loss": 0.8147,
+      "step": 1556
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.973167160201026e-06,
+      "loss": 0.8599,
+      "step": 1557
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.962434067557624e-06,
+      "loss": 0.7522,
+      "step": 1558
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.951701018190359e-06,
+      "loss": 0.7501,
+      "step": 1559
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.940968024463746e-06,
+      "loss": 0.8446,
+      "step": 1560
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.930235098742247e-06,
+      "loss": 0.8264,
+      "step": 1561
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.919502253390246e-06,
+      "loss": 0.8108,
+      "step": 1562
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.908769500772027e-06,
+      "loss": 0.7937,
+      "step": 1563
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.898036853251773e-06,
+      "loss": 0.7939,
+      "step": 1564
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.887304323193548e-06,
+      "loss": 0.8074,
+      "step": 1565
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.876571922961274e-06,
+      "loss": 0.7775,
+      "step": 1566
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.865839664918725e-06,
+      "loss": 0.6961,
+      "step": 1567
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.855107561429518e-06,
+      "loss": 0.7689,
+      "step": 1568
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.844375624857084e-06,
+      "loss": 0.7987,
+      "step": 1569
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.833643867564663e-06,
+      "loss": 0.7302,
+      "step": 1570
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.822912301915294e-06,
+      "loss": 0.7561,
+      "step": 1571
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.812180940271789e-06,
+      "loss": 0.7445,
+      "step": 1572
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.801449794996726e-06,
+      "loss": 0.7731,
+      "step": 1573
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.790718878452439e-06,
+      "loss": 0.8111,
+      "step": 1574
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.77998820300099e-06,
+      "loss": 0.7375,
+      "step": 1575
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.76925778100417e-06,
+      "loss": 0.751,
+      "step": 1576
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.758527624823476e-06,
+      "loss": 0.8973,
+      "step": 1577
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.747797746820098e-06,
+      "loss": 0.7684,
+      "step": 1578
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.737068159354907e-06,
+      "loss": 0.7693,
+      "step": 1579
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.726338874788437e-06,
+      "loss": 0.7587,
+      "step": 1580
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.715609905480875e-06,
+      "loss": 0.7907,
+      "step": 1581
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.704881263792044e-06,
+      "loss": 0.8263,
+      "step": 1582
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.694152962081392e-06,
+      "loss": 0.788,
+      "step": 1583
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.68342501270797e-06,
+      "loss": 0.7722,
+      "step": 1584
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.672697428030429e-06,
+      "loss": 0.7332,
+      "step": 1585
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.661970220406995e-06,
+      "loss": 0.8189,
+      "step": 1586
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.651243402195464e-06,
+      "loss": 0.9797,
+      "step": 1587
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.640516985753179e-06,
+      "loss": 0.7318,
+      "step": 1588
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.629790983437025e-06,
+      "loss": 0.7136,
+      "step": 1589
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.619065407603406e-06,
+      "loss": 0.7653,
+      "step": 1590
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.608340270608233e-06,
+      "loss": 0.7759,
+      "step": 1591
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.59761558480692e-06,
+      "loss": 0.7944,
+      "step": 1592
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.586891362554352e-06,
+      "loss": 0.8004,
+      "step": 1593
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.576167616204883e-06,
+      "loss": 0.6985,
+      "step": 1594
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.565444358112323e-06,
+      "loss": 0.7932,
+      "step": 1595
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.554721600629915e-06,
+      "loss": 0.7065,
+      "step": 1596
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.543999356110321e-06,
+      "loss": 0.754,
+      "step": 1597
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.533277636905626e-06,
+      "loss": 0.6836,
+      "step": 1598
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.522556455367295e-06,
+      "loss": 0.7412,
+      "step": 1599
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.511835823846181e-06,
+      "loss": 0.8502,
+      "step": 1600
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.501115754692503e-06,
+      "loss": 0.7755,
+      "step": 1601
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.49039626025583e-06,
+      "loss": 0.7929,
+      "step": 1602
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.47967735288507e-06,
+      "loss": 0.7557,
+      "step": 1603
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.468959044928459e-06,
+      "loss": 0.7725,
+      "step": 1604
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.458241348733535e-06,
+      "loss": 0.7975,
+      "step": 1605
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.447524276647132e-06,
+      "loss": 0.7023,
+      "step": 1606
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.436807841015371e-06,
+      "loss": 0.7688,
+      "step": 1607
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.426092054183638e-06,
+      "loss": 0.8416,
+      "step": 1608
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.415376928496565e-06,
+      "loss": 0.7416,
+      "step": 1609
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.404662476298033e-06,
+      "loss": 0.7147,
+      "step": 1610
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.393948709931139e-06,
+      "loss": 0.6954,
+      "step": 1611
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.38323564173819e-06,
+      "loss": 0.8764,
+      "step": 1612
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.372523284060695e-06,
+      "loss": 0.7885,
+      "step": 1613
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.36181164923934e-06,
+      "loss": 0.7271,
+      "step": 1614
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.351100749613975e-06,
+      "loss": 0.7704,
+      "step": 1615
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.340390597523611e-06,
+      "loss": 0.8014,
+      "step": 1616
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.329681205306395e-06,
+      "loss": 0.8241,
+      "step": 1617
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.31897258529959e-06,
+      "loss": 0.7604,
+      "step": 1618
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.308264749839586e-06,
+      "loss": 0.7553,
+      "step": 1619
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.297557711261856e-06,
+      "loss": 0.7221,
+      "step": 1620
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.286851481900956e-06,
+      "loss": 0.8027,
+      "step": 1621
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.276146074090517e-06,
+      "loss": 0.7644,
+      "step": 1622
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.26544150016322e-06,
+      "loss": 0.8464,
+      "step": 1623
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.254737772450777e-06,
+      "loss": 0.7709,
+      "step": 1624
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.244034903283944e-06,
+      "loss": 0.7825,
+      "step": 1625
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.233332904992469e-06,
+      "loss": 0.842,
+      "step": 1626
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.222631789905107e-06,
+      "loss": 0.7501,
+      "step": 1627
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.211931570349597e-06,
+      "loss": 0.7647,
+      "step": 1628
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.201232258652632e-06,
+      "loss": 0.797,
+      "step": 1629
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.19053386713988e-06,
+      "loss": 0.7471,
+      "step": 1630
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.179836408135934e-06,
+      "loss": 0.7492,
+      "step": 1631
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.169139893964314e-06,
+      "loss": 0.7216,
+      "step": 1632
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.158444336947462e-06,
+      "loss": 0.7321,
+      "step": 1633
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.147749749406708e-06,
+      "loss": 0.764,
+      "step": 1634
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.13705614366226e-06,
+      "loss": 0.803,
+      "step": 1635
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.126363532033214e-06,
+      "loss": 0.7717,
+      "step": 1636
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.115671926837502e-06,
+      "loss": 0.8796,
+      "step": 1637
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.104981340391904e-06,
+      "loss": 0.6849,
+      "step": 1638
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.094291785012028e-06,
+      "loss": 0.7087,
+      "step": 1639
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.083603273012292e-06,
+      "loss": 0.812,
+      "step": 1640
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.07291581670591e-06,
+      "loss": 0.8163,
+      "step": 1641
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.062229428404886e-06,
+      "loss": 0.7712,
+      "step": 1642
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.051544120419987e-06,
+      "loss": 0.7981,
+      "step": 1643
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.040859905060736e-06,
+      "loss": 0.809,
+      "step": 1644
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.030176794635408e-06,
+      "loss": 0.7591,
+      "step": 1645
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.019494801450988e-06,
+      "loss": 0.7341,
+      "step": 1646
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.008813937813183e-06,
+      "loss": 0.7495,
+      "step": 1647
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.998134216026404e-06,
+      "loss": 0.7541,
+      "step": 1648
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.987455648393738e-06,
+      "loss": 0.7853,
+      "step": 1649
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.976778247216943e-06,
+      "loss": 0.7682,
+      "step": 1650
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.96610202479644e-06,
+      "loss": 0.7187,
+      "step": 1651
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.955426993431286e-06,
+      "loss": 0.7779,
+      "step": 1652
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.944753165419162e-06,
+      "loss": 0.7797,
+      "step": 1653
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.934080553056378e-06,
+      "loss": 0.744,
+      "step": 1654
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.923409168637829e-06,
+      "loss": 0.8173,
+      "step": 1655
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.912739024456997e-06,
+      "loss": 0.8504,
+      "step": 1656
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.902070132805944e-06,
+      "loss": 0.7851,
+      "step": 1657
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.891402505975283e-06,
+      "loss": 0.7213,
+      "step": 1658
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.880736156254163e-06,
+      "loss": 0.7551,
+      "step": 1659
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.87007109593028e-06,
+      "loss": 0.7505,
+      "step": 1660
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.859407337289828e-06,
+      "loss": 0.7891,
+      "step": 1661
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.848744892617507e-06,
+      "loss": 0.8675,
+      "step": 1662
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.838083774196506e-06,
+      "loss": 0.7607,
+      "step": 1663
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.827423994308481e-06,
+      "loss": 0.7594,
+      "step": 1664
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.816765565233549e-06,
+      "loss": 0.7306,
+      "step": 1665
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.806108499250274e-06,
+      "loss": 0.8317,
+      "step": 1666
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.795452808635646e-06,
+      "loss": 0.7357,
+      "step": 1667
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.784798505665063e-06,
+      "loss": 0.7286,
+      "step": 1668
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.774145602612343e-06,
+      "loss": 0.8054,
+      "step": 1669
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.763494111749675e-06,
+      "loss": 0.7458,
+      "step": 1670
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.752844045347626e-06,
+      "loss": 0.9091,
+      "step": 1671
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.742195415675128e-06,
+      "loss": 0.8536,
+      "step": 1672
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.731548234999448e-06,
+      "loss": 0.7861,
+      "step": 1673
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.720902515586185e-06,
+      "loss": 0.7497,
+      "step": 1674
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.710258269699268e-06,
+      "loss": 0.7255,
+      "step": 1675
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.699615509600911e-06,
+      "loss": 0.8932,
+      "step": 1676
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.688974247551626e-06,
+      "loss": 0.8033,
+      "step": 1677
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.678334495810198e-06,
+      "loss": 0.7255,
+      "step": 1678
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.66769626663367e-06,
+      "loss": 0.8083,
+      "step": 1679
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.657059572277331e-06,
+      "loss": 0.7077,
+      "step": 1680
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.646424424994707e-06,
+      "loss": 0.7437,
+      "step": 1681
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.635790837037534e-06,
+      "loss": 0.7095,
+      "step": 1682
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.625158820655755e-06,
+      "loss": 0.7967,
+      "step": 1683
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.614528388097508e-06,
+      "loss": 0.7899,
+      "step": 1684
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.603899551609099e-06,
+      "loss": 0.7431,
+      "step": 1685
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.593272323434994e-06,
+      "loss": 0.7429,
+      "step": 1686
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.582646715817817e-06,
+      "loss": 0.8007,
+      "step": 1687
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.572022740998314e-06,
+      "loss": 0.7847,
+      "step": 1688
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.561400411215352e-06,
+      "loss": 0.7665,
+      "step": 1689
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.550779738705913e-06,
+      "loss": 0.7028,
+      "step": 1690
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.540160735705053e-06,
+      "loss": 0.7391,
+      "step": 1691
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.529543414445916e-06,
+      "loss": 0.7607,
+      "step": 1692
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.51892778715971e-06,
+      "loss": 0.754,
+      "step": 1693
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.508313866075686e-06,
+      "loss": 0.7107,
+      "step": 1694
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.497701663421127e-06,
+      "loss": 0.8004,
+      "step": 1695
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.487091191421346e-06,
+      "loss": 0.7489,
+      "step": 1696
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.476482462299656e-06,
+      "loss": 0.7785,
+      "step": 1697
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.465875488277358e-06,
+      "loss": 0.8164,
+      "step": 1698
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.455270281573745e-06,
+      "loss": 0.742,
+      "step": 1699
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.444666854406057e-06,
+      "loss": 0.7654,
+      "step": 1700
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.434065218989495e-06,
+      "loss": 0.7101,
+      "step": 1701
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.423465387537195e-06,
+      "loss": 0.7383,
+      "step": 1702
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.412867372260211e-06,
+      "loss": 0.742,
+      "step": 1703
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.402271185367504e-06,
+      "loss": 0.7538,
+      "step": 1704
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.391676839065937e-06,
+      "loss": 0.6777,
+      "step": 1705
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.381084345560245e-06,
+      "loss": 0.7997,
+      "step": 1706
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.370493717053026e-06,
+      "loss": 0.7724,
+      "step": 1707
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.35990496574474e-06,
+      "loss": 0.7891,
+      "step": 1708
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.349318103833679e-06,
+      "loss": 0.8498,
+      "step": 1709
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.33873314351595e-06,
+      "loss": 0.7921,
+      "step": 1710
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.328150096985485e-06,
+      "loss": 0.7431,
+      "step": 1711
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.317568976434002e-06,
+      "loss": 0.6997,
+      "step": 1712
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.306989794050998e-06,
+      "loss": 0.7397,
+      "step": 1713
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.296412562023745e-06,
+      "loss": 0.7707,
+      "step": 1714
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.285837292537264e-06,
+      "loss": 0.7653,
+      "step": 1715
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.275263997774313e-06,
+      "loss": 0.8486,
+      "step": 1716
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.26469268991538e-06,
+      "loss": 0.7841,
+      "step": 1717
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.25412338113866e-06,
+      "loss": 0.7721,
+      "step": 1718
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.243556083620046e-06,
+      "loss": 0.7209,
+      "step": 1719
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.232990809533118e-06,
+      "loss": 0.7516,
+      "step": 1720
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.222427571049118e-06,
+      "loss": 0.7674,
+      "step": 1721
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.211866380336947e-06,
+      "loss": 0.7902,
+      "step": 1722
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.20130724956315e-06,
+      "loss": 0.7401,
+      "step": 1723
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.190750190891891e-06,
+      "loss": 0.7078,
+      "step": 1724
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.180195216484953e-06,
+      "loss": 0.7658,
+      "step": 1725
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.169642338501717e-06,
+      "loss": 0.7463,
+      "step": 1726
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.15909156909915e-06,
+      "loss": 0.7723,
+      "step": 1727
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.148542920431784e-06,
+      "loss": 0.7108,
+      "step": 1728
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.137996404651716e-06,
+      "loss": 0.7469,
+      "step": 1729
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.127452033908579e-06,
+      "loss": 0.7225,
+      "step": 1730
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.116909820349537e-06,
+      "loss": 0.745,
+      "step": 1731
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.106369776119275e-06,
+      "loss": 0.7934,
+      "step": 1732
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.09583191335997e-06,
+      "loss": 0.7275,
+      "step": 1733
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.085296244211289e-06,
+      "loss": 0.7738,
+      "step": 1734
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.074762780810375e-06,
+      "loss": 0.8187,
+      "step": 1735
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.064231535291827e-06,
+      "loss": 0.7582,
+      "step": 1736
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.053702519787686e-06,
+      "loss": 0.7686,
+      "step": 1737
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.043175746427433e-06,
+      "loss": 0.8029,
+      "step": 1738
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.032651227337959e-06,
+      "loss": 0.7538,
+      "step": 1739
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.022128974643555e-06,
+      "loss": 0.8434,
+      "step": 1740
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.011609000465915e-06,
+      "loss": 0.7914,
+      "step": 1741
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.00109131692409e-06,
+      "loss": 0.7481,
+      "step": 1742
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.990575936134504e-06,
+      "loss": 0.7443,
+      "step": 1743
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.980062870210928e-06,
+      "loss": 0.7747,
+      "step": 1744
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.969552131264461e-06,
+      "loss": 0.816,
+      "step": 1745
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.959043731403521e-06,
+      "loss": 0.7569,
+      "step": 1746
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.948537682733841e-06,
+      "loss": 0.6954,
+      "step": 1747
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.938033997358437e-06,
+      "loss": 0.7469,
+      "step": 1748
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.927532687377597e-06,
+      "loss": 0.849,
+      "step": 1749
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.917033764888889e-06,
+      "loss": 0.8905,
+      "step": 1750
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.906537241987117e-06,
+      "loss": 0.8014,
+      "step": 1751
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.896043130764324e-06,
+      "loss": 0.8089,
+      "step": 1752
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.885551443309777e-06,
+      "loss": 0.7692,
+      "step": 1753
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.875062191709952e-06,
+      "loss": 0.8397,
+      "step": 1754
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.86457538804851e-06,
+      "loss": 0.741,
+      "step": 1755
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.854091044406304e-06,
+      "loss": 0.7839,
+      "step": 1756
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.843609172861347e-06,
+      "loss": 0.7957,
+      "step": 1757
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.8331297854888e-06,
+      "loss": 0.792,
+      "step": 1758
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.822652894360972e-06,
+      "loss": 0.7869,
+      "step": 1759
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.812178511547287e-06,
+      "loss": 0.7931,
+      "step": 1760
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.801706649114287e-06,
+      "loss": 0.7612,
+      "step": 1761
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.791237319125604e-06,
+      "loss": 0.7373,
+      "step": 1762
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.780770533641958e-06,
+      "loss": 0.7794,
+      "step": 1763
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.770306304721132e-06,
+      "loss": 0.7828,
+      "step": 1764
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.75984464441797e-06,
+      "loss": 0.7965,
+      "step": 1765
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.749385564784355e-06,
+      "loss": 0.7179,
+      "step": 1766
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.738929077869191e-06,
+      "loss": 0.7456,
+      "step": 1767
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.728475195718406e-06,
+      "loss": 0.8249,
+      "step": 1768
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.718023930374919e-06,
+      "loss": 0.7286,
+      "step": 1769
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.707575293878635e-06,
+      "loss": 0.8191,
+      "step": 1770
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.697129298266437e-06,
+      "loss": 0.8698,
+      "step": 1771
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.68668595557216e-06,
+      "loss": 0.7001,
+      "step": 1772
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.67624527782658e-06,
+      "loss": 0.7438,
+      "step": 1773
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.665807277057414e-06,
+      "loss": 0.6985,
+      "step": 1774
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.655371965289284e-06,
+      "loss": 0.8262,
+      "step": 1775
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.64493935454372e-06,
+      "loss": 0.8069,
+      "step": 1776
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.634509456839133e-06,
+      "loss": 0.7769,
+      "step": 1777
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.6240822841908205e-06,
+      "loss": 0.7407,
+      "step": 1778
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.6136578486109336e-06,
+      "loss": 0.7609,
+      "step": 1779
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.603236162108472e-06,
+      "loss": 0.804,
+      "step": 1780
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.592817236689263e-06,
+      "loss": 0.7097,
+      "step": 1781
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.582401084355964e-06,
+      "loss": 0.7587,
+      "step": 1782
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.571987717108029e-06,
+      "loss": 0.7611,
+      "step": 1783
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.561577146941705e-06,
+      "loss": 0.6631,
+      "step": 1784
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5511693858500205e-06,
+      "loss": 0.865,
+      "step": 1785
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.540764445822765e-06,
+      "loss": 0.712,
+      "step": 1786
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.530362338846474e-06,
+      "loss": 0.7717,
+      "step": 1787
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.519963076904432e-06,
+      "loss": 0.7534,
+      "step": 1788
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5095666719766315e-06,
+      "loss": 0.8134,
+      "step": 1789
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.499173136039781e-06,
+      "loss": 0.7534,
+      "step": 1790
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.488782481067286e-06,
+      "loss": 0.7799,
+      "step": 1791
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.478394719029227e-06,
+      "loss": 0.8421,
+      "step": 1792
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4680098618923556e-06,
+      "loss": 0.7975,
+      "step": 1793
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.457627921620078e-06,
+      "loss": 0.7352,
+      "step": 1794
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.447248910172437e-06,
+      "loss": 0.8328,
+      "step": 1795
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4368728395061e-06,
+      "loss": 0.8669,
+      "step": 1796
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.426499721574355e-06,
+      "loss": 0.8234,
+      "step": 1797
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.41612956832708e-06,
+      "loss": 0.9156,
+      "step": 1798
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.405762391710738e-06,
+      "loss": 0.8464,
+      "step": 1799
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.39539820366837e-06,
+      "loss": 0.716,
+      "step": 1800
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.385037016139568e-06,
+      "loss": 0.7842,
+      "step": 1801
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.374678841060465e-06,
+      "loss": 0.6956,
+      "step": 1802
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.364323690363735e-06,
+      "loss": 0.7692,
+      "step": 1803
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.353971575978556e-06,
+      "loss": 0.7015,
+      "step": 1804
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.34362250983061e-06,
+      "loss": 0.815,
+      "step": 1805
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3332765038420775e-06,
+      "loss": 0.7963,
+      "step": 1806
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3229335699316e-06,
+      "loss": 0.78,
+      "step": 1807
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.312593720014286e-06,
+      "loss": 0.7223,
+      "step": 1808
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.302256966001696e-06,
+      "loss": 0.7444,
+      "step": 1809
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.291923319801817e-06,
+      "loss": 0.748,
+      "step": 1810
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2815927933190564e-06,
+      "loss": 0.7848,
+      "step": 1811
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.271265398454233e-06,
+      "loss": 0.7198,
+      "step": 1812
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.260941147104552e-06,
+      "loss": 0.6767,
+      "step": 1813
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.250620051163599e-06,
+      "loss": 0.7131,
+      "step": 1814
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.24030212252133e-06,
+      "loss": 0.7359,
+      "step": 1815
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2299873730640425e-06,
+      "loss": 0.7411,
+      "step": 1816
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2196758146743765e-06,
+      "loss": 0.8645,
+      "step": 1817
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2093674592313e-06,
+      "loss": 0.9405,
+      "step": 1818
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.199062318610085e-06,
+      "loss": 0.7386,
+      "step": 1819
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1887604046823e-06,
+      "loss": 0.7168,
+      "step": 1820
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.178461729315802e-06,
+      "loss": 0.734,
+      "step": 1821
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1681663043747104e-06,
+      "loss": 0.7567,
+      "step": 1822
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.157874141719405e-06,
+      "loss": 0.7814,
+      "step": 1823
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.147585253206508e-06,
+      "loss": 0.7209,
+      "step": 1824
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.137299650688864e-06,
+      "loss": 0.7685,
+      "step": 1825
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.127017346015537e-06,
+      "loss": 0.7495,
+      "step": 1826
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.116738351031791e-06,
+      "loss": 0.6603,
+      "step": 1827
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.1064626775790754e-06,
+      "loss": 0.7572,
+      "step": 1828
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0961903374950144e-06,
+      "loss": 0.8303,
+      "step": 1829
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.085921342613396e-06,
+      "loss": 0.7102,
+      "step": 1830
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.075655704764146e-06,
+      "loss": 0.651,
+      "step": 1831
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0653934357733285e-06,
+      "loss": 0.771,
+      "step": 1832
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.05513454746313e-06,
+      "loss": 0.7451,
+      "step": 1833
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.044879051651836e-06,
+      "loss": 0.7472,
+      "step": 1834
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.034626960153821e-06,
+      "loss": 0.7564,
+      "step": 1835
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.02437828477955e-06,
+      "loss": 0.7997,
+      "step": 1836
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.014133037335543e-06,
+      "loss": 0.7653,
+      "step": 1837
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.003891229624371e-06,
+      "loss": 0.7272,
+      "step": 1838
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.993652873444648e-06,
+      "loss": 0.8121,
+      "step": 1839
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.98341798059101e-06,
+      "loss": 0.7356,
+      "step": 1840
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.973186562854095e-06,
+      "loss": 0.8148,
+      "step": 1841
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9629586320205515e-06,
+      "loss": 0.7254,
+      "step": 1842
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.952734199873002e-06,
+      "loss": 0.6938,
+      "step": 1843
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.942513278190036e-06,
+      "loss": 0.8282,
+      "step": 1844
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.932295878746212e-06,
+      "loss": 0.8342,
+      "step": 1845
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.922082013312017e-06,
+      "loss": 0.7232,
+      "step": 1846
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.911871693653871e-06,
+      "loss": 0.7813,
+      "step": 1847
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.9016649315341156e-06,
+      "loss": 0.7411,
+      "step": 1848
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.891461738710985e-06,
+      "loss": 0.6995,
+      "step": 1849
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.881262126938606e-06,
+      "loss": 0.7219,
+      "step": 1850
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.871066107966981e-06,
+      "loss": 0.7806,
+      "step": 1851
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.860873693541971e-06,
+      "loss": 0.7174,
+      "step": 1852
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.850684895405284e-06,
+      "loss": 0.7868,
+      "step": 1853
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.840499725294467e-06,
+      "loss": 0.7676,
+      "step": 1854
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.830318194942882e-06,
+      "loss": 0.826,
+      "step": 1855
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.820140316079698e-06,
+      "loss": 0.8078,
+      "step": 1856
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.809966100429883e-06,
+      "loss": 0.7606,
+      "step": 1857
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.799795559714182e-06,
+      "loss": 0.697,
+      "step": 1858
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.789628705649102e-06,
+      "loss": 0.7073,
+      "step": 1859
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7794655499469095e-06,
+      "loss": 0.695,
+      "step": 1860
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.769306104315609e-06,
+      "loss": 0.699,
+      "step": 1861
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7591503804589254e-06,
+      "loss": 0.8541,
+      "step": 1862
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7489983900763045e-06,
+      "loss": 0.7064,
+      "step": 1863
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.738850144862887e-06,
+      "loss": 0.8286,
+      "step": 1864
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.728705656509495e-06,
+      "loss": 0.9157,
+      "step": 1865
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.718564936702634e-06,
+      "loss": 0.7529,
+      "step": 1866
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.708427997124456e-06,
+      "loss": 0.7692,
+      "step": 1867
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.698294849452762e-06,
+      "loss": 0.9169,
+      "step": 1868
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.688165505360991e-06,
+      "loss": 0.8032,
+      "step": 1869
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.678039976518193e-06,
+      "loss": 0.7205,
+      "step": 1870
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6679182745890235e-06,
+      "loss": 0.7364,
+      "step": 1871
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.657800411233734e-06,
+      "loss": 0.785,
+      "step": 1872
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.647686398108148e-06,
+      "loss": 0.7223,
+      "step": 1873
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.637576246863655e-06,
+      "loss": 0.7772,
+      "step": 1874
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.627469969147202e-06,
+      "loss": 0.771,
+      "step": 1875
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.617367576601266e-06,
+      "loss": 0.8636,
+      "step": 1876
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.60726908086385e-06,
+      "loss": 0.8073,
+      "step": 1877
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.59717449356847e-06,
+      "loss": 0.8175,
+      "step": 1878
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.587083826344138e-06,
+      "loss": 0.7735,
+      "step": 1879
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.57699709081535e-06,
+      "loss": 0.7578,
+      "step": 1880
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.566914298602072e-06,
+      "loss": 0.7369,
+      "step": 1881
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.556835461319731e-06,
+      "loss": 0.7657,
+      "step": 1882
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.546760590579193e-06,
+      "loss": 0.7841,
+      "step": 1883
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.536689697986756e-06,
+      "loss": 0.7721,
+      "step": 1884
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.526622795144137e-06,
+      "loss": 0.7528,
+      "step": 1885
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.516559893648458e-06,
+      "loss": 0.778,
+      "step": 1886
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.506501005092223e-06,
+      "loss": 0.7649,
+      "step": 1887
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.496446141063324e-06,
+      "loss": 0.7248,
+      "step": 1888
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.486395313145012e-06,
+      "loss": 0.7263,
+      "step": 1889
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.476348532915884e-06,
+      "loss": 0.7626,
+      "step": 1890
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4663058119498825e-06,
+      "loss": 0.708,
+      "step": 1891
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.456267161816269e-06,
+      "loss": 0.696,
+      "step": 1892
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4462325940796114e-06,
+      "loss": 0.8089,
+      "step": 1893
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.436202120299786e-06,
+      "loss": 0.7506,
+      "step": 1894
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.426175752031941e-06,
+      "loss": 0.72,
+      "step": 1895
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.416153500826501e-06,
+      "loss": 0.7679,
+      "step": 1896
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.406135378229147e-06,
+      "loss": 0.8016,
+      "step": 1897
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.396121395780805e-06,
+      "loss": 0.7223,
+      "step": 1898
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.386111565017625e-06,
+      "loss": 0.774,
+      "step": 1899
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.376105897470988e-06,
+      "loss": 0.7677,
+      "step": 1900
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.366104404667462e-06,
+      "loss": 0.7012,
+      "step": 1901
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.356107098128816e-06,
+      "loss": 0.7865,
+      "step": 1902
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.346113989371998e-06,
+      "loss": 0.8749,
+      "step": 1903
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3361250899091154e-06,
+      "loss": 0.7828,
+      "step": 1904
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.326140411247421e-06,
+      "loss": 0.7295,
+      "step": 1905
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.316159964889321e-06,
+      "loss": 0.8084,
+      "step": 1906
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.306183762332332e-06,
+      "loss": 0.7973,
+      "step": 1907
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.296211815069084e-06,
+      "loss": 0.7505,
+      "step": 1908
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2862441345873134e-06,
+      "loss": 0.877,
+      "step": 1909
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.276280732369833e-06,
+      "loss": 0.7019,
+      "step": 1910
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.266321619894525e-06,
+      "loss": 0.7018,
+      "step": 1911
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2563668086343385e-06,
+      "loss": 0.7569,
+      "step": 1912
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.246416310057263e-06,
+      "loss": 0.7326,
+      "step": 1913
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.236470135626316e-06,
+      "loss": 0.7927,
+      "step": 1914
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.226528296799542e-06,
+      "loss": 0.7114,
+      "step": 1915
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.216590805029985e-06,
+      "loss": 0.809,
+      "step": 1916
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.206657671765679e-06,
+      "loss": 0.7131,
+      "step": 1917
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.196728908449646e-06,
+      "loss": 0.74,
+      "step": 1918
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.186804526519865e-06,
+      "loss": 0.8051,
+      "step": 1919
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.176884537409269e-06,
+      "loss": 0.7742,
+      "step": 1920
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.166968952545739e-06,
+      "loss": 0.8241,
+      "step": 1921
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.157057783352068e-06,
+      "loss": 0.7938,
+      "step": 1922
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.147151041245973e-06,
+      "loss": 0.7876,
+      "step": 1923
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.137248737640067e-06,
+      "loss": 0.7934,
+      "step": 1924
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.127350883941851e-06,
+      "loss": 0.7574,
+      "step": 1925
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1174574915537e-06,
+      "loss": 0.7326,
+      "step": 1926
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1075685718728465e-06,
+      "loss": 0.7206,
+      "step": 1927
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.097684136291375e-06,
+      "loss": 0.6658,
+      "step": 1928
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.087804196196201e-06,
+      "loss": 0.7958,
+      "step": 1929
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0779287629690605e-06,
+      "loss": 0.6962,
+      "step": 1930
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.068057847986504e-06,
+      "loss": 0.7377,
+      "step": 1931
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.058191462619867e-06,
+      "loss": 0.7224,
+      "step": 1932
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.048329618235273e-06,
+      "loss": 0.8182,
+      "step": 1933
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.038472326193616e-06,
+      "loss": 0.8477,
+      "step": 1934
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.028619597850541e-06,
+      "loss": 0.8212,
+      "step": 1935
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.018771444556436e-06,
+      "loss": 0.7478,
+      "step": 1936
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.008927877656424e-06,
+      "loss": 0.756,
+      "step": 1937
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.999088908490338e-06,
+      "loss": 0.7711,
+      "step": 1938
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.989254548392717e-06,
+      "loss": 0.7293,
+      "step": 1939
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.979424808692791e-06,
+      "loss": 0.742,
+      "step": 1940
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.969599700714466e-06,
+      "loss": 0.7288,
+      "step": 1941
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.959779235776312e-06,
+      "loss": 0.7755,
+      "step": 1942
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9499634251915535e-06,
+      "loss": 0.7654,
+      "step": 1943
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.940152280268049e-06,
+      "loss": 0.7435,
+      "step": 1944
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.930345812308282e-06,
+      "loss": 0.7551,
+      "step": 1945
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.920544032609353e-06,
+      "loss": 0.7166,
+      "step": 1946
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.910746952462958e-06,
+      "loss": 0.7431,
+      "step": 1947
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.900954583155376e-06,
+      "loss": 0.8743,
+      "step": 1948
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.891166935967468e-06,
+      "loss": 0.75,
+      "step": 1949
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.881384022174648e-06,
+      "loss": 0.6947,
+      "step": 1950
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8716058530468746e-06,
+      "loss": 0.8059,
+      "step": 1951
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.861832439848651e-06,
+      "loss": 0.7238,
+      "step": 1952
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.852063793838991e-06,
+      "loss": 0.7659,
+      "step": 1953
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.842299926271423e-06,
+      "loss": 0.7283,
+      "step": 1954
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.832540848393971e-06,
+      "loss": 0.6987,
+      "step": 1955
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8227865714491325e-06,
+      "loss": 0.6608,
+      "step": 1956
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.813037106673882e-06,
+      "loss": 0.811,
+      "step": 1957
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.803292465299657e-06,
+      "loss": 0.6971,
+      "step": 1958
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.79355265855232e-06,
+      "loss": 0.7692,
+      "step": 1959
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.783817697652178e-06,
+      "loss": 0.827,
+      "step": 1960
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.774087593813956e-06,
+      "loss": 0.8281,
+      "step": 1961
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.764362358246773e-06,
+      "loss": 0.7514,
+      "step": 1962
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7546420021541494e-06,
+      "loss": 0.7633,
+      "step": 1963
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.744926536733986e-06,
+      "loss": 0.7688,
+      "step": 1964
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.735215973178534e-06,
+      "loss": 0.7511,
+      "step": 1965
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.725510322674416e-06,
+      "loss": 0.8242,
+      "step": 1966
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.715809596402588e-06,
+      "loss": 0.7215,
+      "step": 1967
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.7061138055383296e-06,
+      "loss": 0.8112,
+      "step": 1968
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.696422961251236e-06,
+      "loss": 0.7014,
+      "step": 1969
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6867370747052105e-06,
+      "loss": 0.7779,
+      "step": 1970
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.677056157058436e-06,
+      "loss": 0.7806,
+      "step": 1971
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.667380219463377e-06,
+      "loss": 0.7229,
+      "step": 1972
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6577092730667616e-06,
+      "loss": 0.7607,
+      "step": 1973
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6480433290095595e-06,
+      "loss": 0.7248,
+      "step": 1974
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.638382398426987e-06,
+      "loss": 0.767,
+      "step": 1975
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.628726492448488e-06,
+      "loss": 0.8351,
+      "step": 1976
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.619075622197701e-06,
+      "loss": 0.6957,
+      "step": 1977
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.609429798792478e-06,
+      "loss": 0.7792,
+      "step": 1978
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.59978903334486e-06,
+      "loss": 0.7147,
+      "step": 1979
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.590153336961045e-06,
+      "loss": 0.6913,
+      "step": 1980
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.580522720741403e-06,
+      "loss": 0.7227,
+      "step": 1981
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.570897195780456e-06,
+      "loss": 0.7309,
+      "step": 1982
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.561276773166843e-06,
+      "loss": 0.7393,
+      "step": 1983
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5516614639833445e-06,
+      "loss": 0.7351,
+      "step": 1984
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.542051279306842e-06,
+      "loss": 0.7273,
+      "step": 1985
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.532446230208308e-06,
+      "loss": 0.6865,
+      "step": 1986
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.522846327752806e-06,
+      "loss": 0.7763,
+      "step": 1987
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.513251582999477e-06,
+      "loss": 0.8086,
+      "step": 1988
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.503662007001501e-06,
+      "loss": 0.7243,
+      "step": 1989
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.49407761080612e-06,
+      "loss": 0.7034,
+      "step": 1990
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.484498405454608e-06,
+      "loss": 0.7263,
+      "step": 1991
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.474924401982245e-06,
+      "loss": 0.7346,
+      "step": 1992
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.465355611418334e-06,
+      "loss": 0.7351,
+      "step": 1993
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.45579204478617e-06,
+      "loss": 0.6809,
+      "step": 1994
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.44623371310302e-06,
+      "loss": 0.7378,
+      "step": 1995
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.436680627380129e-06,
+      "loss": 0.7432,
+      "step": 1996
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.427132798622703e-06,
+      "loss": 0.715,
+      "step": 1997
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.417590237829878e-06,
+      "loss": 0.7523,
+      "step": 1998
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.4080529559947306e-06,
+      "loss": 0.6767,
+      "step": 1999
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.39852096410426e-06,
+      "loss": 0.7313,
+      "step": 2000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.388994273139358e-06,
+      "loss": 0.9657,
+      "step": 2001
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.379472894074824e-06,
+      "loss": 0.793,
+      "step": 2002
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.369956837879331e-06,
+      "loss": 0.8313,
+      "step": 2003
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3604461155154145e-06,
+      "loss": 0.7205,
+      "step": 2004
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.350940737939476e-06,
+      "loss": 0.8405,
+      "step": 2005
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.34144071610176e-06,
+      "loss": 0.6996,
+      "step": 2006
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3319460609463284e-06,
+      "loss": 0.7786,
+      "step": 2007
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.322456783411071e-06,
+      "loss": 0.7308,
+      "step": 2008
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.312972894427685e-06,
+      "loss": 0.7431,
+      "step": 2009
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.303494404921648e-06,
+      "loss": 0.7445,
+      "step": 2010
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.294021325812229e-06,
+      "loss": 0.7864,
+      "step": 2011
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.284553668012462e-06,
+      "loss": 0.7178,
+      "step": 2012
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.275091442429128e-06,
+      "loss": 0.7474,
+      "step": 2013
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.265634659962758e-06,
+      "loss": 0.6243,
+      "step": 2014
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.256183331507612e-06,
+      "loss": 0.6625,
+      "step": 2015
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.246737467951662e-06,
+      "loss": 0.6236,
+      "step": 2016
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.237297080176586e-06,
+      "loss": 0.6219,
+      "step": 2017
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.227862179057763e-06,
+      "loss": 0.6594,
+      "step": 2018
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.218432775464236e-06,
+      "loss": 0.648,
+      "step": 2019
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.209008880258723e-06,
+      "loss": 0.6997,
+      "step": 2020
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.199590504297605e-06,
+      "loss": 0.6036,
+      "step": 2021
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.190177658430881e-06,
+      "loss": 0.6412,
+      "step": 2022
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.180770353502202e-06,
+      "loss": 0.5224,
+      "step": 2023
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.171368600348832e-06,
+      "loss": 0.6201,
+      "step": 2024
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.161972409801625e-06,
+      "loss": 0.6361,
+      "step": 2025
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.152581792685043e-06,
+      "loss": 0.6317,
+      "step": 2026
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.143196759817122e-06,
+      "loss": 0.6456,
+      "step": 2027
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.13381732200946e-06,
+      "loss": 0.6734,
+      "step": 2028
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.124443490067215e-06,
+      "loss": 0.6506,
+      "step": 2029
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.115075274789091e-06,
+      "loss": 0.6158,
+      "step": 2030
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.105712686967308e-06,
+      "loss": 0.5772,
+      "step": 2031
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.096355737387615e-06,
+      "loss": 0.6701,
+      "step": 2032
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.087004436829265e-06,
+      "loss": 0.6603,
+      "step": 2033
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.077658796064995e-06,
+      "loss": 0.5882,
+      "step": 2034
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0683188258610296e-06,
+      "loss": 0.6811,
+      "step": 2035
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.058984536977062e-06,
+      "loss": 0.6855,
+      "step": 2036
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.04965594016623e-06,
+      "loss": 0.5852,
+      "step": 2037
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.040333046175125e-06,
+      "loss": 0.606,
+      "step": 2038
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.03101586574377e-06,
+      "loss": 0.5763,
+      "step": 2039
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.021704409605591e-06,
+      "loss": 0.5564,
+      "step": 2040
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.012398688487434e-06,
+      "loss": 0.6035,
+      "step": 2041
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0030987131095395e-06,
+      "loss": 0.566,
+      "step": 2042
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.993804494185512e-06,
+      "loss": 0.5904,
+      "step": 2043
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9845160424223415e-06,
+      "loss": 0.6115,
+      "step": 2044
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9752333685203715e-06,
+      "loss": 0.6584,
+      "step": 2045
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9659564831732786e-06,
+      "loss": 0.6211,
+      "step": 2046
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.956685397068082e-06,
+      "loss": 0.6445,
+      "step": 2047
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.947420120885121e-06,
+      "loss": 0.7305,
+      "step": 2048
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.938160665298031e-06,
+      "loss": 0.5883,
+      "step": 2049
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9289070409737496e-06,
+      "loss": 0.5784,
+      "step": 2050
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.919659258572501e-06,
+      "loss": 0.6296,
+      "step": 2051
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.910417328747767e-06,
+      "loss": 0.6753,
+      "step": 2052
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.901181262146298e-06,
+      "loss": 0.6486,
+      "step": 2053
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.89195106940809e-06,
+      "loss": 0.5897,
+      "step": 2054
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.88272676116636e-06,
+      "loss": 0.7349,
+      "step": 2055
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8735083480475594e-06,
+      "loss": 0.5733,
+      "step": 2056
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8642958406713485e-06,
+      "loss": 0.6676,
+      "step": 2057
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8550892496505685e-06,
+      "loss": 0.6027,
+      "step": 2058
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.845888585591263e-06,
+      "loss": 0.6004,
+      "step": 2059
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.836693859092643e-06,
+      "loss": 0.7052,
+      "step": 2060
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.827505080747069e-06,
+      "loss": 0.6804,
+      "step": 2061
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8183222611400606e-06,
+      "loss": 0.6018,
+      "step": 2062
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.809145410850273e-06,
+      "loss": 0.6575,
+      "step": 2063
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.799974540449475e-06,
+      "loss": 0.6477,
+      "step": 2064
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.790809660502555e-06,
+      "loss": 0.6565,
+      "step": 2065
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.781650781567501e-06,
+      "loss": 0.6336,
+      "step": 2066
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.772497914195382e-06,
+      "loss": 0.6281,
+      "step": 2067
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.763351068930341e-06,
+      "loss": 0.632,
+      "step": 2068
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.754210256309591e-06,
+      "loss": 0.6141,
+      "step": 2069
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.745075486863394e-06,
+      "loss": 0.5917,
+      "step": 2070
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.73594677111504e-06,
+      "loss": 0.6357,
+      "step": 2071
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.726824119580857e-06,
+      "loss": 0.5792,
+      "step": 2072
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.717707542770187e-06,
+      "loss": 0.6594,
+      "step": 2073
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.708597051185359e-06,
+      "loss": 0.5889,
+      "step": 2074
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.699492655321711e-06,
+      "loss": 0.6256,
+      "step": 2075
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.69039436566755e-06,
+      "loss": 0.6027,
+      "step": 2076
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.681302192704141e-06,
+      "loss": 0.627,
+      "step": 2077
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.672216146905717e-06,
+      "loss": 0.6678,
+      "step": 2078
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.663136238739446e-06,
+      "loss": 0.6499,
+      "step": 2079
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.654062478665421e-06,
+      "loss": 0.5966,
+      "step": 2080
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.644994877136656e-06,
+      "loss": 0.6129,
+      "step": 2081
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.635933444599079e-06,
+      "loss": 0.6109,
+      "step": 2082
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.626878191491492e-06,
+      "loss": 0.6259,
+      "step": 2083
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6178291282455945e-06,
+      "loss": 0.6591,
+      "step": 2084
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6087862652859535e-06,
+      "loss": 0.7531,
+      "step": 2085
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.599749613029984e-06,
+      "loss": 0.6741,
+      "step": 2086
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.590719181887952e-06,
+      "loss": 0.6303,
+      "step": 2087
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.581694982262965e-06,
+      "loss": 0.6663,
+      "step": 2088
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.572677024550933e-06,
+      "loss": 0.5724,
+      "step": 2089
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5636653191405915e-06,
+      "loss": 0.6367,
+      "step": 2090
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.55465987641347e-06,
+      "loss": 0.6652,
+      "step": 2091
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.545660706743874e-06,
+      "loss": 0.5973,
+      "step": 2092
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.536667820498894e-06,
+      "loss": 0.6315,
+      "step": 2093
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.527681228038379e-06,
+      "loss": 0.6018,
+      "step": 2094
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.518700939714921e-06,
+      "loss": 0.697,
+      "step": 2095
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5097269658738564e-06,
+      "loss": 0.5974,
+      "step": 2096
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5007593168532506e-06,
+      "loss": 0.6605,
+      "step": 2097
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.4917980029838695e-06,
+      "loss": 0.5861,
+      "step": 2098
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4828430345891924e-06,
+      "loss": 0.6176,
+      "step": 2099
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4738944219853895e-06,
+      "loss": 0.6656,
+      "step": 2100
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.464952175481296e-06,
+      "loss": 0.6507,
+      "step": 2101
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.456016305378429e-06,
+      "loss": 0.614,
+      "step": 2102
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.447086821970953e-06,
+      "loss": 0.6232,
+      "step": 2103
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.438163735545668e-06,
+      "loss": 0.6229,
+      "step": 2104
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.429247056382019e-06,
+      "loss": 0.6219,
+      "step": 2105
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.420336794752063e-06,
+      "loss": 0.5691,
+      "step": 2106
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.411432960920456e-06,
+      "loss": 0.5759,
+      "step": 2107
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.402535565144462e-06,
+      "loss": 0.6835,
+      "step": 2108
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.393644617673926e-06,
+      "loss": 0.576,
+      "step": 2109
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.384760128751253e-06,
+      "loss": 0.6014,
+      "step": 2110
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.375882108611421e-06,
+      "loss": 0.6153,
+      "step": 2111
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.367010567481956e-06,
+      "loss": 0.639,
+      "step": 2112
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.358145515582904e-06,
+      "loss": 0.6748,
+      "step": 2113
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3492869631268545e-06,
+      "loss": 0.6878,
+      "step": 2114
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.340434920318902e-06,
+      "loss": 0.5973,
+      "step": 2115
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.331589397356637e-06,
+      "loss": 0.635,
+      "step": 2116
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3227504044301475e-06,
+      "loss": 0.5855,
+      "step": 2117
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3139179517219955e-06,
+      "loss": 0.7202,
+      "step": 2118
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.305092049407203e-06,
+      "loss": 0.6054,
+      "step": 2119
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.296272707653256e-06,
+      "loss": 0.6382,
+      "step": 2120
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.28745993662008e-06,
+      "loss": 0.5988,
+      "step": 2121
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.278653746460022e-06,
+      "loss": 0.5707,
+      "step": 2122
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.269854147317857e-06,
+      "loss": 0.5871,
+      "step": 2123
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.26106114933077e-06,
+      "loss": 0.6578,
+      "step": 2124
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.252274762628329e-06,
+      "loss": 0.6834,
+      "step": 2125
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.243494997332495e-06,
+      "loss": 0.5968,
+      "step": 2126
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.234721863557604e-06,
+      "loss": 0.5949,
+      "step": 2127
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.22595537141034e-06,
+      "loss": 0.59,
+      "step": 2128
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.217195530989746e-06,
+      "loss": 0.6204,
+      "step": 2129
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.2084423523872056e-06,
+      "loss": 0.5774,
+      "step": 2130
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1996958456864125e-06,
+      "loss": 0.6516,
+      "step": 2131
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.190956020963387e-06,
+      "loss": 0.6364,
+      "step": 2132
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.182222888286453e-06,
+      "loss": 0.6129,
+      "step": 2133
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.173496457716214e-06,
+      "loss": 0.6168,
+      "step": 2134
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.164776739305563e-06,
+      "loss": 0.5819,
+      "step": 2135
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.156063743099658e-06,
+      "loss": 0.573,
+      "step": 2136
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.147357479135906e-06,
+      "loss": 0.6516,
+      "step": 2137
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.138657957443969e-06,
+      "loss": 0.6143,
+      "step": 2138
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1299651880457396e-06,
+      "loss": 0.6322,
+      "step": 2139
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.121279180955321e-06,
+      "loss": 0.6607,
+      "step": 2140
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.112599946179039e-06,
+      "loss": 0.5854,
+      "step": 2141
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.103927493715416e-06,
+      "loss": 0.5433,
+      "step": 2142
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.095261833555152e-06,
+      "loss": 0.6163,
+      "step": 2143
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.086602975681132e-06,
+      "loss": 0.596,
+      "step": 2144
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.077950930068404e-06,
+      "loss": 0.5952,
+      "step": 2145
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.069305706684158e-06,
+      "loss": 0.6056,
+      "step": 2146
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.060667315487736e-06,
+      "loss": 0.5671,
+      "step": 2147
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.05203576643061e-06,
+      "loss": 0.5514,
+      "step": 2148
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.043411069456356e-06,
+      "loss": 0.7218,
+      "step": 2149
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.034793234500672e-06,
+      "loss": 0.6662,
+      "step": 2150
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0261822714913454e-06,
+      "loss": 0.6561,
+      "step": 2151
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.017578190348242e-06,
+      "loss": 0.5675,
+      "step": 2152
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.008981000983304e-06,
+      "loss": 0.6126,
+      "step": 2153
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.00039071330054e-06,
+      "loss": 0.601,
+      "step": 2154
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.991807337195995e-06,
+      "loss": 0.6047,
+      "step": 2155
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.983230882557761e-06,
+      "loss": 0.6306,
+      "step": 2156
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.97466135926596e-06,
+      "loss": 0.5898,
+      "step": 2157
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.966098777192713e-06,
+      "loss": 0.5997,
+      "step": 2158
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.957543146202161e-06,
+      "loss": 0.5637,
+      "step": 2159
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9489944761504336e-06,
+      "loss": 0.6007,
+      "step": 2160
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.940452776885633e-06,
+      "loss": 0.5773,
+      "step": 2161
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.931918058247839e-06,
+      "loss": 0.628,
+      "step": 2162
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.923390330069092e-06,
+      "loss": 0.6018,
+      "step": 2163
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.91486960217337e-06,
+      "loss": 0.5437,
+      "step": 2164
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.906355884376593e-06,
+      "loss": 0.5833,
+      "step": 2165
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.897849186486607e-06,
+      "loss": 0.6827,
+      "step": 2166
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.8893495183031646e-06,
+      "loss": 0.6366,
+      "step": 2167
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.8808568896179234e-06,
+      "loss": 0.6953,
+      "step": 2168
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.872371310214438e-06,
+      "loss": 0.63,
+      "step": 2169
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.863892789868127e-06,
+      "loss": 0.5672,
+      "step": 2170
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.855421338346289e-06,
+      "loss": 0.6141,
+      "step": 2171
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.846956965408082e-06,
+      "loss": 0.602,
+      "step": 2172
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.838499680804495e-06,
+      "loss": 0.5818,
+      "step": 2173
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.830049494278364e-06,
+      "loss": 0.6331,
+      "step": 2174
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.821606415564348e-06,
+      "loss": 0.6282,
+      "step": 2175
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8131704543889024e-06,
+      "loss": 0.6021,
+      "step": 2176
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8047416204703023e-06,
+      "loss": 0.6206,
+      "step": 2177
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7963199235186064e-06,
+      "loss": 0.6582,
+      "step": 2178
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.787905373235643e-06,
+      "loss": 0.6322,
+      "step": 2179
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7794979793150153e-06,
+      "loss": 0.647,
+      "step": 2180
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7710977514420853e-06,
+      "loss": 0.5608,
+      "step": 2181
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7627046992939486e-06,
+      "loss": 0.606,
+      "step": 2182
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7543188325394443e-06,
+      "loss": 0.6198,
+      "step": 2183
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.745940160839133e-06,
+      "loss": 0.5901,
+      "step": 2184
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.737568693845278e-06,
+      "loss": 0.6767,
+      "step": 2185
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7292044412018523e-06,
+      "loss": 0.6278,
+      "step": 2186
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.720847412544517e-06,
+      "loss": 0.6538,
+      "step": 2187
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7124976175006034e-06,
+      "loss": 0.6844,
+      "step": 2188
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7041550656891168e-06,
+      "loss": 0.5956,
+      "step": 2189
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6958197667207198e-06,
+      "loss": 0.6119,
+      "step": 2190
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.68749173019771e-06,
+      "loss": 0.5991,
+      "step": 2191
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6791709657140275e-06,
+      "loss": 0.5925,
+      "step": 2192
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6708574828552356e-06,
+      "loss": 0.5941,
+      "step": 2193
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6625512911984995e-06,
+      "loss": 0.593,
+      "step": 2194
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.654252400312596e-06,
+      "loss": 0.5992,
+      "step": 2195
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6459608197578867e-06,
+      "loss": 0.5715,
+      "step": 2196
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6376765590863073e-06,
+      "loss": 0.5756,
+      "step": 2197
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6293996278413667e-06,
+      "loss": 0.5939,
+      "step": 2198
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6211300355581347e-06,
+      "loss": 0.6181,
+      "step": 2199
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6128677917632117e-06,
+      "loss": 0.5297,
+      "step": 2200
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6046129059747493e-06,
+      "loss": 0.5941,
+      "step": 2201
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5963653877024075e-06,
+      "loss": 0.6238,
+      "step": 2202
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5881252464473705e-06,
+      "loss": 0.5805,
+      "step": 2203
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5798924917023227e-06,
+      "loss": 0.5428,
+      "step": 2204
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5716671329514296e-06,
+      "loss": 0.6496,
+      "step": 2205
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5634491796703476e-06,
+      "loss": 0.6728,
+      "step": 2206
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5552386413261998e-06,
+      "loss": 0.5828,
+      "step": 2207
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5470355273775593e-06,
+      "loss": 0.5686,
+      "step": 2208
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5388398472744545e-06,
+      "loss": 0.6017,
+      "step": 2209
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5306516104583523e-06,
+      "loss": 0.5906,
+      "step": 2210
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.522470826362132e-06,
+      "loss": 0.6378,
+      "step": 2211
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5142975044101e-06,
+      "loss": 0.635,
+      "step": 2212
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5061316540179648e-06,
+      "loss": 0.6266,
+      "step": 2213
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4979732845928217e-06,
+      "loss": 0.6415,
+      "step": 2214
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4898224055331453e-06,
+      "loss": 0.6343,
+      "step": 2215
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4816790262287938e-06,
+      "loss": 0.5755,
+      "step": 2216
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4735431560609776e-06,
+      "loss": 0.5721,
+      "step": 2217
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.465414804402255e-06,
+      "loss": 0.6395,
+      "step": 2218
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.457293980616526e-06,
+      "loss": 0.6264,
+      "step": 2219
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.449180694059022e-06,
+      "loss": 0.5488,
+      "step": 2220
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.441074954076281e-06,
+      "loss": 0.6496,
+      "step": 2221
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4329767700061577e-06,
+      "loss": 0.6973,
+      "step": 2222
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.424886151177801e-06,
+      "loss": 0.5935,
+      "step": 2223
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4168031069116345e-06,
+      "loss": 0.6071,
+      "step": 2224
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4087276465193686e-06,
+      "loss": 0.567,
+      "step": 2225
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4006597793039732e-06,
+      "loss": 0.6041,
+      "step": 2226
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3925995145596635e-06,
+      "loss": 0.5629,
+      "step": 2227
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.384546861571907e-06,
+      "loss": 0.5835,
+      "step": 2228
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.376501829617399e-06,
+      "loss": 0.6231,
+      "step": 2229
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.36846442796405e-06,
+      "loss": 0.6769,
+      "step": 2230
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3604346658709865e-06,
+      "loss": 0.5244,
+      "step": 2231
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3524125525885355e-06,
+      "loss": 0.6596,
+      "step": 2232
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3443980973582033e-06,
+      "loss": 0.5734,
+      "step": 2233
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3363913094126843e-06,
+      "loss": 0.6124,
+      "step": 2234
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3283921979758384e-06,
+      "loss": 0.5617,
+      "step": 2235
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.320400772262672e-06,
+      "loss": 0.5884,
+      "step": 2236
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.312417041479351e-06,
+      "loss": 0.5711,
+      "step": 2237
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.304441014823172e-06,
+      "loss": 0.5863,
+      "step": 2238
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.2964727014825505e-06,
+      "loss": 0.6192,
+      "step": 2239
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2885121106370234e-06,
+      "loss": 0.5902,
+      "step": 2240
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2805592514572327e-06,
+      "loss": 0.641,
+      "step": 2241
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2726141331049035e-06,
+      "loss": 0.5683,
+      "step": 2242
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2646767647328515e-06,
+      "loss": 0.5677,
+      "step": 2243
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2567471554849663e-06,
+      "loss": 0.5541,
+      "step": 2244
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.24882531449619e-06,
+      "loss": 0.6217,
+      "step": 2245
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.240911250892522e-06,
+      "loss": 0.594,
+      "step": 2246
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2330049737910063e-06,
+      "loss": 0.6791,
+      "step": 2247
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2251064922997032e-06,
+      "loss": 0.5748,
+      "step": 2248
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.217215815517706e-06,
+      "loss": 0.6549,
+      "step": 2249
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2093329525351127e-06,
+      "loss": 0.6742,
+      "step": 2250
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2014579124330125e-06,
+      "loss": 0.6344,
+      "step": 2251
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1935907042834926e-06,
+      "loss": 0.5713,
+      "step": 2252
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.185731337149617e-06,
+      "loss": 0.6515,
+      "step": 2253
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1778798200854078e-06,
+      "loss": 0.6091,
+      "step": 2254
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.170036162135852e-06,
+      "loss": 0.5572,
+      "step": 2255
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1622003723368845e-06,
+      "loss": 0.5299,
+      "step": 2256
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.154372459715366e-06,
+      "loss": 0.5947,
+      "step": 2257
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.146552433289091e-06,
+      "loss": 0.5885,
+      "step": 2258
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1387403020667708e-06,
+      "loss": 0.6032,
+      "step": 2259
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.13093607504801e-06,
+      "loss": 0.5671,
+      "step": 2260
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1231397612233193e-06,
+      "loss": 0.5532,
+      "step": 2261
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.115351369574092e-06,
+      "loss": 0.5563,
+      "step": 2262
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.107570909072587e-06,
+      "loss": 0.601,
+      "step": 2263
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0997983886819314e-06,
+      "loss": 0.6233,
+      "step": 2264
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0920338173561117e-06,
+      "loss": 0.6487,
+      "step": 2265
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.084277204039943e-06,
+      "loss": 0.5923,
+      "step": 2266
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.076528557669083e-06,
+      "loss": 0.6002,
+      "step": 2267
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0687878871700126e-06,
+      "loss": 0.581,
+      "step": 2268
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0610552014600158e-06,
+      "loss": 0.597,
+      "step": 2269
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0533305094471844e-06,
+      "loss": 0.6174,
+      "step": 2270
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.045613820030404e-06,
+      "loss": 0.5279,
+      "step": 2271
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.037905142099331e-06,
+      "loss": 0.5972,
+      "step": 2272
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0302044845344015e-06,
+      "loss": 0.6111,
+      "step": 2273
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0225118562068125e-06,
+      "loss": 0.5981,
+      "step": 2274
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0148272659785027e-06,
+      "loss": 0.6494,
+      "step": 2275
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.007150722702158e-06,
+      "loss": 0.6326,
+      "step": 2276
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.999482235221197e-06,
+      "loss": 0.5645,
+      "step": 2277
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9918218123697475e-06,
+      "loss": 0.5341,
+      "step": 2278
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.984169462972655e-06,
+      "loss": 0.5757,
+      "step": 2279
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9765251958454645e-06,
+      "loss": 0.6102,
+      "step": 2280
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.968889019794402e-06,
+      "loss": 0.5987,
+      "step": 2281
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.961260943616383e-06,
+      "loss": 0.5708,
+      "step": 2282
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9536409760989893e-06,
+      "loss": 0.6379,
+      "step": 2283
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.946029126020453e-06,
+      "loss": 0.6753,
+      "step": 2284
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.938425402149665e-06,
+      "loss": 0.6833,
+      "step": 2285
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9308298132461533e-06,
+      "loss": 0.602,
+      "step": 2286
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9232423680600675e-06,
+      "loss": 0.673,
+      "step": 2287
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.915663075332181e-06,
+      "loss": 0.6605,
+      "step": 2288
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9080919437938815e-06,
+      "loss": 0.6226,
+      "step": 2289
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9005289821671377e-06,
+      "loss": 0.6915,
+      "step": 2290
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8929741991645245e-06,
+      "loss": 0.6061,
+      "step": 2291
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.885427603489187e-06,
+      "loss": 0.577,
+      "step": 2292
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8778892038348347e-06,
+      "loss": 0.5793,
+      "step": 2293
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8703590088857424e-06,
+      "loss": 0.6977,
+      "step": 2294
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8628370273167337e-06,
+      "loss": 0.5793,
+      "step": 2295
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8553232677931606e-06,
+      "loss": 0.5938,
+      "step": 2296
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8478177389709118e-06,
+      "loss": 0.6104,
+      "step": 2297
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8403204494963966e-06,
+      "loss": 0.5544,
+      "step": 2298
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.83283140800652e-06,
+      "loss": 0.5673,
+      "step": 2299
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8253506231286963e-06,
+      "loss": 0.5661,
+      "step": 2300
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.81787810348083e-06,
+      "loss": 0.6197,
+      "step": 2301
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.81041385767129e-06,
+      "loss": 0.5865,
+      "step": 2302
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.802957894298929e-06,
+      "loss": 0.6214,
+      "step": 2303
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.795510221953053e-06,
+      "loss": 0.584,
+      "step": 2304
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.788070849213409e-06,
+      "loss": 0.5656,
+      "step": 2305
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7806397846501944e-06,
+      "loss": 0.5595,
+      "step": 2306
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.773217036824033e-06,
+      "loss": 0.6258,
+      "step": 2307
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7658026142859594e-06,
+      "loss": 0.5405,
+      "step": 2308
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7583965255774246e-06,
+      "loss": 0.584,
+      "step": 2309
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.750998779230284e-06,
+      "loss": 0.6219,
+      "step": 2310
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7436093837667687e-06,
+      "loss": 0.6221,
+      "step": 2311
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.736228347699499e-06,
+      "loss": 0.6142,
+      "step": 2312
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.728855679531468e-06,
+      "loss": 0.6055,
+      "step": 2313
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7214913877560166e-06,
+      "loss": 0.5638,
+      "step": 2314
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7141354808568487e-06,
+      "loss": 0.5804,
+      "step": 2315
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7067879673080055e-06,
+      "loss": 0.6073,
+      "step": 2316
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.699448855573852e-06,
+      "loss": 0.5731,
+      "step": 2317
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.692118154109082e-06,
+      "loss": 0.5893,
+      "step": 2318
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6847958713587007e-06,
+      "loss": 0.6261,
+      "step": 2319
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6774820157580084e-06,
+      "loss": 0.5965,
+      "step": 2320
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.670176595732602e-06,
+      "loss": 0.5772,
+      "step": 2321
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6628796196983632e-06,
+      "loss": 0.7064,
+      "step": 2322
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.655591096061438e-06,
+      "loss": 0.6266,
+      "step": 2323
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6483110332182425e-06,
+      "loss": 0.6092,
+      "step": 2324
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6410394395554463e-06,
+      "loss": 0.5731,
+      "step": 2325
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6337763234499535e-06,
+      "loss": 0.6592,
+      "step": 2326
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6265216932689118e-06,
+      "loss": 0.5597,
+      "step": 2327
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6192755573696916e-06,
+      "loss": 0.59,
+      "step": 2328
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6120379240998708e-06,
+      "loss": 0.5436,
+      "step": 2329
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6048088017972385e-06,
+      "loss": 0.6364,
+      "step": 2330
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.597588198789782e-06,
+      "loss": 0.6714,
+      "step": 2331
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5903761233956636e-06,
+      "loss": 0.5418,
+      "step": 2332
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5831725839232315e-06,
+      "loss": 0.568,
+      "step": 2333
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.575977588670999e-06,
+      "loss": 0.5785,
+      "step": 2334
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5687911459276292e-06,
+      "loss": 0.5882,
+      "step": 2335
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5616132639719425e-06,
+      "loss": 0.6505,
+      "step": 2336
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5544439510728947e-06,
+      "loss": 0.5743,
+      "step": 2337
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.547283215489562e-06,
+      "loss": 0.5478,
+      "step": 2338
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5401310654711485e-06,
+      "loss": 0.7006,
+      "step": 2339
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.532987509256968e-06,
+      "loss": 0.6267,
+      "step": 2340
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5258525550764246e-06,
+      "loss": 0.6337,
+      "step": 2341
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5187262111490228e-06,
+      "loss": 0.6289,
+      "step": 2342
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.511608485684347e-06,
+      "loss": 0.6753,
+      "step": 2343
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5044993868820455e-06,
+      "loss": 0.551,
+      "step": 2344
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.497398922931835e-06,
+      "loss": 0.567,
+      "step": 2345
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.490307102013487e-06,
+      "loss": 0.5818,
+      "step": 2346
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4832239322968086e-06,
+      "loss": 0.6551,
+      "step": 2347
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4761494219416447e-06,
+      "loss": 0.5992,
+      "step": 2348
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4690835790978705e-06,
+      "loss": 0.5512,
+      "step": 2349
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4620264119053627e-06,
+      "loss": 0.6753,
+      "step": 2350
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4549779284940135e-06,
+      "loss": 0.6165,
+      "step": 2351
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.447938136983714e-06,
+      "loss": 0.6115,
+      "step": 2352
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.44090704548433e-06,
+      "loss": 0.5957,
+      "step": 2353
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4338846620957157e-06,
+      "loss": 0.5943,
+      "step": 2354
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.426870994907692e-06,
+      "loss": 0.5479,
+      "step": 2355
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4198660520000318e-06,
+      "loss": 0.6222,
+      "step": 2356
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4128698414424646e-06,
+      "loss": 0.5189,
+      "step": 2357
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.405882371294661e-06,
+      "loss": 0.6203,
+      "step": 2358
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3989036496062147e-06,
+      "loss": 0.5596,
+      "step": 2359
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3919336844166497e-06,
+      "loss": 0.536,
+      "step": 2360
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3849724837553957e-06,
+      "loss": 0.6046,
+      "step": 2361
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.378020055641792e-06,
+      "loss": 0.5995,
+      "step": 2362
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3710764080850646e-06,
+      "loss": 0.5825,
+      "step": 2363
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3641415490843312e-06,
+      "loss": 0.626,
+      "step": 2364
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3572154866285833e-06,
+      "loss": 0.5389,
+      "step": 2365
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3502982286966737e-06,
+      "loss": 0.5906,
+      "step": 2366
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.343389783257317e-06,
+      "loss": 0.5922,
+      "step": 2367
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3364901582690804e-06,
+      "loss": 0.5413,
+      "step": 2368
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.329599361680357e-06,
+      "loss": 0.6662,
+      "step": 2369
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3227174014293798e-06,
+      "loss": 0.5817,
+      "step": 2370
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.315844285444203e-06,
+      "loss": 0.5835,
+      "step": 2371
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.308980021642683e-06,
+      "loss": 0.5628,
+      "step": 2372
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3021246179324886e-06,
+      "loss": 0.5426,
+      "step": 2373
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.295278082211073e-06,
+      "loss": 0.5711,
+      "step": 2374
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.288440422365681e-06,
+      "loss": 0.6231,
+      "step": 2375
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2816116462733295e-06,
+      "loss": 0.5862,
+      "step": 2376
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2747917618007963e-06,
+      "loss": 0.5301,
+      "step": 2377
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.267980776804625e-06,
+      "loss": 0.56,
+      "step": 2378
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2611786991311025e-06,
+      "loss": 0.5452,
+      "step": 2379
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2543855366162516e-06,
+      "loss": 0.5715,
+      "step": 2380
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2476012970858275e-06,
+      "loss": 0.5783,
+      "step": 2381
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.240825988355311e-06,
+      "loss": 0.5457,
+      "step": 2382
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2340596182298824e-06,
+      "loss": 0.5914,
+      "step": 2383
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.227302194504437e-06,
+      "loss": 0.6071,
+      "step": 2384
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.22055372496356e-06,
+      "loss": 0.5514,
+      "step": 2385
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2138142173815156e-06,
+      "loss": 0.5702,
+      "step": 2386
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2070836795222506e-06,
+      "loss": 0.6662,
+      "step": 2387
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2003621191393797e-06,
+      "loss": 0.5779,
+      "step": 2388
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.193649543976165e-06,
+      "loss": 0.5668,
+      "step": 2389
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1869459617655297e-06,
+      "loss": 0.604,
+      "step": 2390
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.180251380230034e-06,
+      "loss": 0.62,
+      "step": 2391
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.173565807081861e-06,
+      "loss": 0.6828,
+      "step": 2392
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.166889250022827e-06,
+      "loss": 0.6088,
+      "step": 2393
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1602217167443582e-06,
+      "loss": 0.5785,
+      "step": 2394
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.153563214927481e-06,
+      "loss": 0.6232,
+      "step": 2395
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.146913752242822e-06,
+      "loss": 0.623,
+      "step": 2396
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1402733363505966e-06,
+      "loss": 0.5886,
+      "step": 2397
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1336419749005898e-06,
+      "loss": 0.64,
+      "step": 2398
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.127019675532165e-06,
+      "loss": 0.5417,
+      "step": 2399
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1204064458742448e-06,
+      "loss": 0.5977,
+      "step": 2400
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.113802293545294e-06,
+      "loss": 0.6279,
+      "step": 2401
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1072072261533317e-06,
+      "loss": 0.602,
+      "step": 2402
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.100621251295909e-06,
+      "loss": 0.6353,
+      "step": 2403
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0940443765600936e-06,
+      "loss": 0.586,
+      "step": 2404
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0874766095224797e-06,
+      "loss": 0.7776,
+      "step": 2405
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0809179577491702e-06,
+      "loss": 0.5912,
+      "step": 2406
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0743684287957542e-06,
+      "loss": 0.5654,
+      "step": 2407
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.067828030207325e-06,
+      "loss": 0.5788,
+      "step": 2408
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.061296769518455e-06,
+      "loss": 0.5909,
+      "step": 2409
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.054774654253181e-06,
+      "loss": 0.6048,
+      "step": 2410
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0482616919250155e-06,
+      "loss": 0.5393,
+      "step": 2411
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0417578900369218e-06,
+      "loss": 0.5721,
+      "step": 2412
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0352632560813066e-06,
+      "loss": 0.5741,
+      "step": 2413
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0287777975400214e-06,
+      "loss": 0.6424,
+      "step": 2414
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.022301521884347e-06,
+      "loss": 0.5414,
+      "step": 2415
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0158344365749793e-06,
+      "loss": 0.5609,
+      "step": 2416
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.009376549062032e-06,
+      "loss": 0.5957,
+      "step": 2417
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.002927866785027e-06,
+      "loss": 0.5933,
+      "step": 2418
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.996488397172871e-06,
+      "loss": 0.616,
+      "step": 2419
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.990058147643864e-06,
+      "loss": 0.5763,
+      "step": 2420
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9836371256056885e-06,
+      "loss": 0.5971,
+      "step": 2421
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9772253384553854e-06,
+      "loss": 0.5955,
+      "step": 2422
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.970822793579368e-06,
+      "loss": 0.6381,
+      "step": 2423
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9644294983533998e-06,
+      "loss": 0.6217,
+      "step": 2424
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9580454601425826e-06,
+      "loss": 0.5823,
+      "step": 2425
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.951670686301359e-06,
+      "loss": 0.5655,
+      "step": 2426
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.945305184173505e-06,
+      "loss": 0.5901,
+      "step": 2427
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9389489610921007e-06,
+      "loss": 0.5908,
+      "step": 2428
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9326020243795483e-06,
+      "loss": 0.5976,
+      "step": 2429
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9262643813475524e-06,
+      "loss": 0.5908,
+      "step": 2430
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9199360392971023e-06,
+      "loss": 0.6632,
+      "step": 2431
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.91361700551848e-06,
+      "loss": 0.6335,
+      "step": 2432
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9073072872912447e-06,
+      "loss": 0.6072,
+      "step": 2433
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9010068918842162e-06,
+      "loss": 0.5628,
+      "step": 2434
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8947158265554843e-06,
+      "loss": 0.6273,
+      "step": 2435
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8884340985523864e-06,
+      "loss": 0.5383,
+      "step": 2436
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.882161715111498e-06,
+      "loss": 0.6277,
+      "step": 2437
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.875898683458638e-06,
+      "loss": 0.5843,
+      "step": 2438
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8696450108088505e-06,
+      "loss": 0.5414,
+      "step": 2439
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8634007043663905e-06,
+      "loss": 0.597,
+      "step": 2440
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8571657713247304e-06,
+      "loss": 0.5881,
+      "step": 2441
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.850940218866547e-06,
+      "loss": 0.5621,
+      "step": 2442
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.844724054163699e-06,
+      "loss": 0.5523,
+      "step": 2443
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8385172843772414e-06,
+      "loss": 0.6614,
+      "step": 2444
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8323199166574046e-06,
+      "loss": 0.681,
+      "step": 2445
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.82613195814358e-06,
+      "loss": 0.6033,
+      "step": 2446
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8199534159643274e-06,
+      "loss": 0.5602,
+      "step": 2447
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8137842972373598e-06,
+      "loss": 0.5744,
+      "step": 2448
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.807624609069526e-06,
+      "loss": 0.5629,
+      "step": 2449
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8014743585568172e-06,
+      "loss": 0.558,
+      "step": 2450
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7953335527843552e-06,
+      "loss": 0.5745,
+      "step": 2451
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7892021988263707e-06,
+      "loss": 0.5703,
+      "step": 2452
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7830803037462152e-06,
+      "loss": 0.5097,
+      "step": 2453
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7769678745963449e-06,
+      "loss": 0.5813,
+      "step": 2454
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7708649184183003e-06,
+      "loss": 0.5663,
+      "step": 2455
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7647714422427188e-06,
+      "loss": 0.583,
+      "step": 2456
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.758687453089315e-06,
+      "loss": 0.678,
+      "step": 2457
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7526129579668704e-06,
+      "loss": 0.5702,
+      "step": 2458
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.746547963873233e-06,
+      "loss": 0.5798,
+      "step": 2459
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.740492477795308e-06,
+      "loss": 0.6601,
+      "step": 2460
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7344465067090388e-06,
+      "loss": 0.5923,
+      "step": 2461
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7284100575794137e-06,
+      "loss": 0.5304,
+      "step": 2462
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.722383137360455e-06,
+      "loss": 0.5994,
+      "step": 2463
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.716365752995197e-06,
+      "loss": 0.6148,
+      "step": 2464
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7103579114156976e-06,
+      "loss": 0.6096,
+      "step": 2465
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7043596195430223e-06,
+      "loss": 0.6487,
+      "step": 2466
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6983708842872248e-06,
+      "loss": 0.5906,
+      "step": 2467
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.692391712547361e-06,
+      "loss": 0.5739,
+      "step": 2468
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6864221112114653e-06,
+      "loss": 0.6284,
+      "step": 2469
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6804620871565436e-06,
+      "loss": 0.538,
+      "step": 2470
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6745116472485722e-06,
+      "loss": 0.5796,
+      "step": 2471
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6685707983424903e-06,
+      "loss": 0.5131,
+      "step": 2472
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6626395472821778e-06,
+      "loss": 0.5662,
+      "step": 2473
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6567179009004675e-06,
+      "loss": 0.6131,
+      "step": 2474
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.650805866019125e-06,
+      "loss": 0.5685,
+      "step": 2475
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6449034494488381e-06,
+      "loss": 0.6029,
+      "step": 2476
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6390106579892217e-06,
+      "loss": 0.6014,
+      "step": 2477
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6331274984288004e-06,
+      "loss": 0.5545,
+      "step": 2478
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.627253977544998e-06,
+      "loss": 0.5878,
+      "step": 2479
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6213901021041412e-06,
+      "loss": 0.681,
+      "step": 2480
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6155358788614428e-06,
+      "loss": 0.5884,
+      "step": 2481
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6096913145609927e-06,
+      "loss": 0.5955,
+      "step": 2482
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6038564159357584e-06,
+      "loss": 0.6151,
+      "step": 2483
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5980311897075717e-06,
+      "loss": 0.5726,
+      "step": 2484
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5922156425871183e-06,
+      "loss": 0.5937,
+      "step": 2485
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5864097812739355e-06,
+      "loss": 0.5999,
+      "step": 2486
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5806136124564076e-06,
+      "loss": 0.5821,
+      "step": 2487
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5748271428117434e-06,
+      "loss": 0.5266,
+      "step": 2488
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5690503790059852e-06,
+      "loss": 0.6567,
+      "step": 2489
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5632833276939941e-06,
+      "loss": 0.5509,
+      "step": 2490
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5575259955194354e-06,
+      "loss": 0.5456,
+      "step": 2491
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5517783891147875e-06,
+      "loss": 0.638,
+      "step": 2492
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5460405151013202e-06,
+      "loss": 0.5828,
+      "step": 2493
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5403123800890874e-06,
+      "loss": 0.5041,
+      "step": 2494
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5345939906769302e-06,
+      "loss": 0.5702,
+      "step": 2495
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5288853534524628e-06,
+      "loss": 0.592,
+      "step": 2496
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5231864749920578e-06,
+      "loss": 0.5556,
+      "step": 2497
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5174973618608514e-06,
+      "loss": 0.7133,
+      "step": 2498
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.511818020612731e-06,
+      "loss": 0.4995,
+      "step": 2499
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5061484577903217e-06,
+      "loss": 0.6014,
+      "step": 2500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5004886799249862e-06,
+      "loss": 0.5298,
+      "step": 2501
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4948386935368208e-06,
+      "loss": 0.587,
+      "step": 2502
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4891985051346292e-06,
+      "loss": 0.6348,
+      "step": 2503
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4835681212159391e-06,
+      "loss": 0.5555,
+      "step": 2504
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4779475482669803e-06,
+      "loss": 0.5724,
+      "step": 2505
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4723367927626785e-06,
+      "loss": 0.6388,
+      "step": 2506
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4667358611666472e-06,
+      "loss": 0.5842,
+      "step": 2507
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.461144759931188e-06,
+      "loss": 0.5809,
+      "step": 2508
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4555634954972796e-06,
+      "loss": 0.587,
+      "step": 2509
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4499920742945594e-06,
+      "loss": 0.6073,
+      "step": 2510
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.444430502741333e-06,
+      "loss": 0.5541,
+      "step": 2511
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4388787872445597e-06,
+      "loss": 0.5992,
+      "step": 2512
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4333369341998382e-06,
+      "loss": 0.6106,
+      "step": 2513
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4278049499914116e-06,
+      "loss": 0.5716,
+      "step": 2514
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4222828409921519e-06,
+      "loss": 0.5989,
+      "step": 2515
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4167706135635528e-06,
+      "loss": 0.6204,
+      "step": 2516
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.411268274055727e-06,
+      "loss": 0.6017,
+      "step": 2517
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4057758288073985e-06,
+      "loss": 0.742,
+      "step": 2518
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4002932841458861e-06,
+      "loss": 0.6157,
+      "step": 2519
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3948206463871084e-06,
+      "loss": 0.5347,
+      "step": 2520
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3893579218355723e-06,
+      "loss": 0.5743,
+      "step": 2521
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3839051167843588e-06,
+      "loss": 0.5289,
+      "step": 2522
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3784622375151258e-06,
+      "loss": 0.5419,
+      "step": 2523
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.373029290298099e-06,
+      "loss": 0.5371,
+      "step": 2524
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3676062813920554e-06,
+      "loss": 0.5791,
+      "step": 2525
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3621932170443275e-06,
+      "loss": 0.5933,
+      "step": 2526
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3567901034907949e-06,
+      "loss": 0.5342,
+      "step": 2527
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.351396946955864e-06,
+      "loss": 0.5948,
+      "step": 2528
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3460137536524798e-06,
+      "loss": 0.602,
+      "step": 2529
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.340640529782109e-06,
+      "loss": 0.5889,
+      "step": 2530
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3352772815347247e-06,
+      "loss": 0.6288,
+      "step": 2531
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3299240150888181e-06,
+      "loss": 0.558,
+      "step": 2532
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3245807366113784e-06,
+      "loss": 0.5879,
+      "step": 2533
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3192474522578847e-06,
+      "loss": 0.6138,
+      "step": 2534
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3139241681723058e-06,
+      "loss": 0.5879,
+      "step": 2535
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3086108904870954e-06,
+      "loss": 0.526,
+      "step": 2536
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.303307625323167e-06,
+      "loss": 0.526,
+      "step": 2537
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2980143787899103e-06,
+      "loss": 0.5067,
+      "step": 2538
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.292731156985173e-06,
+      "loss": 0.7123,
+      "step": 2539
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2874579659952468e-06,
+      "loss": 0.6428,
+      "step": 2540
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2821948118948758e-06,
+      "loss": 0.5828,
+      "step": 2541
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2769417007472384e-06,
+      "loss": 0.6039,
+      "step": 2542
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.271698638603941e-06,
+      "loss": 0.5885,
+      "step": 2543
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2664656315050184e-06,
+      "loss": 0.6529,
+      "step": 2544
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2612426854789195e-06,
+      "loss": 0.5741,
+      "step": 2545
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2560298065425003e-06,
+      "loss": 0.5861,
+      "step": 2546
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2508270007010225e-06,
+      "loss": 0.5688,
+      "step": 2547
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2456342739481463e-06,
+      "loss": 0.5576,
+      "step": 2548
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2404516322659145e-06,
+      "loss": 0.5876,
+      "step": 2549
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.235279081624754e-06,
+      "loss": 0.6257,
+      "step": 2550
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2301166279834708e-06,
+      "loss": 0.5896,
+      "step": 2551
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2249642772892324e-06,
+      "loss": 0.5827,
+      "step": 2552
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2198220354775726e-06,
+      "loss": 0.582,
+      "step": 2553
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2146899084723808e-06,
+      "loss": 0.6351,
+      "step": 2554
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2095679021858876e-06,
+      "loss": 0.5921,
+      "step": 2555
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2044560225186686e-06,
+      "loss": 0.5875,
+      "step": 2556
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1993542753596376e-06,
+      "loss": 0.6094,
+      "step": 2557
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1942626665860257e-06,
+      "loss": 0.5328,
+      "step": 2558
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.189181202063393e-06,
+      "loss": 0.554,
+      "step": 2559
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1841098876456136e-06,
+      "loss": 0.6101,
+      "step": 2560
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.17904872917486e-06,
+      "loss": 0.5513,
+      "step": 2561
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1739977324816131e-06,
+      "loss": 0.5859,
+      "step": 2562
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1689569033846482e-06,
+      "loss": 0.5756,
+      "step": 2563
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1639262476910184e-06,
+      "loss": 0.6563,
+      "step": 2564
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.158905771196065e-06,
+      "loss": 0.5555,
+      "step": 2565
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1538954796834045e-06,
+      "loss": 0.5874,
+      "step": 2566
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1488953789249103e-06,
+      "loss": 0.5437,
+      "step": 2567
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1439054746807254e-06,
+      "loss": 0.6118,
+      "step": 2568
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1389257726992442e-06,
+      "loss": 0.5805,
+      "step": 2569
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1339562787171022e-06,
+      "loss": 0.5786,
+      "step": 2570
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1289969984591842e-06,
+      "loss": 0.5987,
+      "step": 2571
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.124047937638605e-06,
+      "loss": 0.5749,
+      "step": 2572
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1191091019567024e-06,
+      "loss": 0.5795,
+      "step": 2573
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.114180497103039e-06,
+      "loss": 0.5676,
+      "step": 2574
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1092621287553952e-06,
+      "loss": 0.5909,
+      "step": 2575
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1043540025797496e-06,
+      "loss": 0.5636,
+      "step": 2576
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.099456124230288e-06,
+      "loss": 0.5942,
+      "step": 2577
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.094568499349392e-06,
+      "loss": 0.5998,
+      "step": 2578
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0896911335676252e-06,
+      "loss": 0.6575,
+      "step": 2579
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0848240325037374e-06,
+      "loss": 0.585,
+      "step": 2580
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0799672017646524e-06,
+      "loss": 0.5636,
+      "step": 2581
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0751206469454612e-06,
+      "loss": 0.5682,
+      "step": 2582
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0702843736294178e-06,
+      "loss": 0.5972,
+      "step": 2583
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0654583873879342e-06,
+      "loss": 0.5857,
+      "step": 2584
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0606426937805659e-06,
+      "loss": 0.5598,
+      "step": 2585
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0558372983550158e-06,
+      "loss": 0.6049,
+      "step": 2586
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0510422066471249e-06,
+      "loss": 0.5682,
+      "step": 2587
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0462574241808565e-06,
+      "loss": 0.5905,
+      "step": 2588
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.041482956468306e-06,
+      "loss": 0.5631,
+      "step": 2589
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0367188090096836e-06,
+      "loss": 0.565,
+      "step": 2590
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.031964987293307e-06,
+      "loss": 0.6395,
+      "step": 2591
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0272214967956006e-06,
+      "loss": 0.5406,
+      "step": 2592
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0224883429810929e-06,
+      "loss": 0.6172,
+      "step": 2593
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0177655313023938e-06,
+      "loss": 0.5652,
+      "step": 2594
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.013053067200207e-06,
+      "loss": 0.5978,
+      "step": 2595
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0083509561033155e-06,
+      "loss": 0.6727,
+      "step": 2596
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0036592034285696e-06,
+      "loss": 0.5703,
+      "step": 2597
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.989778145808915e-07,
+      "loss": 0.6357,
+      "step": 2598
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.94306794953267e-07,
+      "loss": 0.5642,
+      "step": 2599
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.896461499267273e-07,
+      "loss": 0.6335,
+      "step": 2600
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.8499588487036e-07,
+      "loss": 0.6374,
+      "step": 2601
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.803560051412941e-07,
+      "loss": 0.5708,
+      "step": 2602
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.757265160846896e-07,
+      "loss": 0.5858,
+      "step": 2603
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.711074230337414e-07,
+      "loss": 0.555,
+      "step": 2604
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.664987313096673e-07,
+      "loss": 0.5323,
+      "step": 2605
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.619004462217007e-07,
+      "loss": 0.6478,
+      "step": 2606
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.573125730670874e-07,
+      "loss": 0.5711,
+      "step": 2607
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.527351171310816e-07,
+      "loss": 0.5899,
+      "step": 2608
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.4816808368693e-07,
+      "loss": 0.5791,
+      "step": 2609
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.436114779958794e-07,
+      "loss": 0.6104,
+      "step": 2610
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.39065305307163e-07,
+      "loss": 0.5679,
+      "step": 2611
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.345295708579883e-07,
+      "loss": 0.5796,
+      "step": 2612
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.300042798735443e-07,
+      "loss": 0.5732,
+      "step": 2613
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.254894375669921e-07,
+      "loss": 0.6528,
+      "step": 2614
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.209850491394445e-07,
+      "loss": 0.6144,
+      "step": 2615
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.164911197799819e-07,
+      "loss": 0.5771,
+      "step": 2616
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.120076546656342e-07,
+      "loss": 0.6296,
+      "step": 2617
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.075346589613698e-07,
+      "loss": 0.5716,
+      "step": 2618
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.030721378201024e-07,
+      "loss": 0.5942,
+      "step": 2619
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.986200963826808e-07,
+      "loss": 0.5778,
+      "step": 2620
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.941785397778724e-07,
+      "loss": 0.5835,
+      "step": 2621
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.897474731223743e-07,
+      "loss": 0.5573,
+      "step": 2622
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.853269015207977e-07,
+      "loss": 0.614,
+      "step": 2623
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.809168300656578e-07,
+      "loss": 0.5633,
+      "step": 2624
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.765172638373809e-07,
+      "loss": 0.6002,
+      "step": 2625
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.721282079042881e-07,
+      "loss": 0.5706,
+      "step": 2626
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.677496673225905e-07,
+      "loss": 0.5824,
+      "step": 2627
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.633816471363887e-07,
+      "loss": 0.5293,
+      "step": 2628
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.590241523776644e-07,
+      "loss": 0.5657,
+      "step": 2629
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.546771880662686e-07,
+      "loss": 0.5143,
+      "step": 2630
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.503407592099256e-07,
+      "loss": 0.5722,
+      "step": 2631
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.460148708042248e-07,
+      "loss": 0.5622,
+      "step": 2632
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.416995278326068e-07,
+      "loss": 0.5729,
+      "step": 2633
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.37394735266368e-07,
+      "loss": 0.6565,
+      "step": 2634
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.331004980646517e-07,
+      "loss": 0.553,
+      "step": 2635
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.288168211744363e-07,
+      "loss": 0.5879,
+      "step": 2636
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.245437095305386e-07,
+      "loss": 0.5906,
+      "step": 2637
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.202811680556066e-07,
+      "loss": 0.5474,
+      "step": 2638
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.160292016601034e-07,
+      "loss": 0.6217,
+      "step": 2639
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.117878152423164e-07,
+      "loss": 0.5945,
+      "step": 2640
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.075570136883448e-07,
+      "loss": 0.6471,
+      "step": 2641
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.033368018720888e-07,
+      "loss": 0.7155,
+      "step": 2642
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.991271846552529e-07,
+      "loss": 0.5948,
+      "step": 2643
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.949281668873399e-07,
+      "loss": 0.5944,
+      "step": 2644
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.907397534056316e-07,
+      "loss": 0.5689,
+      "step": 2645
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.86561949035205e-07,
+      "loss": 0.551,
+      "step": 2646
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.823947585889102e-07,
+      "loss": 0.5783,
+      "step": 2647
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.78238186867366e-07,
+      "loss": 0.5324,
+      "step": 2648
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.740922386589677e-07,
+      "loss": 0.5551,
+      "step": 2649
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.699569187398658e-07,
+      "loss": 0.5665,
+      "step": 2650
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.658322318739653e-07,
+      "loss": 0.5971,
+      "step": 2651
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.617181828129283e-07,
+      "loss": 0.5163,
+      "step": 2652
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.576147762961617e-07,
+      "loss": 0.5529,
+      "step": 2653
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.535220170508006e-07,
+      "loss": 0.583,
+      "step": 2654
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.494399097917282e-07,
+      "loss": 0.6128,
+      "step": 2655
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.453684592215526e-07,
+      "loss": 0.5878,
+      "step": 2656
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.41307670030601e-07,
+      "loss": 0.5752,
+      "step": 2657
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.372575468969223e-07,
+      "loss": 0.5499,
+      "step": 2658
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.33218094486281e-07,
+      "loss": 0.5689,
+      "step": 2659
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.291893174521391e-07,
+      "loss": 0.601,
+      "step": 2660
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.251712204356709e-07,
+      "loss": 0.646,
+      "step": 2661
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.211638080657446e-07,
+      "loss": 0.5447,
+      "step": 2662
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.171670849589141e-07,
+      "loss": 0.5058,
+      "step": 2663
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.131810557194252e-07,
+      "loss": 0.5675,
+      "step": 2664
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.092057249392037e-07,
+      "loss": 0.7061,
+      "step": 2665
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.052410971978463e-07,
+      "loss": 0.6368,
+      "step": 2666
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.012871770626251e-07,
+      "loss": 0.5441,
+      "step": 2667
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.973439690884754e-07,
+      "loss": 0.5778,
+      "step": 2668
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.934114778179891e-07,
+      "loss": 0.6282,
+      "step": 2669
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.894897077814167e-07,
+      "loss": 0.5264,
+      "step": 2670
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.855786634966555e-07,
+      "loss": 0.5739,
+      "step": 2671
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.816783494692447e-07,
+      "loss": 0.6501,
+      "step": 2672
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.777887701923669e-07,
+      "loss": 0.5724,
+      "step": 2673
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.739099301468366e-07,
+      "loss": 0.5876,
+      "step": 2674
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.700418338010939e-07,
+      "loss": 0.5303,
+      "step": 2675
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.661844856112032e-07,
+      "loss": 0.5678,
+      "step": 2676
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.62337890020851e-07,
+      "loss": 0.5507,
+      "step": 2677
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.585020514613305e-07,
+      "loss": 0.5759,
+      "step": 2678
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.546769743515469e-07,
+      "loss": 0.553,
+      "step": 2679
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.508626630980086e-07,
+      "loss": 0.4815,
+      "step": 2680
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.470591220948175e-07,
+      "loss": 0.5605,
+      "step": 2681
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.43266355723673e-07,
+      "loss": 0.5602,
+      "step": 2682
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.394843683538599e-07,
+      "loss": 0.5532,
+      "step": 2683
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.357131643422431e-07,
+      "loss": 0.5758,
+      "step": 2684
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.319527480332688e-07,
+      "loss": 0.5798,
+      "step": 2685
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.282031237589559e-07,
+      "loss": 0.5859,
+      "step": 2686
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.244642958388869e-07,
+      "loss": 0.5504,
+      "step": 2687
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.207362685802098e-07,
+      "loss": 0.5249,
+      "step": 2688
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.170190462776315e-07,
+      "loss": 0.5941,
+      "step": 2689
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.133126332134065e-07,
+      "loss": 0.5951,
+      "step": 2690
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.096170336573415e-07,
+      "loss": 0.5558,
+      "step": 2691
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.059322518667865e-07,
+      "loss": 0.5663,
+      "step": 2692
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.022582920866249e-07,
+      "loss": 0.6082,
+      "step": 2693
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.985951585492766e-07,
+      "loss": 0.5316,
+      "step": 2694
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.949428554746905e-07,
+      "loss": 0.5968,
+      "step": 2695
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.913013870703333e-07,
+      "loss": 0.5432,
+      "step": 2696
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.876707575311968e-07,
+      "loss": 0.5252,
+      "step": 2697
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.840509710397824e-07,
+      "loss": 0.646,
+      "step": 2698
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.804420317661008e-07,
+      "loss": 0.5938,
+      "step": 2699
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.768439438676676e-07,
+      "loss": 0.5143,
+      "step": 2700
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.732567114894982e-07,
+      "loss": 0.5797,
+      "step": 2701
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.696803387640981e-07,
+      "loss": 0.6206,
+      "step": 2702
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.661148298114682e-07,
+      "loss": 0.6552,
+      "step": 2703
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.625601887390919e-07,
+      "loss": 0.5838,
+      "step": 2704
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.590164196419301e-07,
+      "loss": 0.5574,
+      "step": 2705
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.554835266024227e-07,
+      "loss": 0.593,
+      "step": 2706
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.51961513690481e-07,
+      "loss": 0.5627,
+      "step": 2707
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.48450384963477e-07,
+      "loss": 0.5861,
+      "step": 2708
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.449501444662497e-07,
+      "loss": 0.5909,
+      "step": 2709
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.414607962310947e-07,
+      "loss": 0.5696,
+      "step": 2710
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.379823442777532e-07,
+      "loss": 0.5375,
+      "step": 2711
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.345147926134198e-07,
+      "loss": 0.5795,
+      "step": 2712
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.310581452327335e-07,
+      "loss": 0.6041,
+      "step": 2713
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.276124061177645e-07,
+      "loss": 0.6112,
+      "step": 2714
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.24177579238021e-07,
+      "loss": 0.5143,
+      "step": 2715
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.207536685504433e-07,
+      "loss": 0.6101,
+      "step": 2716
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.173406779993884e-07,
+      "loss": 0.6021,
+      "step": 2717
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.139386115166389e-07,
+      "loss": 0.6067,
+      "step": 2718
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.105474730213933e-07,
+      "loss": 0.5582,
+      "step": 2719
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.071672664202565e-07,
+      "loss": 0.5862,
+      "step": 2720
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.037979956072436e-07,
+      "loss": 0.5298,
+      "step": 2721
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.00439664463771e-07,
+      "loss": 0.6011,
+      "step": 2722
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.970922768586517e-07,
+      "loss": 0.6327,
+      "step": 2723
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.937558366480899e-07,
+      "loss": 0.5695,
+      "step": 2724
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.904303476756856e-07,
+      "loss": 0.5548,
+      "step": 2725
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.871158137724119e-07,
+      "loss": 0.5582,
+      "step": 2726
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.838122387566313e-07,
+      "loss": 0.6058,
+      "step": 2727
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.805196264340772e-07,
+      "loss": 0.5574,
+      "step": 2728
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.772379805978533e-07,
+      "loss": 0.648,
+      "step": 2729
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.739673050284321e-07,
+      "loss": 0.569,
+      "step": 2730
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.707076034936475e-07,
+      "loss": 0.5655,
+      "step": 2731
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.674588797486901e-07,
+      "loss": 0.5926,
+      "step": 2732
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.64221137536105e-07,
+      "loss": 0.6075,
+      "step": 2733
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.609943805857886e-07,
+      "loss": 0.5765,
+      "step": 2734
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.577786126149775e-07,
+      "loss": 0.5926,
+      "step": 2735
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5457383732825266e-07,
+      "loss": 0.5704,
+      "step": 2736
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.51380058417531e-07,
+      "loss": 0.5698,
+      "step": 2737
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.481972795620593e-07,
+      "loss": 0.5556,
+      "step": 2738
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4502550442841466e-07,
+      "loss": 0.6339,
+      "step": 2739
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4186473667049867e-07,
+      "loss": 0.5719,
+      "step": 2740
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.387149799295276e-07,
+      "loss": 0.5852,
+      "step": 2741
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.355762378340378e-07,
+      "loss": 0.5741,
+      "step": 2742
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3244851399987574e-07,
+      "loss": 0.5754,
+      "step": 2743
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.293318120301926e-07,
+      "loss": 0.5455,
+      "step": 2744
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.26226135515444e-07,
+      "loss": 0.549,
+      "step": 2745
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.231314880333848e-07,
+      "loss": 0.6013,
+      "step": 2746
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2004787314906204e-07,
+      "loss": 0.5636,
+      "step": 2747
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.169752944148153e-07,
+      "loss": 0.5878,
+      "step": 2748
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1391375537027214e-07,
+      "loss": 0.5441,
+      "step": 2749
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1086325954233564e-07,
+      "loss": 0.5633,
+      "step": 2750
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.078238104451948e-07,
+      "loss": 0.5405,
+      "step": 2751
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.047954115803088e-07,
+      "loss": 0.5624,
+      "step": 2752
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.017780664364046e-07,
+      "loss": 0.5687,
+      "step": 2753
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9877177848948177e-07,
+      "loss": 0.5869,
+      "step": 2754
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.957765512027967e-07,
+      "loss": 0.6486,
+      "step": 2755
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.927923880268647e-07,
+      "loss": 0.5935,
+      "step": 2756
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.898192923994559e-07,
+      "loss": 0.5903,
+      "step": 2757
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8685726774559286e-07,
+      "loss": 0.5805,
+      "step": 2758
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8390631747753926e-07,
+      "loss": 0.6259,
+      "step": 2759
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8096644499480364e-07,
+      "loss": 0.5694,
+      "step": 2760
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7803765368413567e-07,
+      "loss": 0.6006,
+      "step": 2761
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7511994691951435e-07,
+      "loss": 0.6379,
+      "step": 2762
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.72213328062152e-07,
+      "loss": 0.6313,
+      "step": 2763
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.69317800460488e-07,
+      "loss": 0.6057,
+      "step": 2764
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.664333674501841e-07,
+      "loss": 0.5493,
+      "step": 2765
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6356003235411887e-07,
+      "loss": 0.6186,
+      "step": 2766
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.606977984823901e-07,
+      "loss": 0.5996,
+      "step": 2767
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.578466691323035e-07,
+      "loss": 0.5967,
+      "step": 2768
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5500664758837287e-07,
+      "loss": 0.5389,
+      "step": 2769
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5217773712232005e-07,
+      "loss": 0.566,
+      "step": 2770
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4935994099305814e-07,
+      "loss": 0.5229,
+      "step": 2771
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4655326244670386e-07,
+      "loss": 0.5987,
+      "step": 2772
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4375770471656635e-07,
+      "loss": 0.5722,
+      "step": 2773
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4097327102313726e-07,
+      "loss": 0.6078,
+      "step": 2774
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3819996457410075e-07,
+      "loss": 0.5966,
+      "step": 2775
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3543778856431676e-07,
+      "loss": 0.6583,
+      "step": 2776
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.326867461758265e-07,
+      "loss": 0.6083,
+      "step": 2777
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.299468405778417e-07,
+      "loss": 0.5408,
+      "step": 2778
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.272180749267506e-07,
+      "loss": 0.6207,
+      "step": 2779
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2450045236609995e-07,
+      "loss": 0.5486,
+      "step": 2780
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2179397602660424e-07,
+      "loss": 0.6194,
+      "step": 2781
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.190986490261405e-07,
+      "loss": 0.6163,
+      "step": 2782
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1641447446973397e-07,
+      "loss": 0.5874,
+      "step": 2783
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.137414554495688e-07,
+      "loss": 0.5324,
+      "step": 2784
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.110795950449752e-07,
+      "loss": 0.6003,
+      "step": 2785
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.08428896322428e-07,
+      "loss": 0.5283,
+      "step": 2786
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.057893623355457e-07,
+      "loss": 0.5561,
+      "step": 2787
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.031609961250848e-07,
+      "loss": 0.5939,
+      "step": 2788
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0054380071893206e-07,
+      "loss": 0.5094,
+      "step": 2789
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9793777913211143e-07,
+      "loss": 0.5989,
+      "step": 2790
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.953429343667735e-07,
+      "loss": 0.5409,
+      "step": 2791
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9275926941218834e-07,
+      "loss": 0.6621,
+      "step": 2792
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.901867872447528e-07,
+      "loss": 0.6177,
+      "step": 2793
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.876254908279774e-07,
+      "loss": 0.5806,
+      "step": 2794
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.850753831124875e-07,
+      "loss": 0.6025,
+      "step": 2795
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.825364670360198e-07,
+      "loss": 0.5588,
+      "step": 2796
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8000874552341795e-07,
+      "loss": 0.6113,
+      "step": 2797
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7749222148662714e-07,
+      "loss": 0.5528,
+      "step": 2798
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7498689782469724e-07,
+      "loss": 0.5996,
+      "step": 2799
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.724927774237718e-07,
+      "loss": 0.6174,
+      "step": 2800
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.700098631570891e-07,
+      "loss": 0.6275,
+      "step": 2801
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.675381578849789e-07,
+      "loss": 0.5018,
+      "step": 2802
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6507766445485583e-07,
+      "loss": 0.5977,
+      "step": 2803
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.626283857012235e-07,
+      "loss": 0.5554,
+      "step": 2804
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.601903244456572e-07,
+      "loss": 0.5915,
+      "step": 2805
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.57763483496819e-07,
+      "loss": 0.5966,
+      "step": 2806
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.553478656504416e-07,
+      "loss": 0.5721,
+      "step": 2807
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5294347368932435e-07,
+      "loss": 0.6041,
+      "step": 2808
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5055031038334174e-07,
+      "loss": 0.5347,
+      "step": 2809
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.481683784894273e-07,
+      "loss": 0.5675,
+      "step": 2810
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.457976807515783e-07,
+      "loss": 0.5789,
+      "step": 2811
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4343821990085005e-07,
+      "loss": 0.5776,
+      "step": 2812
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4108999865535277e-07,
+      "loss": 0.5373,
+      "step": 2813
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3875301972024703e-07,
+      "loss": 0.6703,
+      "step": 2814
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3642728578774478e-07,
+      "loss": 0.5582,
+      "step": 2815
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3411279953710397e-07,
+      "loss": 0.5863,
+      "step": 2816
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.318095636346196e-07,
+      "loss": 0.5879,
+      "step": 2817
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2951758073363362e-07,
+      "loss": 0.5848,
+      "step": 2818
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2723685347452062e-07,
+      "loss": 0.6114,
+      "step": 2819
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.249673844846878e-07,
+      "loss": 0.596,
+      "step": 2820
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2270917637857602e-07,
+      "loss": 0.5812,
+      "step": 2821
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2046223175765104e-07,
+      "loss": 0.5505,
+      "step": 2822
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.182265532104022e-07,
+      "loss": 0.6423,
+      "step": 2823
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1600214331234382e-07,
+      "loss": 0.5407,
+      "step": 2824
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1378900462600605e-07,
+      "loss": 0.5663,
+      "step": 2825
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1158713970093393e-07,
+      "loss": 0.5405,
+      "step": 2826
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0939655107368728e-07,
+      "loss": 0.5943,
+      "step": 2827
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0721724126783526e-07,
+      "loss": 0.5584,
+      "step": 2828
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0504921279395074e-07,
+      "loss": 0.5329,
+      "step": 2829
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0289246814961473e-07,
+      "loss": 0.5603,
+      "step": 2830
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0074700981940755e-07,
+      "loss": 0.6299,
+      "step": 2831
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9861284027490545e-07,
+      "loss": 0.5077,
+      "step": 2832
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9648996197468184e-07,
+      "loss": 0.6439,
+      "step": 2833
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9437837736430597e-07,
+      "loss": 0.6202,
+      "step": 2834
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.922780888763298e-07,
+      "loss": 0.5714,
+      "step": 2835
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.901890989302957e-07,
+      "loss": 0.569,
+      "step": 2836
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8811140993273191e-07,
+      "loss": 0.5648,
+      "step": 2837
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.86045024277145e-07,
+      "loss": 0.6359,
+      "step": 2838
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8398994434402074e-07,
+      "loss": 0.5732,
+      "step": 2839
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.819461725008209e-07,
+      "loss": 0.592,
+      "step": 2840
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.799137111019811e-07,
+      "loss": 0.6213,
+      "step": 2841
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7789256248890497e-07,
+      "loss": 0.5233,
+      "step": 2842
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7588272898996672e-07,
+      "loss": 0.584,
+      "step": 2843
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7388421292050318e-07,
+      "loss": 0.6525,
+      "step": 2844
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.718970165828149e-07,
+      "loss": 0.581,
+      "step": 2845
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6992114226616175e-07,
+      "loss": 0.5508,
+      "step": 2846
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6795659224675854e-07,
+      "loss": 0.5295,
+      "step": 2847
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.660033687877771e-07,
+      "loss": 0.5645,
+      "step": 2848
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6406147413934092e-07,
+      "loss": 0.6029,
+      "step": 2849
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6213091053852047e-07,
+      "loss": 0.5875,
+      "step": 2850
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6021168020933565e-07,
+      "loss": 0.5412,
+      "step": 2851
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5830378536274782e-07,
+      "loss": 0.564,
+      "step": 2852
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.564072281966611e-07,
+      "loss": 0.6485,
+      "step": 2853
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5452201089591667e-07,
+      "loss": 0.5844,
+      "step": 2854
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.526481356322973e-07,
+      "loss": 0.5828,
+      "step": 2855
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5078560456451395e-07,
+      "loss": 0.6548,
+      "step": 2856
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4893441983821032e-07,
+      "loss": 0.6135,
+      "step": 2857
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4709458358596162e-07,
+      "loss": 0.5537,
+      "step": 2858
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.45266097927268e-07,
+      "loss": 0.5887,
+      "step": 2859
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4344896496855222e-07,
+      "loss": 0.5391,
+      "step": 2860
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.416431868031609e-07,
+      "loss": 0.577,
+      "step": 2861
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.3984876551135783e-07,
+      "loss": 0.6568,
+      "step": 2862
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.3806570316032719e-07,
+      "loss": 0.5913,
+      "step": 2863
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3629400180416362e-07,
+      "loss": 0.5552,
+      "step": 2864
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3453366348387453e-07,
+      "loss": 0.533,
+      "step": 2865
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3278469022737995e-07,
+      "loss": 0.5014,
+      "step": 2866
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3104708404950482e-07,
+      "loss": 0.5714,
+      "step": 2867
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.293208469519791e-07,
+      "loss": 0.5425,
+      "step": 2868
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2760598092343534e-07,
+      "loss": 0.646,
+      "step": 2869
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2590248793940997e-07,
+      "loss": 0.5651,
+      "step": 2870
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2421036996233204e-07,
+      "loss": 0.5585,
+      "step": 2871
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.225296289415301e-07,
+      "loss": 0.5773,
+      "step": 2872
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.208602668132275e-07,
+      "loss": 0.5593,
+      "step": 2873
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1920228550053592e-07,
+      "loss": 0.5805,
+      "step": 2874
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1755568691345753e-07,
+      "loss": 0.5507,
+      "step": 2875
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1592047294888164e-07,
+      "loss": 0.6033,
+      "step": 2876
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1429664549058362e-07,
+      "loss": 0.6248,
+      "step": 2877
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1268420640921817e-07,
+      "loss": 0.5678,
+      "step": 2878
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1108315756232613e-07,
+      "loss": 0.5471,
+      "step": 2879
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0949350079431986e-07,
+      "loss": 0.5817,
+      "step": 2880
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0791523793649228e-07,
+      "loss": 0.5776,
+      "step": 2881
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0634837080701121e-07,
+      "loss": 0.6013,
+      "step": 2882
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.047929012109139e-07,
+      "loss": 0.5635,
+      "step": 2883
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0324883094010918e-07,
+      "loss": 0.5881,
+      "step": 2884
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0171616177337196e-07,
+      "loss": 0.5875,
+      "step": 2885
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0019489547634654e-07,
+      "loss": 0.6773,
+      "step": 2886
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.86850338015366e-08,
+      "loss": 0.5607,
+      "step": 2887
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.718657848831193e-08,
+      "loss": 0.5867,
+      "step": 2888
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.569953126289832e-08,
+      "loss": 0.5517,
+      "step": 2889
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.422389383838215e-08,
+      "loss": 0.5356,
+      "step": 2890
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.27596679147047e-08,
+      "loss": 0.5465,
+      "step": 2891
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.130685517866111e-08,
+      "loss": 0.5892,
+      "step": 2892
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.986545730389706e-08,
+      "loss": 0.6087,
+      "step": 2893
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.843547595091317e-08,
+      "loss": 0.59,
+      "step": 2894
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.70169127670517e-08,
+      "loss": 0.5974,
+      "step": 2895
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.560976938650434e-08,
+      "loss": 0.5832,
+      "step": 2896
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.42140474303088e-08,
+      "loss": 0.6481,
+      "step": 2897
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.282974850634229e-08,
+      "loss": 0.5953,
+      "step": 2898
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.145687420932247e-08,
+      "loss": 0.6647,
+      "step": 2899
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.009542612080867e-08,
+      "loss": 0.6133,
+      "step": 2900
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.874540580919521e-08,
+      "loss": 0.608,
+      "step": 2901
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.740681482971135e-08,
+      "loss": 0.6391,
+      "step": 2902
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.607965472442025e-08,
+      "loss": 0.5484,
+      "step": 2903
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.476392702221669e-08,
+      "loss": 0.6137,
+      "step": 2904
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.345963323882598e-08,
+      "loss": 0.5789,
+      "step": 2905
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.216677487680068e-08,
+      "loss": 0.5437,
+      "step": 2906
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.088535342551828e-08,
+      "loss": 0.5759,
+      "step": 2907
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.961537036118238e-08,
+      "loss": 0.6203,
+      "step": 2908
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.835682714682157e-08,
+      "loss": 0.5394,
+      "step": 2909
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.710972523228165e-08,
+      "loss": 0.6253,
+      "step": 2910
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.587406605423119e-08,
+      "loss": 0.5312,
+      "step": 2911
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.464985103615485e-08,
+      "loss": 0.5996,
+      "step": 2912
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.343708158835338e-08,
+      "loss": 0.5925,
+      "step": 2913
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.223575910794366e-08,
+      "loss": 0.5516,
+      "step": 2914
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.104588497885533e-08,
+      "loss": 0.582,
+      "step": 2915
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.986746057182746e-08,
+      "loss": 0.6126,
+      "step": 2916
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.870048724441191e-08,
+      "loss": 0.6531,
+      "step": 2917
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.754496634096773e-08,
+      "loss": 0.5335,
+      "step": 2918
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.640089919265901e-08,
+      "loss": 0.5634,
+      "step": 2919
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.526828711745702e-08,
+      "loss": 0.5799,
+      "step": 2920
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.4147131420136945e-08,
+      "loss": 0.6046,
+      "step": 2921
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.3037433392275625e-08,
+      "loss": 0.6881,
+      "step": 2922
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.1939194312248255e-08,
+      "loss": 0.5619,
+      "step": 2923
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.0852415445233896e-08,
+      "loss": 0.5735,
+      "step": 2924
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.977709804320663e-08,
+      "loss": 0.5385,
+      "step": 2925
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.871324334493555e-08,
+      "loss": 0.7488,
+      "step": 2926
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.766085257598696e-08,
+      "loss": 0.5321,
+      "step": 2927
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.6619926948721085e-08,
+      "loss": 0.5522,
+      "step": 2928
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.5590467662288695e-08,
+      "loss": 0.5808,
+      "step": 2929
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.457247590263114e-08,
+      "loss": 0.5521,
+      "step": 2930
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.356595284247922e-08,
+      "loss": 0.5853,
+      "step": 2931
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.2570899641354305e-08,
+      "loss": 0.5336,
+      "step": 2932
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.158731744556055e-08,
+      "loss": 0.5878,
+      "step": 2933
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.0615207388189356e-08,
+      "loss": 0.5797,
+      "step": 2934
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.965457058911715e-08,
+      "loss": 0.5558,
+      "step": 2935
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.870540815500201e-08,
+      "loss": 0.6358,
+      "step": 2936
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.7767721179281516e-08,
+      "loss": 0.5215,
+      "step": 2937
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.6841510742179343e-08,
+      "loss": 0.5519,
+      "step": 2938
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.592677791068977e-08,
+      "loss": 0.5708,
+      "step": 2939
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.502352373859319e-08,
+      "loss": 0.5591,
+      "step": 2940
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.4131749266440586e-08,
+      "loss": 0.6006,
+      "step": 2941
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.3251455521560174e-08,
+      "loss": 0.6466,
+      "step": 2942
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.23826435180552e-08,
+      "loss": 0.5463,
+      "step": 2943
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.152531425680172e-08,
+      "loss": 0.5601,
+      "step": 2944
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.067946872544636e-08,
+      "loss": 0.5621,
+      "step": 2945
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9845107898407446e-08,
+      "loss": 0.5952,
+      "step": 2946
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9022232736871658e-08,
+      "loss": 0.555,
+      "step": 2947
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.821084418879738e-08,
+      "loss": 0.5652,
+      "step": 2948
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7410943188905802e-08,
+      "loss": 0.5372,
+      "step": 2949
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6622530658686475e-08,
+      "loss": 0.5402,
+      "step": 2950
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5845607506396198e-08,
+      "loss": 0.6515,
+      "step": 2951
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5080174627052366e-08,
+      "loss": 0.5648,
+      "step": 2952
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.432623290243963e-08,
+      "loss": 0.6171,
+      "step": 2953
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3583783201099887e-08,
+      "loss": 0.6094,
+      "step": 2954
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2852826378338966e-08,
+      "loss": 0.6177,
+      "step": 2955
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2133363276224396e-08,
+      "loss": 0.5732,
+      "step": 2956
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1425394723579852e-08,
+      "loss": 0.5139,
+      "step": 2957
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.072892153598738e-08,
+      "loss": 0.605,
+      "step": 2958
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0043944515789615e-08,
+      "loss": 0.4814,
+      "step": 2959
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9370464452082018e-08,
+      "loss": 0.5527,
+      "step": 2960
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.87084821207173e-08,
+      "loss": 0.5454,
+      "step": 2961
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8057998284302104e-08,
+      "loss": 0.5814,
+      "step": 2962
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7419013692195896e-08,
+      "loss": 0.5687,
+      "step": 2963
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.679152908051429e-08,
+      "loss": 0.6237,
+      "step": 2964
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6175545172119056e-08,
+      "loss": 0.5648,
+      "step": 2965
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5571062676630334e-08,
+      "loss": 0.6599,
+      "step": 2966
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.49780822904122e-08,
+      "loss": 0.5538,
+      "step": 2967
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4396604696581552e-08,
+      "loss": 0.5675,
+      "step": 2968
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3826630565003663e-08,
+      "loss": 0.55,
+      "step": 2969
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3268160552291076e-08,
+      "loss": 0.5905,
+      "step": 2970
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2721195301803602e-08,
+      "loss": 0.6525,
+      "step": 2971
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2185735443649427e-08,
+      "loss": 0.5258,
+      "step": 2972
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.1661781594678456e-08,
+      "loss": 0.5353,
+      "step": 2973
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1149334358490082e-08,
+      "loss": 0.5402,
+      "step": 2974
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0648394325424305e-08,
+      "loss": 0.606,
+      "step": 2975
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.015896207256728e-08,
+      "loss": 0.5754,
+      "step": 2976
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.68103816374688e-09,
+      "loss": 0.6136,
+      "step": 2977
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.214623149533808e-09,
+      "loss": 0.5916,
+      "step": 2978
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.759717567241588e-09,
+      "loss": 0.5338,
+      "step": 2979
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.316321940921023e-09,
+      "loss": 0.5834,
+      "step": 2980
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.884436781369075e-09,
+      "loss": 0.565,
+      "step": 2981
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.464062586116649e-09,
+      "loss": 0.6049,
+      "step": 2982
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.055199839437476e-09,
+      "loss": 0.6182,
+      "step": 2983
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.657849012344786e-09,
+      "loss": 0.5936,
+      "step": 2984
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.272010562587972e-09,
+      "loss": 0.5292,
+      "step": 2985
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.8976849346548126e-09,
+      "loss": 0.5389,
+      "step": 2986
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.534872559769256e-09,
+      "loss": 0.6047,
+      "step": 2987
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.1835738558925205e-09,
+      "loss": 0.5672,
+      "step": 2988
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.843789227724216e-09,
+      "loss": 0.5709,
+      "step": 2989
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.515519066697893e-09,
+      "loss": 0.5667,
+      "step": 2990
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.198763750979939e-09,
+      "loss": 0.5559,
+      "step": 2991
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.893523645476238e-09,
+      "loss": 0.6063,
+      "step": 2992
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5997991018243973e-09,
+      "loss": 0.5921,
+      "step": 2993
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.3175904583948592e-09,
+      "loss": 0.5405,
+      "step": 2994
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.046898040295343e-09,
+      "loss": 0.5466,
+      "step": 2995
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.787722159364181e-09,
+      "loss": 0.6158,
+      "step": 2996
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.540063114172542e-09,
+      "loss": 0.5626,
+      "step": 2997
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.3039211900255377e-09,
+      "loss": 0.561,
+      "step": 2998
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.079296658960006e-09,
+      "loss": 0.5755,
+      "step": 2999
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.866189779743399e-09,
+      "loss": 0.5111,
+      "step": 3000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.664600797877114e-09,
+      "loss": 0.5778,
+      "step": 3001
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4745299455920514e-09,
+      "loss": 0.6429,
+      "step": 3002
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.2959774418497273e-09,
+      "loss": 0.706,
+      "step": 3003
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.128943492344492e-09,
+      "loss": 0.5722,
+      "step": 3004
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.734282895013102e-10,
+      "loss": 0.5708,
+      "step": 3005
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.294320124724309e-10,
+      "loss": 0.6168,
+      "step": 3006
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.969548271429372e-10,
+      "loss": 0.6246,
+      "step": 3007
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.759968861274167e-10,
+      "loss": 0.5464,
+      "step": 3008
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.665583287699615e-10,
+      "loss": 0.5683,
+      "step": 3009
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.6863928114416747e-10,
+      "loss": 0.603,
+      "step": 3010
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.8223985605313474e-10,
+      "loss": 0.5903,
+      "step": 3011
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.0736015302946777e-10,
+      "loss": 0.6052,
+      "step": 3012
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.440002583352751e-10,
+      "loss": 0.5289,
+      "step": 3013
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.216024496105925e-11,
+      "loss": 0.5758,
+      "step": 3014
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.184017262793717e-11,
+      "loss": 0.5483,
+      "step": 3015
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.3040087783199326e-11,
+      "loss": 0.5205,
+      "step": 3016
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.760023604750587e-12,
+      "loss": 0.5746,
+      "step": 3017
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.5668,
+      "step": 3018
+    },
+    {
+      "epoch": 3.0,
+      "step": 3018,
+      "total_flos": 1.8950842896498033e+18,
+      "train_loss": 0.20122119833729454,
+      "train_runtime": 20495.6759,
+      "train_samples_per_second": 37.717,
+      "train_steps_per_second": 0.147
+    }
+  ],
+  "max_steps": 3018,
+  "num_train_epochs": 3,
+  "total_flos": 1.8950842896498033e+18,
+  "trial_name": null,
+  "trial_params": null
+}