diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,8416 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.7450980392156863,
+  "global_step": 1400,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.347826086956522e-07,
+      "loss": 1.6791,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.695652173913044e-07,
+      "loss": 1.7297,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3043478260869566e-06,
+      "loss": 1.5988,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7391304347826088e-06,
+      "loss": 1.6378,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.173913043478261e-06,
+      "loss": 1.6101,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6086956521739132e-06,
+      "loss": 1.5025,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.043478260869566e-06,
+      "loss": 1.4691,
+      "step": 7
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4782608695652175e-06,
+      "loss": 1.5057,
+      "step": 8
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.91304347826087e-06,
+      "loss": 1.4645,
+      "step": 9
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.347826086956522e-06,
+      "loss": 1.4841,
+      "step": 10
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.782608695652174e-06,
+      "loss": 1.3484,
+      "step": 11
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.2173913043478265e-06,
+      "loss": 1.3993,
+      "step": 12
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.652173913043479e-06,
+      "loss": 1.2537,
+      "step": 13
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.086956521739132e-06,
+      "loss": 1.3931,
+      "step": 14
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.521739130434783e-06,
+      "loss": 1.4166,
+      "step": 15
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.956521739130435e-06,
+      "loss": 1.328,
+      "step": 16
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.391304347826087e-06,
+      "loss": 1.3269,
+      "step": 17
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.82608695652174e-06,
+      "loss": 1.3398,
+      "step": 18
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.260869565217392e-06,
+      "loss": 1.2918,
+      "step": 19
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.695652173913044e-06,
+      "loss": 1.3524,
+      "step": 20
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.130434782608697e-06,
+      "loss": 1.2914,
+      "step": 21
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.565217391304349e-06,
+      "loss": 1.3488,
+      "step": 22
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1e-05,
+      "loss": 1.3098,
+      "step": 23
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0434782608695653e-05,
+      "loss": 1.2008,
+      "step": 24
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0869565217391305e-05,
+      "loss": 1.2517,
+      "step": 25
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1304347826086957e-05,
+      "loss": 1.2726,
+      "step": 26
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1739130434782611e-05,
+      "loss": 1.3406,
+      "step": 27
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2173913043478263e-05,
+      "loss": 1.2348,
+      "step": 28
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2608695652173915e-05,
+      "loss": 1.2797,
+      "step": 29
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3043478260869566e-05,
+      "loss": 1.2801,
+      "step": 30
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3478260869565218e-05,
+      "loss": 1.2752,
+      "step": 31
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.391304347826087e-05,
+      "loss": 1.239,
+      "step": 32
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4347826086956522e-05,
+      "loss": 1.1827,
+      "step": 33
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4782608695652174e-05,
+      "loss": 1.2833,
+      "step": 34
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5217391304347828e-05,
+      "loss": 1.2571,
+      "step": 35
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.565217391304348e-05,
+      "loss": 1.1767,
+      "step": 36
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6086956521739132e-05,
+      "loss": 1.228,
+      "step": 37
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6521739130434785e-05,
+      "loss": 1.2042,
+      "step": 38
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6956521739130437e-05,
+      "loss": 1.209,
+      "step": 39
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.739130434782609e-05,
+      "loss": 1.203,
+      "step": 40
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.782608695652174e-05,
+      "loss": 1.234,
+      "step": 41
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8260869565217393e-05,
+      "loss": 1.1969,
+      "step": 42
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8695652173913045e-05,
+      "loss": 1.1641,
+      "step": 43
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9130434782608697e-05,
+      "loss": 1.2476,
+      "step": 44
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.956521739130435e-05,
+      "loss": 1.1758,
+      "step": 45
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 1.1608,
+      "step": 46
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999977592067603e-05,
+      "loss": 1.1936,
+      "step": 47
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999910368370826e-05,
+      "loss": 1.1984,
+      "step": 48
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999798329210938e-05,
+      "loss": 1.1492,
+      "step": 49
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999964147509006e-05,
+      "loss": 1.1872,
+      "step": 50
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999439806711137e-05,
+      "loss": 1.0389,
+      "step": 51
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999193324977974e-05,
+      "loss": 1.2634,
+      "step": 52
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999890203099519e-05,
+      "loss": 1.1964,
+      "step": 53
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998565926068253e-05,
+      "loss": 1.1295,
+      "step": 54
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999818501170344e-05,
+      "loss": 1.1741,
+      "step": 55
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997759289607854e-05,
+      "loss": 1.1604,
+      "step": 56
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997288761689404e-05,
+      "loss": 1.142,
+      "step": 57
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996773430056806e-05,
+      "loss": 1.1625,
+      "step": 58
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996213297019556e-05,
+      "loss": 1.1561,
+      "step": 59
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995608365087945e-05,
+      "loss": 1.12,
+      "step": 60
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994958636973024e-05,
+      "loss": 1.2068,
+      "step": 61
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999426411558661e-05,
+      "loss": 1.1427,
+      "step": 62
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993524804041255e-05,
+      "loss": 1.1102,
+      "step": 63
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992740705650252e-05,
+      "loss": 1.1538,
+      "step": 64
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991911823927607e-05,
+      "loss": 1.0986,
+      "step": 65
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991038162588018e-05,
+      "loss": 1.1527,
+      "step": 66
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999011972554688e-05,
+      "loss": 1.1093,
+      "step": 67
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989156516920248e-05,
+      "loss": 1.1408,
+      "step": 68
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998814854102482e-05,
+      "loss": 1.1066,
+      "step": 69
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987095802377933e-05,
+      "loss": 1.1342,
+      "step": 70
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998599830569752e-05,
+      "loss": 1.1122,
+      "step": 71
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998485605590211e-05,
+      "loss": 1.1034,
+      "step": 72
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.99836690581108e-05,
+      "loss": 1.0664,
+      "step": 73
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9982437317643218e-05,
+      "loss": 1.0988,
+      "step": 74
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981160840019513e-05,
+      "loss": 1.1115,
+      "step": 75
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979839630960333e-05,
+      "loss": 1.0505,
+      "step": 76
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978473696386787e-05,
+      "loss": 1.0887,
+      "step": 77
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977063042420438e-05,
+      "loss": 1.0579,
+      "step": 78
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9975607675383243e-05,
+      "loss": 1.139,
+      "step": 79
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997410760179756e-05,
+      "loss": 1.1476,
+      "step": 80
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99725628283861e-05,
+      "loss": 1.1369,
+      "step": 81
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99709733620719e-05,
+      "loss": 1.1403,
+      "step": 82
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969339209978287e-05,
+      "loss": 1.0507,
+      "step": 83
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9967660379428855e-05,
+      "loss": 1.1241,
+      "step": 84
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996593687794743e-05,
+      "loss": 1.0748,
+      "step": 85
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996416871325803e-05,
+      "loss": 1.0404,
+      "step": 86
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996235589328484e-05,
+      "loss": 1.1234,
+      "step": 87
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996049842615217e-05,
+      "loss": 1.0707,
+      "step": 88
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995859632018442e-05,
+      "loss": 1.0819,
+      "step": 89
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995664958390604e-05,
+      "loss": 1.1184,
+      "step": 90
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.99546582260415e-05,
+      "loss": 1.0907,
+      "step": 91
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995262225551524e-05,
+      "loss": 1.0984,
+      "step": 92
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9950541681451644e-05,
+      "loss": 1.0967,
+      "step": 93
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948416513174976e-05,
+      "loss": 1.0085,
+      "step": 94
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994624676020936e-05,
+      "loss": 1.1173,
+      "step": 95
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944032432278743e-05,
+      "loss": 1.0637,
+      "step": 96
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941773539306818e-05,
+      "loss": 1.1113,
+      "step": 97
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939470091417012e-05,
+      "loss": 1.0784,
+      "step": 98
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937122098932428e-05,
+      "loss": 1.0924,
+      "step": 99
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934729572375792e-05,
+      "loss": 1.1565,
+      "step": 100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932292522469424e-05,
+      "loss": 1.0751,
+      "step": 101
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992981096013517e-05,
+      "loss": 1.0475,
+      "step": 102
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992728489649437e-05,
+      "loss": 1.195,
+      "step": 103
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992471434286779e-05,
+      "loss": 1.0231,
+      "step": 104
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99220993107756e-05,
+      "loss": 1.1734,
+      "step": 105
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9919439811937283e-05,
+      "loss": 1.114,
+      "step": 106
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9916735858271615e-05,
+      "loss": 1.0786,
+      "step": 107
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9913987461896597e-05,
+      "loss": 1.0595,
+      "step": 108
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991119463512941e-05,
+      "loss": 1.0714,
+      "step": 109
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9908357390486342e-05,
+      "loss": 1.0782,
+      "step": 110
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9905475740682756e-05,
+      "loss": 1.0971,
+      "step": 111
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990254969863301e-05,
+      "loss": 1.0634,
+      "step": 112
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9899579277450417e-05,
+      "loss": 1.189,
+      "step": 113
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989656449044718e-05,
+      "loss": 1.1077,
+      "step": 114
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9893505351134322e-05,
+      "loss": 1.0724,
+      "step": 115
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9890401873221642e-05,
+      "loss": 1.0403,
+      "step": 116
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9887254070617644e-05,
+      "loss": 1.0682,
+      "step": 117
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988406195742948e-05,
+      "loss": 1.1245,
+      "step": 118
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9880825547962882e-05,
+      "loss": 1.0633,
+      "step": 119
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9877544856722094e-05,
+      "loss": 1.0761,
+      "step": 120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987421989840982e-05,
+      "loss": 1.0576,
+      "step": 121
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987085068792715e-05,
+      "loss": 1.0742,
+      "step": 122
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986743724037349e-05,
+      "loss": 1.0786,
+      "step": 123
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98639795710465e-05,
+      "loss": 1.1654,
+      "step": 124
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.986047769544203e-05,
+      "loss": 1.1486,
+      "step": 125
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9856931629254032e-05,
+      "loss": 1.0743,
+      "step": 126
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9853341388374504e-05,
+      "loss": 1.1069,
+      "step": 127
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9849706988893433e-05,
+      "loss": 1.1149,
+      "step": 128
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9846028447098687e-05,
+      "loss": 1.1201,
+      "step": 129
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984230577947597e-05,
+      "loss": 1.0538,
+      "step": 130
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983853900270874e-05,
+      "loss": 1.1556,
+      "step": 131
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983472813367813e-05,
+      "loss": 1.0697,
+      "step": 132
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9830873189462886e-05,
+      "loss": 1.1127,
+      "step": 133
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9826974187339267e-05,
+      "loss": 1.0996,
+      "step": 134
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982303114478099e-05,
+      "loss": 1.0341,
+      "step": 135
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9819044079459142e-05,
+      "loss": 0.9972,
+      "step": 136
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9815013009242103e-05,
+      "loss": 1.0652,
+      "step": 137
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981093795219546e-05,
+      "loss": 1.0404,
+      "step": 138
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980681892658193e-05,
+      "loss": 1.0773,
+      "step": 139
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980265595086129e-05,
+      "loss": 0.9761,
+      "step": 140
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9798449043690272e-05,
+      "loss": 1.0444,
+      "step": 141
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9794198223922496e-05,
+      "loss": 1.0361,
+      "step": 142
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9789903510608374e-05,
+      "loss": 1.0651,
+      "step": 143
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9785564922995042e-05,
+      "loss": 1.0096,
+      "step": 144
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978118248052625e-05,
+      "loss": 1.012,
+      "step": 145
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9776756202842297e-05,
+      "loss": 1.022,
+      "step": 146
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977228610977992e-05,
+      "loss": 1.0339,
+      "step": 147
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976777222137224e-05,
+      "loss": 1.1113,
+      "step": 148
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9763214557848634e-05,
+      "loss": 1.0431,
+      "step": 149
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9758613139634662e-05,
+      "loss": 1.0113,
+      "step": 150
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975396798735198e-05,
+      "loss": 1.0248,
+      "step": 151
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9749279121818235e-05,
+      "loss": 1.1174,
+      "step": 152
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9744546564046995e-05,
+      "loss": 1.0334,
+      "step": 153
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9739770335247616e-05,
+      "loss": 1.0291,
+      "step": 154
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9734950456825187e-05,
+      "loss": 1.0916,
+      "step": 155
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9730086950380404e-05,
+      "loss": 1.045,
+      "step": 156
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9725179837709494e-05,
+      "loss": 1.0836,
+      "step": 157
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972022914080411e-05,
+      "loss": 1.0648,
+      "step": 158
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9715234881851226e-05,
+      "loss": 1.0938,
+      "step": 159
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9710197083233044e-05,
+      "loss": 1.0282,
+      "step": 160
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9705115767526894e-05,
+      "loss": 1.0196,
+      "step": 161
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9699990957505136e-05,
+      "loss": 1.0088,
+      "step": 162
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9694822676135045e-05,
+      "loss": 1.0781,
+      "step": 163
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9689610946578725e-05,
+      "loss": 1.0253,
+      "step": 164
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968435579219299e-05,
+      "loss": 1.0348,
+      "step": 165
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9679057236529266e-05,
+      "loss": 1.079,
+      "step": 166
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9673715303333495e-05,
+      "loss": 0.9829,
+      "step": 167
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9668330016546004e-05,
+      "loss": 1.0186,
+      "step": 168
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9662901400301432e-05,
+      "loss": 1.0874,
+      "step": 169
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965742947892858e-05,
+      "loss": 1.0479,
+      "step": 170
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.965191427695035e-05,
+      "loss": 1.0716,
+      "step": 171
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964635581908359e-05,
+      "loss": 1.0624,
+      "step": 172
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9640754130239004e-05,
+      "loss": 1.0209,
+      "step": 173
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9635109235521057e-05,
+      "loss": 1.044,
+      "step": 174
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9629421160227825e-05,
+      "loss": 1.0349,
+      "step": 175
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962368992985091e-05,
+      "loss": 1.052,
+      "step": 176
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9617915570075317e-05,
+      "loss": 1.1025,
+      "step": 177
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961209810677934e-05,
+      "loss": 1.0033,
+      "step": 178
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9606237566034443e-05,
+      "loss": 1.0843,
+      "step": 179
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9600333974105145e-05,
+      "loss": 0.9795,
+      "step": 180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9594387357448908e-05,
+      "loss": 1.0159,
+      "step": 181
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9588397742716004e-05,
+      "loss": 1.04,
+      "step": 182
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958236515674941e-05,
+      "loss": 1.048,
+      "step": 183
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9576289626584685e-05,
+      "loss": 1.0624,
+      "step": 184
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957017117944984e-05,
+      "loss": 1.0334,
+      "step": 185
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9564009842765225e-05,
+      "loss": 1.0801,
+      "step": 186
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9557805644143408e-05,
+      "loss": 1.038,
+      "step": 187
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955155861138903e-05,
+      "loss": 1.0932,
+      "step": 188
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9545268772498723e-05,
+      "loss": 1.0315,
+      "step": 189
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9538936155660934e-05,
+      "loss": 1.0132,
+      "step": 190
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9532560789255838e-05,
+      "loss": 1.0522,
+      "step": 191
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9526142701855187e-05,
+      "loss": 1.026,
+      "step": 192
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9519681922222195e-05,
+      "loss": 1.0,
+      "step": 193
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951317847931141e-05,
+      "loss": 1.0203,
+      "step": 194
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950663240226857e-05,
+      "loss": 1.0467,
+      "step": 195
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9500043720430484e-05,
+      "loss": 1.021,
+      "step": 196
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9493412463324903e-05,
+      "loss": 1.05,
+      "step": 197
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9486738660670373e-05,
+      "loss": 0.9964,
+      "step": 198
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9480022342376125e-05,
+      "loss": 1.0743,
+      "step": 199
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9473263538541916e-05,
+      "loss": 1.1076,
+      "step": 200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946646227945791e-05,
+      "loss": 1.0902,
+      "step": 201
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945961859560454e-05,
+      "loss": 1.0278,
+      "step": 202
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9452732517652366e-05,
+      "loss": 1.0401,
+      "step": 203
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9445804076461946e-05,
+      "loss": 1.0421,
+      "step": 204
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9438833303083677e-05,
+      "loss": 1.0122,
+      "step": 205
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943182022875769e-05,
+      "loss": 1.0435,
+      "step": 206
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942476488491369e-05,
+      "loss": 1.0039,
+      "step": 207
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9417667303170803e-05,
+      "loss": 1.0056,
+      "step": 208
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941052751533745e-05,
+      "loss": 1.0392,
+      "step": 209
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940334555341122e-05,
+      "loss": 1.0488,
+      "step": 210
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9396121449578694e-05,
+      "loss": 1.0113,
+      "step": 211
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938885523621531e-05,
+      "loss": 1.0754,
+      "step": 212
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9381546945885236e-05,
+      "loss": 0.9854,
+      "step": 213
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9374196611341212e-05,
+      "loss": 1.0239,
+      "step": 214
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9366804265524388e-05,
+      "loss": 1.0045,
+      "step": 215
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.935936994156421e-05,
+      "loss": 1.0204,
+      "step": 216
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.935189367277824e-05,
+      "loss": 1.041,
+      "step": 217
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9344375492672024e-05,
+      "loss": 1.033,
+      "step": 218
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9336815434938935e-05,
+      "loss": 1.0796,
+      "step": 219
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932921353346003e-05,
+      "loss": 1.0618,
+      "step": 220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932156982230388e-05,
+      "loss": 1.0393,
+      "step": 221
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9313884335726443e-05,
+      "loss": 0.9881,
+      "step": 222
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9306157108170892e-05,
+      "loss": 0.9971,
+      "step": 223
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9298388174267463e-05,
+      "loss": 1.0703,
+      "step": 224
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9290577568833305e-05,
+      "loss": 1.0303,
+      "step": 225
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9282725326872324e-05,
+      "loss": 0.9709,
+      "step": 226
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.927483148357502e-05,
+      "loss": 0.9944,
+      "step": 227
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9266896074318335e-05,
+      "loss": 0.999,
+      "step": 228
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9258919134665494e-05,
+      "loss": 1.0798,
+      "step": 229
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9250900700365837e-05,
+      "loss": 0.9999,
+      "step": 230
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9242840807354676e-05,
+      "loss": 1.1024,
+      "step": 231
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.923473949175311e-05,
+      "loss": 0.9646,
+      "step": 232
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9226596789867895e-05,
+      "loss": 0.9891,
+      "step": 233
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921841273819125e-05,
+      "loss": 1.0312,
+      "step": 234
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921018737340071e-05,
+      "loss": 1.0352,
+      "step": 235
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920192073235896e-05,
+      "loss": 0.9777,
+      "step": 236
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9193612852113666e-05,
+      "loss": 0.9957,
+      "step": 237
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918526376989731e-05,
+      "loss": 1.0297,
+      "step": 238
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9176873523127025e-05,
+      "loss": 1.0033,
+      "step": 239
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9168442149404438e-05,
+      "loss": 0.9653,
+      "step": 240
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9159969686515462e-05,
+      "loss": 0.9856,
+      "step": 241
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9151456172430186e-05,
+      "loss": 1.0076,
+      "step": 242
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9142901645302654e-05,
+      "loss": 1.0059,
+      "step": 243
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9134306143470722e-05,
+      "loss": 1.0037,
+      "step": 244
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912566970545587e-05,
+      "loss": 1.0349,
+      "step": 245
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911699236996305e-05,
+      "loss": 1.0429,
+      "step": 246
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9108274175880483e-05,
+      "loss": 0.9261,
+      "step": 247
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9099515162279515e-05,
+      "loss": 1.0949,
+      "step": 248
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909071536841442e-05,
+      "loss": 1.0227,
+      "step": 249
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9081874833722234e-05,
+      "loss": 0.9658,
+      "step": 250
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9072993597822586e-05,
+      "loss": 1.0351,
+      "step": 251
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906407170051749e-05,
+      "loss": 1.0135,
+      "step": 252
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905510918179121e-05,
+      "loss": 0.9789,
+      "step": 253
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9046106081810047e-05,
+      "loss": 1.0324,
+      "step": 254
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903706244092217e-05,
+      "loss": 0.9867,
+      "step": 255
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9027978299657436e-05,
+      "loss": 1.0064,
+      "step": 256
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9018853698727216e-05,
+      "loss": 1.0491,
+      "step": 257
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900968867902419e-05,
+      "loss": 1.0126,
+      "step": 258
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9000483281622198e-05,
+      "loss": 1.0314,
+      "step": 259
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8991237547776014e-05,
+      "loss": 0.9933,
+      "step": 260
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8981951518921194e-05,
+      "loss": 1.0303,
+      "step": 261
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8972625236673887e-05,
+      "loss": 0.9933,
+      "step": 262
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896325874283063e-05,
+      "loss": 1.0585,
+      "step": 263
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895385207936817e-05,
+      "loss": 1.0379,
+      "step": 264
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894440528844329e-05,
+      "loss": 1.0336,
+      "step": 265
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8934918412392596e-05,
+      "loss": 1.0545,
+      "step": 266
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8925391493732346e-05,
+      "loss": 1.0106,
+      "step": 267
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891582457515825e-05,
+      "loss": 0.9264,
+      "step": 268
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890621769954528e-05,
+      "loss": 1.0619,
+      "step": 269
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8896570909947477e-05,
+      "loss": 0.9478,
+      "step": 270
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8886884249597763e-05,
+      "loss": 1.0347,
+      "step": 271
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887715776190775e-05,
+      "loss": 1.0231,
+      "step": 272
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8867391490467534e-05,
+      "loss": 0.9832,
+      "step": 273
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8857585479045493e-05,
+      "loss": 0.9978,
+      "step": 274
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884773977158813e-05,
+      "loss": 0.9926,
+      "step": 275
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8837854412219828e-05,
+      "loss": 0.9447,
+      "step": 276
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8827929445242678e-05,
+      "loss": 1.0038,
+      "step": 277
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8817964915136277e-05,
+      "loss": 0.9859,
+      "step": 278
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8807960866557535e-05,
+      "loss": 1.024,
+      "step": 279
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8797917344340453e-05,
+      "loss": 0.9742,
+      "step": 280
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8787834393495952e-05,
+      "loss": 0.9774,
+      "step": 281
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8777712059211643e-05,
+      "loss": 1.0539,
+      "step": 282
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8767550386851646e-05,
+      "loss": 1.0163,
+      "step": 283
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875734942195637e-05,
+      "loss": 1.0675,
+      "step": 284
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8747109210242326e-05,
+      "loss": 1.0296,
+      "step": 285
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8736829797601903e-05,
+      "loss": 1.003,
+      "step": 286
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8726511230103183e-05,
+      "loss": 0.9922,
+      "step": 287
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8716153553989716e-05,
+      "loss": 0.9657,
+      "step": 288
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8705756815680324e-05,
+      "loss": 0.9735,
+      "step": 289
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8695321061768886e-05,
+      "loss": 0.9142,
+      "step": 290
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8684846339024145e-05,
+      "loss": 0.9359,
+      "step": 291
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8674332694389464e-05,
+      "loss": 0.9541,
+      "step": 292
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8663780174982656e-05,
+      "loss": 1.0237,
+      "step": 293
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8653188828095754e-05,
+      "loss": 0.9368,
+      "step": 294
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8642558701194787e-05,
+      "loss": 1.0263,
+      "step": 295
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8631889841919596e-05,
+      "loss": 0.9919,
+      "step": 296
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.862118229808359e-05,
+      "loss": 1.001,
+      "step": 297
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8610436117673557e-05,
+      "loss": 1.0349,
+      "step": 298
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.859965134884943e-05,
+      "loss": 0.9706,
+      "step": 299
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8588828039944086e-05,
+      "loss": 1.0074,
+      "step": 300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857796623946312e-05,
+      "loss": 0.9529,
+      "step": 301
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8567065996084628e-05,
+      "loss": 0.9359,
+      "step": 302
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8556127358658995e-05,
+      "loss": 0.9875,
+      "step": 303
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854515037620867e-05,
+      "loss": 1.0257,
+      "step": 304
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853413509792795e-05,
+      "loss": 1.0232,
+      "step": 305
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8523081573182754e-05,
+      "loss": 0.9977,
+      "step": 306
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8511989851510416e-05,
+      "loss": 1.0244,
+      "step": 307
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8500859982619438e-05,
+      "loss": 0.9651,
+      "step": 308
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.848969201638929e-05,
+      "loss": 1.0111,
+      "step": 309
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847848600287019e-05,
+      "loss": 1.0351,
+      "step": 310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8467241992282842e-05,
+      "loss": 0.9433,
+      "step": 311
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845596003501826e-05,
+      "loss": 1.0271,
+      "step": 312
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8444640181637514e-05,
+      "loss": 0.9916,
+      "step": 313
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8433282482871497e-05,
+      "loss": 0.9801,
+      "step": 314
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8421886989620724e-05,
+      "loss": 0.9983,
+      "step": 315
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.841045375295508e-05,
+      "loss": 0.9564,
+      "step": 316
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839898282411361e-05,
+      "loss": 1.0238,
+      "step": 317
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8387474254504265e-05,
+      "loss": 0.9463,
+      "step": 318
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8375928095703703e-05,
+      "loss": 0.9783,
+      "step": 319
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8364344399457028e-05,
+      "loss": 0.9805,
+      "step": 320
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.835272321767758e-05,
+      "loss": 0.9801,
+      "step": 321
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8341064602446686e-05,
+      "loss": 0.9734,
+      "step": 322
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832936860601344e-05,
+      "loss": 0.9487,
+      "step": 323
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8317635280794466e-05,
+      "loss": 0.9983,
+      "step": 324
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8305864679373667e-05,
+      "loss": 0.9424,
+      "step": 325
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829405685450202e-05,
+      "loss": 1.0011,
+      "step": 326
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8282211859097308e-05,
+      "loss": 0.9649,
+      "step": 327
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8270329746243903e-05,
+      "loss": 0.9664,
+      "step": 328
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8258410569192526e-05,
+      "loss": 1.0741,
+      "step": 329
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824645438135999e-05,
+      "loss": 0.9481,
+      "step": 330
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8234461236328992e-05,
+      "loss": 0.9758,
+      "step": 331
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.822243118784785e-05,
+      "loss": 0.9374,
+      "step": 332
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821036428983026e-05,
+      "loss": 1.0554,
+      "step": 333
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8198260596355077e-05,
+      "loss": 0.9472,
+      "step": 334
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.818612016166604e-05,
+      "loss": 0.9414,
+      "step": 335
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8173943040171567e-05,
+      "loss": 1.0546,
+      "step": 336
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8161729286444474e-05,
+      "loss": 1.058,
+      "step": 337
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.814947895522176e-05,
+      "loss": 0.9709,
+      "step": 338
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8137192101404343e-05,
+      "loss": 0.9947,
+      "step": 339
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8124868780056814e-05,
+      "loss": 0.9748,
+      "step": 340
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.811250904640721e-05,
+      "loss": 0.9811,
+      "step": 341
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8100112955846746e-05,
+      "loss": 1.018,
+      "step": 342
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8087680563929574e-05,
+      "loss": 0.9788,
+      "step": 343
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.807521192637253e-05,
+      "loss": 0.9772,
+      "step": 344
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.806270709905489e-05,
+      "loss": 0.9929,
+      "step": 345
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805016613801813e-05,
+      "loss": 1.007,
+      "step": 346
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8037589099465637e-05,
+      "loss": 0.9805,
+      "step": 347
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8024976039762507e-05,
+      "loss": 0.9382,
+      "step": 348
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8012327015435255e-05,
+      "loss": 0.9942,
+      "step": 349
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7999642083171576e-05,
+      "loss": 1.0483,
+      "step": 350
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.798692129982009e-05,
+      "loss": 0.977,
+      "step": 351
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7974164722390094e-05,
+      "loss": 0.9409,
+      "step": 352
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796137240805129e-05,
+      "loss": 0.9485,
+      "step": 353
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7948544414133534e-05,
+      "loss": 0.9804,
+      "step": 354
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7935680798126597e-05,
+      "loss": 1.0293,
+      "step": 355
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.792278161767989e-05,
+      "loss": 1.0039,
+      "step": 356
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.79098469306022e-05,
+      "loss": 0.9322,
+      "step": 357
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7896876794861443e-05,
+      "loss": 0.9763,
+      "step": 358
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.788387126858441e-05,
+      "loss": 0.9941,
+      "step": 359
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7870830410056487e-05,
+      "loss": 1.0175,
+      "step": 360
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7857754277721417e-05,
+      "loss": 0.9435,
+      "step": 361
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7844642930181008e-05,
+      "loss": 0.9538,
+      "step": 362
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7831496426194906e-05,
+      "loss": 1.0082,
+      "step": 363
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.78183148246803e-05,
+      "loss": 1.0124,
+      "step": 364
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.780509818471168e-05,
+      "loss": 0.9355,
+      "step": 365
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.779184656552056e-05,
+      "loss": 1.076,
+      "step": 366
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7778560026495218e-05,
+      "loss": 0.9652,
+      "step": 367
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7765238627180424e-05,
+      "loss": 0.9779,
+      "step": 368
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775188242727719e-05,
+      "loss": 0.9888,
+      "step": 369
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.773849148664247e-05,
+      "loss": 0.956,
+      "step": 370
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.772506586528893e-05,
+      "loss": 0.9886,
+      "step": 371
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7711605623384653e-05,
+      "loss": 0.9966,
+      "step": 372
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7698110821252873e-05,
+      "loss": 0.9489,
+      "step": 373
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7684581519371714e-05,
+      "loss": 0.9426,
+      "step": 374
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7671017778373914e-05,
+      "loss": 0.9119,
+      "step": 375
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7657419659046555e-05,
+      "loss": 0.9933,
+      "step": 376
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7643787222330773e-05,
+      "loss": 0.9659,
+      "step": 377
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7630120529321518e-05,
+      "loss": 0.9642,
+      "step": 378
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7616419641267267e-05,
+      "loss": 1.0612,
+      "step": 379
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.760268461956972e-05,
+      "loss": 0.9142,
+      "step": 380
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758891552578357e-05,
+      "loss": 0.987,
+      "step": 381
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7575112421616203e-05,
+      "loss": 0.9447,
+      "step": 382
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7561275368927422e-05,
+      "loss": 1.0042,
+      "step": 383
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7547404429729174e-05,
+      "loss": 0.9916,
+      "step": 384
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.753349966618528e-05,
+      "loss": 0.9748,
+      "step": 385
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.751956114061113e-05,
+      "loss": 0.9446,
+      "step": 386
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7505588915473437e-05,
+      "loss": 0.9261,
+      "step": 387
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7491583053389937e-05,
+      "loss": 0.9571,
+      "step": 388
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.747754361712911e-05,
+      "loss": 0.9491,
+      "step": 389
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7463470669609907e-05,
+      "loss": 0.9405,
+      "step": 390
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7449364273901457e-05,
+      "loss": 0.931,
+      "step": 391
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.743522449322279e-05,
+      "loss": 0.9834,
+      "step": 392
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.742105139094256e-05,
+      "loss": 0.9414,
+      "step": 393
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7406845030578747e-05,
+      "loss": 0.9359,
+      "step": 394
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7392605475798387e-05,
+      "loss": 0.9825,
+      "step": 395
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7378332790417275e-05,
+      "loss": 0.9679,
+      "step": 396
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7364027038399682e-05,
+      "loss": 0.9437,
+      "step": 397
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.734968828385808e-05,
+      "loss": 0.9085,
+      "step": 398
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7335316591052828e-05,
+      "loss": 0.9587,
+      "step": 399
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7320912024391915e-05,
+      "loss": 1.014,
+      "step": 400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7306474648430653e-05,
+      "loss": 0.9894,
+      "step": 401
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.729200452787139e-05,
+      "loss": 0.9684,
+      "step": 402
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7277501727563227e-05,
+      "loss": 1.0049,
+      "step": 403
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.726296631250171e-05,
+      "loss": 0.9846,
+      "step": 404
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.724839834782856e-05,
+      "loss": 0.9728,
+      "step": 405
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7233797898831376e-05,
+      "loss": 0.9802,
+      "step": 406
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7219165030943327e-05,
+      "loss": 0.9484,
+      "step": 407
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.720449980974288e-05,
+      "loss": 0.9905,
+      "step": 408
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7189802300953488e-05,
+      "loss": 0.9704,
+      "step": 409
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717507257044331e-05,
+      "loss": 0.9493,
+      "step": 410
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.716031068422491e-05,
+      "loss": 0.9792,
+      "step": 411
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7145516708454953e-05,
+      "loss": 0.9278,
+      "step": 412
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7130690709433927e-05,
+      "loss": 0.9525,
+      "step": 413
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.711583275360582e-05,
+      "loss": 0.9911,
+      "step": 414
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.710094290755785e-05,
+      "loss": 0.8888,
+      "step": 415
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7086021238020154e-05,
+      "loss": 0.9815,
+      "step": 416
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 0.9102,
+      "step": 417
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7056082696108896e-05,
+      "loss": 0.9677,
+      "step": 418
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.70410659579075e-05,
+      "loss": 0.8997,
+      "step": 419
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7026017664560108e-05,
+      "loss": 0.9211,
+      "step": 420
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7010937883506936e-05,
+      "loss": 0.9918,
+      "step": 421
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.699582668232934e-05,
+      "loss": 0.9244,
+      "step": 422
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.698068412874946e-05,
+      "loss": 0.9245,
+      "step": 423
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6965510290629973e-05,
+      "loss": 0.9079,
+      "step": 424
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.695030523597374e-05,
+      "loss": 0.9392,
+      "step": 425
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6935069032923525e-05,
+      "loss": 0.9178,
+      "step": 426
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6919801749761696e-05,
+      "loss": 1.0048,
+      "step": 427
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6904503454909905e-05,
+      "loss": 0.9467,
+      "step": 428
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6889174216928776e-05,
+      "loss": 0.9948,
+      "step": 429
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6873814104517617e-05,
+      "loss": 0.9644,
+      "step": 430
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6858423186514108e-05,
+      "loss": 0.9621,
+      "step": 431
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6843001531893967e-05,
+      "loss": 0.932,
+      "step": 432
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6827549209770676e-05,
+      "loss": 0.9902,
+      "step": 433
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6812066289395157e-05,
+      "loss": 0.9288,
+      "step": 434
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6796552840155453e-05,
+      "loss": 0.9686,
+      "step": 435
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6781008931576433e-05,
+      "loss": 0.9789,
+      "step": 436
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.676543463331946e-05,
+      "loss": 0.9389,
+      "step": 437
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6749830015182106e-05,
+      "loss": 0.9272,
+      "step": 438
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6734195147097815e-05,
+      "loss": 0.9309,
+      "step": 439
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6718530099135596e-05,
+      "loss": 0.9441,
+      "step": 440
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6702834941499718e-05,
+      "loss": 0.9532,
+      "step": 441
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6687109744529394e-05,
+      "loss": 0.9301,
+      "step": 442
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6671354578698444e-05,
+      "loss": 0.9464,
+      "step": 443
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665556951461501e-05,
+      "loss": 0.9501,
+      "step": 444
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6639754623021227e-05,
+      "loss": 0.8981,
+      "step": 445
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6623909974792888e-05,
+      "loss": 1.0658,
+      "step": 446
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6608035640939165e-05,
+      "loss": 0.9744,
+      "step": 447
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6592131692602257e-05,
+      "loss": 0.9901,
+      "step": 448
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.657619820105708e-05,
+      "loss": 0.9162,
+      "step": 449
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.656023523771095e-05,
+      "loss": 0.9462,
+      "step": 450
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6544242874103278e-05,
+      "loss": 0.909,
+      "step": 451
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6528221181905217e-05,
+      "loss": 0.9495,
+      "step": 452
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6512170232919372e-05,
+      "loss": 0.9533,
+      "step": 453
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6496090099079452e-05,
+      "loss": 0.9985,
+      "step": 454
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6479980852449974e-05,
+      "loss": 0.9243,
+      "step": 455
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6463842565225914e-05,
+      "loss": 0.9321,
+      "step": 456
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6447675309732403e-05,
+      "loss": 0.9812,
+      "step": 457
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.64314791584244e-05,
+      "loss": 0.9181,
+      "step": 458
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6415254183886356e-05,
+      "loss": 0.9205,
+      "step": 459
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6399000458831893e-05,
+      "loss": 0.9581,
+      "step": 460
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6382718056103486e-05,
+      "loss": 0.9824,
+      "step": 461
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6366407048672135e-05,
+      "loss": 0.9144,
+      "step": 462
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6350067509637028e-05,
+      "loss": 0.9512,
+      "step": 463
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6333699512225224e-05,
+      "loss": 0.9713,
+      "step": 464
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6317303129791315e-05,
+      "loss": 0.9305,
+      "step": 465
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6300878435817115e-05,
+      "loss": 0.9706,
+      "step": 466
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.62844255039113e-05,
+      "loss": 1.0021,
+      "step": 467
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.626794440780911e-05,
+      "loss": 0.8972,
+      "step": 468
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6251435221372007e-05,
+      "loss": 0.9839,
+      "step": 469
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6234898018587336e-05,
+      "loss": 0.9504,
+      "step": 470
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6218332873568e-05,
+      "loss": 0.9305,
+      "step": 471
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.620173986055213e-05,
+      "loss": 0.93,
+      "step": 472
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6185119053902748e-05,
+      "loss": 0.9323,
+      "step": 473
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.616847052810744e-05,
+      "loss": 0.8579,
+      "step": 474
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6151794357778006e-05,
+      "loss": 0.9636,
+      "step": 475
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.613509061765015e-05,
+      "loss": 0.9515,
+      "step": 476
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.611835938258313e-05,
+      "loss": 0.9269,
+      "step": 477
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6101600727559423e-05,
+      "loss": 0.901,
+      "step": 478
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6084814727684386e-05,
+      "loss": 0.9334,
+      "step": 479
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6068001458185934e-05,
+      "loss": 0.9775,
+      "step": 480
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6051160994414187e-05,
+      "loss": 0.961,
+      "step": 481
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.603429341184114e-05,
+      "loss": 0.9844,
+      "step": 482
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.601739878606032e-05,
+      "loss": 1.0144,
+      "step": 483
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.600047719278645e-05,
+      "loss": 0.9194,
+      "step": 484
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5983528707855128e-05,
+      "loss": 0.9532,
+      "step": 485
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.596655340722244e-05,
+      "loss": 0.939,
+      "step": 486
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5949551366964675e-05,
+      "loss": 0.9464,
+      "step": 487
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.593252266327794e-05,
+      "loss": 0.99,
+      "step": 488
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5915467372477843e-05,
+      "loss": 0.9315,
+      "step": 489
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5898385570999146e-05,
+      "loss": 0.9315,
+      "step": 490
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5881277335395425e-05,
+      "loss": 0.9095,
+      "step": 491
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5864142742338707e-05,
+      "loss": 0.9562,
+      "step": 492
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.584698186861916e-05,
+      "loss": 0.9453,
+      "step": 493
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5829794791144723e-05,
+      "loss": 0.8973,
+      "step": 494
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5812581586940767e-05,
+      "loss": 0.8567,
+      "step": 495
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5795342333149757e-05,
+      "loss": 0.9142,
+      "step": 496
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.57780771070309e-05,
+      "loss": 0.8777,
+      "step": 497
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.57607859859598e-05,
+      "loss": 0.9424,
+      "step": 498
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5743469047428113e-05,
+      "loss": 0.908,
+      "step": 499
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.572612636904319e-05,
+      "loss": 0.9652,
+      "step": 500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5708758028527754e-05,
+      "loss": 0.9331,
+      "step": 501
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5691364103719515e-05,
+      "loss": 0.9343,
+      "step": 502
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5673944672570858e-05,
+      "loss": 0.9205,
+      "step": 503
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.565649981314847e-05,
+      "loss": 0.9272,
+      "step": 504
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5639029603632995e-05,
+      "loss": 0.9948,
+      "step": 505
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5621534122318682e-05,
+      "loss": 0.9337,
+      "step": 506
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5604013447613057e-05,
+      "loss": 0.9129,
+      "step": 507
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5586467658036526e-05,
+      "loss": 0.9152,
+      "step": 508
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.556889683222207e-05,
+      "loss": 0.9607,
+      "step": 509
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5551301048914863e-05,
+      "loss": 1.0045,
+      "step": 510
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5533680386971927e-05,
+      "loss": 0.8021,
+      "step": 511
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5516034925361785e-05,
+      "loss": 0.7985,
+      "step": 512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5498364743164098e-05,
+      "loss": 0.7471,
+      "step": 513
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5480669919569313e-05,
+      "loss": 0.6998,
+      "step": 514
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5462950533878318e-05,
+      "loss": 0.7766,
+      "step": 515
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.544520666550207e-05,
+      "loss": 0.718,
+      "step": 516
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5427438393961246e-05,
+      "loss": 0.7452,
+      "step": 517
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.54096457988859e-05,
+      "loss": 0.7485,
+      "step": 518
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5391828960015074e-05,
+      "loss": 0.7644,
+      "step": 519
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.537398795719648e-05,
+      "loss": 0.8274,
+      "step": 520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5356122870386117e-05,
+      "loss": 0.7165,
+      "step": 521
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.533823377964791e-05,
+      "loss": 0.8017,
+      "step": 522
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5320320765153367e-05,
+      "loss": 0.6981,
+      "step": 523
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5302383907181212e-05,
+      "loss": 0.7214,
+      "step": 524
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5284423286117035e-05,
+      "loss": 0.7203,
+      "step": 525
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.52664389824529e-05,
+      "loss": 0.6644,
+      "step": 526
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.524843107678702e-05,
+      "loss": 0.8209,
+      "step": 527
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5230399649823389e-05,
+      "loss": 0.7709,
+      "step": 528
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.52123447823714e-05,
+      "loss": 0.7315,
+      "step": 529
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5194266555345505e-05,
+      "loss": 0.7588,
+      "step": 530
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5176165049764846e-05,
+      "loss": 0.7513,
+      "step": 531
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5158040346752878e-05,
+      "loss": 0.7339,
+      "step": 532
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.513989252753703e-05,
+      "loss": 0.7708,
+      "step": 533
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5121721673448319e-05,
+      "loss": 0.7232,
+      "step": 534
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5103527865921005e-05,
+      "loss": 0.7556,
+      "step": 535
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5085311186492206e-05,
+      "loss": 0.7018,
+      "step": 536
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5067071716801546e-05,
+      "loss": 0.6902,
+      "step": 537
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5048809538590789e-05,
+      "loss": 0.7054,
+      "step": 538
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5030524733703464e-05,
+      "loss": 0.796,
+      "step": 539
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5012217384084501e-05,
+      "loss": 0.7685,
+      "step": 540
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4993887571779877e-05,
+      "loss": 0.7459,
+      "step": 541
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4975535378936228e-05,
+      "loss": 0.6768,
+      "step": 542
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4957160887800494e-05,
+      "loss": 0.7152,
+      "step": 543
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4938764180719548e-05,
+      "loss": 0.7735,
+      "step": 544
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4920345340139819e-05,
+      "loss": 0.7751,
+      "step": 545
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.490190444860694e-05,
+      "loss": 0.7263,
+      "step": 546
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4883441588765355e-05,
+      "loss": 0.6805,
+      "step": 547
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4864956843357967e-05,
+      "loss": 0.7375,
+      "step": 548
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4846450295225769e-05,
+      "loss": 0.7809,
+      "step": 549
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.482792202730745e-05,
+      "loss": 0.6919,
+      "step": 550
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4809372122639047e-05,
+      "loss": 0.7146,
+      "step": 551
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4790800664353562e-05,
+      "loss": 0.7679,
+      "step": 552
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.477220773568059e-05,
+      "loss": 0.7337,
+      "step": 553
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.475359341994595e-05,
+      "loss": 0.7381,
+      "step": 554
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.473495780057131e-05,
+      "loss": 0.7658,
+      "step": 555
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4716300961073808e-05,
+      "loss": 0.7123,
+      "step": 556
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.469762298506568e-05,
+      "loss": 0.6939,
+      "step": 557
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4678923956253894e-05,
+      "loss": 0.7063,
+      "step": 558
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4660203958439768e-05,
+      "loss": 0.7473,
+      "step": 559
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4641463075518587e-05,
+      "loss": 0.7373,
+      "step": 560
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4622701391479237e-05,
+      "loss": 0.7771,
+      "step": 561
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.460391899040383e-05,
+      "loss": 0.718,
+      "step": 562
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4585115956467322e-05,
+      "loss": 0.7029,
+      "step": 563
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4566292373937133e-05,
+      "loss": 0.7365,
+      "step": 564
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4547448327172778e-05,
+      "loss": 0.7436,
+      "step": 565
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4528583900625481e-05,
+      "loss": 0.7444,
+      "step": 566
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.45096991788378e-05,
+      "loss": 0.7441,
+      "step": 567
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4490794246443249e-05,
+      "loss": 0.7738,
+      "step": 568
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4471869188165923e-05,
+      "loss": 0.7806,
+      "step": 569
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4452924088820101e-05,
+      "loss": 0.6921,
+      "step": 570
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4433959033309888e-05,
+      "loss": 0.7087,
+      "step": 571
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4414974106628816e-05,
+      "loss": 0.7206,
+      "step": 572
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4395969393859478e-05,
+      "loss": 0.7585,
+      "step": 573
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4376944980173138e-05,
+      "loss": 0.7575,
+      "step": 574
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.435790095082935e-05,
+      "loss": 0.7606,
+      "step": 575
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4338837391175582e-05,
+      "loss": 0.6768,
+      "step": 576
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4319754386646824e-05,
+      "loss": 0.7467,
+      "step": 577
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4300652022765207e-05,
+      "loss": 0.7047,
+      "step": 578
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4281530385139627e-05,
+      "loss": 0.7383,
+      "step": 579
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4262389559465364e-05,
+      "loss": 0.7478,
+      "step": 580
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4243229631523678e-05,
+      "loss": 0.722,
+      "step": 581
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4224050687181442e-05,
+      "loss": 0.6896,
+      "step": 582
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4204852812390757e-05,
+      "loss": 0.689,
+      "step": 583
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4185636093188558e-05,
+      "loss": 0.7549,
+      "step": 584
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4166400615696232e-05,
+      "loss": 0.7865,
+      "step": 585
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4147146466119235e-05,
+      "loss": 0.7154,
+      "step": 586
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4127873730746706e-05,
+      "loss": 0.7073,
+      "step": 587
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4108582495951077e-05,
+      "loss": 0.7755,
+      "step": 588
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4089272848187678e-05,
+      "loss": 0.7189,
+      "step": 589
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.406994487399437e-05,
+      "loss": 0.754,
+      "step": 590
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4050598659991144e-05,
+      "loss": 0.7286,
+      "step": 591
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4031234292879726e-05,
+      "loss": 0.6891,
+      "step": 592
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.40118518594432e-05,
+      "loss": 0.7654,
+      "step": 593
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3992451446545624e-05,
+      "loss": 0.7363,
+      "step": 594
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.397303314113162e-05,
+      "loss": 0.7878,
+      "step": 595
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3953597030226006e-05,
+      "loss": 0.7313,
+      "step": 596
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3934143200933398e-05,
+      "loss": 0.6866,
+      "step": 597
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3914671740437811e-05,
+      "loss": 0.7035,
+      "step": 598
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3895182736002276e-05,
+      "loss": 0.663,
+      "step": 599
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3875676274968453e-05,
+      "loss": 0.7065,
+      "step": 600
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3856152444756235e-05,
+      "loss": 0.7959,
+      "step": 601
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3836611332863356e-05,
+      "loss": 0.7671,
+      "step": 602
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3817053026865e-05,
+      "loss": 0.7599,
+      "step": 603
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3797477614413404e-05,
+      "loss": 0.7261,
+      "step": 604
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3777885183237473e-05,
+      "loss": 0.7651,
+      "step": 605
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3758275821142382e-05,
+      "loss": 0.7575,
+      "step": 606
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3738649616009183e-05,
+      "loss": 0.7243,
+      "step": 607
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3719006655794414e-05,
+      "loss": 0.7304,
+      "step": 608
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3699347028529694e-05,
+      "loss": 0.7264,
+      "step": 609
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3679670822321347e-05,
+      "loss": 0.7004,
+      "step": 610
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3659978125349991e-05,
+      "loss": 0.7229,
+      "step": 611
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3640269025870152e-05,
+      "loss": 0.6133,
+      "step": 612
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3620543612209861e-05,
+      "loss": 0.7549,
+      "step": 613
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3600801972770272e-05,
+      "loss": 0.7877,
+      "step": 614
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3581044196025243e-05,
+      "loss": 0.7258,
+      "step": 615
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3561270370520957e-05,
+      "loss": 0.7429,
+      "step": 616
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3541480584875531e-05,
+      "loss": 0.7792,
+      "step": 617
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3521674927778594e-05,
+      "loss": 0.717,
+      "step": 618
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3501853487990917e-05,
+      "loss": 0.7192,
+      "step": 619
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.348201635434399e-05,
+      "loss": 0.7398,
+      "step": 620
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3462163615739647e-05,
+      "loss": 0.7234,
+      "step": 621
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3442295361149651e-05,
+      "loss": 0.7678,
+      "step": 622
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3422411679615305e-05,
+      "loss": 0.7264,
+      "step": 623
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3402512660247046e-05,
+      "loss": 0.7632,
+      "step": 624
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3382598392224053e-05,
+      "loss": 0.7015,
+      "step": 625
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.336266896479384e-05,
+      "loss": 0.7232,
+      "step": 626
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3342724467271852e-05,
+      "loss": 0.765,
+      "step": 627
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3322764989041086e-05,
+      "loss": 0.7443,
+      "step": 628
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3302790619551673e-05,
+      "loss": 0.7564,
+      "step": 629
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.328280144832047e-05,
+      "loss": 0.7023,
+      "step": 630
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.326279756493068e-05,
+      "loss": 0.7387,
+      "step": 631
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3242779059031436e-05,
+      "loss": 0.6726,
+      "step": 632
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3222746020337406e-05,
+      "loss": 0.7099,
+      "step": 633
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3202698538628376e-05,
+      "loss": 0.7268,
+      "step": 634
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3182636703748881e-05,
+      "loss": 0.7254,
+      "step": 635
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3162560605607763e-05,
+      "loss": 0.6561,
+      "step": 636
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3142470334177794e-05,
+      "loss": 0.7076,
+      "step": 637
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3122365979495259e-05,
+      "loss": 0.6512,
+      "step": 638
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3102247631659564e-05,
+      "loss": 0.6363,
+      "step": 639
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3082115380832829e-05,
+      "loss": 0.7633,
+      "step": 640
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.306196931723947e-05,
+      "loss": 0.716,
+      "step": 641
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3041809531165819e-05,
+      "loss": 0.7556,
+      "step": 642
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3021636112959696e-05,
+      "loss": 0.7228,
+      "step": 643
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3001449153030024e-05,
+      "loss": 0.756,
+      "step": 644
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2981248741846404e-05,
+      "loss": 0.773,
+      "step": 645
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2961034969938732e-05,
+      "loss": 0.7269,
+      "step": 646
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2940807927896773e-05,
+      "loss": 0.6693,
+      "step": 647
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.292056770636976e-05,
+      "loss": 0.6987,
+      "step": 648
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2900314396066e-05,
+      "loss": 0.6602,
+      "step": 649
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.288004808775246e-05,
+      "loss": 0.7263,
+      "step": 650
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2859768872254344e-05,
+      "loss": 0.7176,
+      "step": 651
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2839476840454714e-05,
+      "loss": 0.7196,
+      "step": 652
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2819172083294068e-05,
+      "loss": 0.727,
+      "step": 653
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2798854691769927e-05,
+      "loss": 0.7524,
+      "step": 654
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2778524756936438e-05,
+      "loss": 0.7204,
+      "step": 655
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2758182369903969e-05,
+      "loss": 0.7483,
+      "step": 656
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2737827621838678e-05,
+      "loss": 0.7504,
+      "step": 657
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2717460603962132e-05,
+      "loss": 0.739,
+      "step": 658
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2697081407550886e-05,
+      "loss": 0.7273,
+      "step": 659
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2676690123936066e-05,
+      "loss": 0.7328,
+      "step": 660
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2656286844502978e-05,
+      "loss": 0.693,
+      "step": 661
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2635871660690677e-05,
+      "loss": 0.7774,
+      "step": 662
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.261544466399158e-05,
+      "loss": 0.7683,
+      "step": 663
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.259500594595104e-05,
+      "loss": 0.7132,
+      "step": 664
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2574555598166932e-05,
+      "loss": 0.7224,
+      "step": 665
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2554093712289267e-05,
+      "loss": 0.7322,
+      "step": 666
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2533620380019752e-05,
+      "loss": 0.7392,
+      "step": 667
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2513135693111399e-05,
+      "loss": 0.7136,
+      "step": 668
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2492639743368096e-05,
+      "loss": 0.7039,
+      "step": 669
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2472132622644222e-05,
+      "loss": 0.7226,
+      "step": 670
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2451614422844208e-05,
+      "loss": 0.7374,
+      "step": 671
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2431085235922143e-05,
+      "loss": 0.7166,
+      "step": 672
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2410545153881355e-05,
+      "loss": 0.7439,
+      "step": 673
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2389994268773995e-05,
+      "loss": 0.7176,
+      "step": 674
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2369432672700634e-05,
+      "loss": 0.7174,
+      "step": 675
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.234886045780984e-05,
+      "loss": 0.7341,
+      "step": 676
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2328277716297771e-05,
+      "loss": 0.7129,
+      "step": 677
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2307684540407775e-05,
+      "loss": 0.7583,
+      "step": 678
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2287081022429935e-05,
+      "loss": 0.7298,
+      "step": 679
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2266467254700712e-05,
+      "loss": 0.7048,
+      "step": 680
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.224584332960248e-05,
+      "loss": 0.7671,
+      "step": 681
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2225209339563144e-05,
+      "loss": 0.7035,
+      "step": 682
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.220456537705572e-05,
+      "loss": 0.7289,
+      "step": 683
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2183911534597908e-05,
+      "loss": 0.7258,
+      "step": 684
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2163247904751688e-05,
+      "loss": 0.7627,
+      "step": 685
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2142574580122903e-05,
+      "loss": 0.6861,
+      "step": 686
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2121891653360851e-05,
+      "loss": 0.6678,
+      "step": 687
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.210119921715785e-05,
+      "loss": 0.7285,
+      "step": 688
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2080497364248849e-05,
+      "loss": 0.7086,
+      "step": 689
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2059786187410984e-05,
+      "loss": 0.6572,
+      "step": 690
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2039065779463195e-05,
+      "loss": 0.7477,
+      "step": 691
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2018336233265775e-05,
+      "loss": 0.7336,
+      "step": 692
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1997597641719976e-05,
+      "loss": 0.6717,
+      "step": 693
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1976850097767598e-05,
+      "loss": 0.7632,
+      "step": 694
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1956093694390547e-05,
+      "loss": 0.6646,
+      "step": 695
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1935328524610443e-05,
+      "loss": 0.7142,
+      "step": 696
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1914554681488188e-05,
+      "loss": 0.7662,
+      "step": 697
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1893772258123554e-05,
+      "loss": 0.6749,
+      "step": 698
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1872981347654778e-05,
+      "loss": 0.6461,
+      "step": 699
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1852182043258112e-05,
+      "loss": 0.7511,
+      "step": 700
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1831374438147445e-05,
+      "loss": 0.7491,
+      "step": 701
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1810558625573856e-05,
+      "loss": 0.7359,
+      "step": 702
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1789734698825209e-05,
+      "loss": 0.7372,
+      "step": 703
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.176890275122573e-05,
+      "loss": 0.7181,
+      "step": 704
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.17480628761356e-05,
+      "loss": 0.7166,
+      "step": 705
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1727215166950519e-05,
+      "loss": 0.7076,
+      "step": 706
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1706359717101295e-05,
+      "loss": 0.6737,
+      "step": 707
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1685496620053434e-05,
+      "loss": 0.7461,
+      "step": 708
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1664625969306707e-05,
+      "loss": 0.7243,
+      "step": 709
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1643747858394743e-05,
+      "loss": 0.6899,
+      "step": 710
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.16228623808846e-05,
+      "loss": 0.8022,
+      "step": 711
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1601969630376353e-05,
+      "loss": 0.7704,
+      "step": 712
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.158106970050267e-05,
+      "loss": 0.7285,
+      "step": 713
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.156016268492839e-05,
+      "loss": 0.6435,
+      "step": 714
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1539248677350114e-05,
+      "loss": 0.7342,
+      "step": 715
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.151832777149578e-05,
+      "loss": 0.7427,
+      "step": 716
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1497400061124231e-05,
+      "loss": 0.6811,
+      "step": 717
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1476465640024814e-05,
+      "loss": 0.7563,
+      "step": 718
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1455524602016946e-05,
+      "loss": 0.6817,
+      "step": 719
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1434577040949699e-05,
+      "loss": 0.7575,
+      "step": 720
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1413623050701381e-05,
+      "loss": 0.7814,
+      "step": 721
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1392662725179114e-05,
+      "loss": 0.7329,
+      "step": 722
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1371696158318404e-05,
+      "loss": 0.7242,
+      "step": 723
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1350723444082736e-05,
+      "loss": 0.722,
+      "step": 724
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1329744676463144e-05,
+      "loss": 0.6673,
+      "step": 725
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1308759949477786e-05,
+      "loss": 0.7053,
+      "step": 726
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1287769357171536e-05,
+      "loss": 0.724,
+      "step": 727
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1266772993615543e-05,
+      "loss": 0.6969,
+      "step": 728
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1245770952906829e-05,
+      "loss": 0.6824,
+      "step": 729
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1224763329167859e-05,
+      "loss": 0.785,
+      "step": 730
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1203750216546105e-05,
+      "loss": 0.6677,
+      "step": 731
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1182731709213658e-05,
+      "loss": 0.7184,
+      "step": 732
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1161707901366776e-05,
+      "loss": 0.736,
+      "step": 733
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1140678887225468e-05,
+      "loss": 0.7286,
+      "step": 734
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1119644761033079e-05,
+      "loss": 0.7466,
+      "step": 735
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1098605617055871e-05,
+      "loss": 0.7381,
+      "step": 736
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1077561549582578e-05,
+      "loss": 0.7431,
+      "step": 737
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1056512652924014e-05,
+      "loss": 0.7195,
+      "step": 738
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1035459021412631e-05,
+      "loss": 0.6959,
+      "step": 739
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1014400749402096e-05,
+      "loss": 0.6758,
+      "step": 740
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0993337931266879e-05,
+      "loss": 0.6869,
+      "step": 741
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0972270661401812e-05,
+      "loss": 0.697,
+      "step": 742
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.09511990342217e-05,
+      "loss": 0.6798,
+      "step": 743
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0930123144160857e-05,
+      "loss": 0.7043,
+      "step": 744
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0909043085672703e-05,
+      "loss": 0.705,
+      "step": 745
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0887958953229349e-05,
+      "loss": 0.6515,
+      "step": 746
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0866870841321156e-05,
+      "loss": 0.6996,
+      "step": 747
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0845778844456319e-05,
+      "loss": 0.6817,
+      "step": 748
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0824683057160445e-05,
+      "loss": 0.68,
+      "step": 749
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0803583573976137e-05,
+      "loss": 0.6627,
+      "step": 750
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0782480489462546e-05,
+      "loss": 0.6957,
+      "step": 751
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0761373898194972e-05,
+      "loss": 0.6439,
+      "step": 752
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.074026389476443e-05,
+      "loss": 0.7282,
+      "step": 753
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0719150573777226e-05,
+      "loss": 0.717,
+      "step": 754
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0698034029854533e-05,
+      "loss": 0.7458,
+      "step": 755
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0676914357631968e-05,
+      "loss": 0.6983,
+      "step": 756
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0655791651759174e-05,
+      "loss": 0.7865,
+      "step": 757
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0634666006899375e-05,
+      "loss": 0.7022,
+      "step": 758
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.061353751772898e-05,
+      "loss": 0.6665,
+      "step": 759
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0592406278937143e-05,
+      "loss": 0.6883,
+      "step": 760
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0571272385225338e-05,
+      "loss": 0.7024,
+      "step": 761
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.055013593130693e-05,
+      "loss": 0.7366,
+      "step": 762
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0528997011906775e-05,
+      "loss": 0.6668,
+      "step": 763
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0507855721760757e-05,
+      "loss": 0.6919,
+      "step": 764
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0486712155615408e-05,
+      "loss": 0.6995,
+      "step": 765
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.046556640822744e-05,
+      "loss": 0.707,
+      "step": 766
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0444418574363354e-05,
+      "loss": 0.6681,
+      "step": 767
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0423268748798992e-05,
+      "loss": 0.7561,
+      "step": 768
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0402117026319128e-05,
+      "loss": 0.6574,
+      "step": 769
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0380963501717034e-05,
+      "loss": 0.7075,
+      "step": 770
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0359808269794063e-05,
+      "loss": 0.7032,
+      "step": 771
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0338651425359217e-05,
+      "loss": 0.7179,
+      "step": 772
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0317493063228716e-05,
+      "loss": 0.6361,
+      "step": 773
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0296333278225599e-05,
+      "loss": 0.6996,
+      "step": 774
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0275172165179268e-05,
+      "loss": 0.7163,
+      "step": 775
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0254009818925082e-05,
+      "loss": 0.7143,
+      "step": 776
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0232846334303925e-05,
+      "loss": 0.6926,
+      "step": 777
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0211681806161787e-05,
+      "loss": 0.6692,
+      "step": 778
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0190516329349328e-05,
+      "loss": 0.7666,
+      "step": 779
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0169349998721459e-05,
+      "loss": 0.6846,
+      "step": 780
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0148182909136928e-05,
+      "loss": 0.7338,
+      "step": 781
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0127015155457875e-05,
+      "loss": 0.7034,
+      "step": 782
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0105846832549417e-05,
+      "loss": 0.6797,
+      "step": 783
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0084678035279227e-05,
+      "loss": 0.6752,
+      "step": 784
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.00635088585171e-05,
+      "loss": 0.7522,
+      "step": 785
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0042339397134528e-05,
+      "loss": 0.7096,
+      "step": 786
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0021169746004287e-05,
+      "loss": 0.7339,
+      "step": 787
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1e-05,
+      "loss": 0.7047,
+      "step": 788
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.978830253995713e-06,
+      "loss": 0.7288,
+      "step": 789
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.957660602865477e-06,
+      "loss": 0.6157,
+      "step": 790
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.936491141482902e-06,
+      "loss": 0.764,
+      "step": 791
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.915321964720773e-06,
+      "loss": 0.5981,
+      "step": 792
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.894153167450586e-06,
+      "loss": 0.6801,
+      "step": 793
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.872984844542128e-06,
+      "loss": 0.6251,
+      "step": 794
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.851817090863073e-06,
+      "loss": 0.7278,
+      "step": 795
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.830650001278543e-06,
+      "loss": 0.682,
+      "step": 796
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.809483670650676e-06,
+      "loss": 0.7366,
+      "step": 797
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.788318193838218e-06,
+      "loss": 0.7158,
+      "step": 798
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.767153665696077e-06,
+      "loss": 0.6826,
+      "step": 799
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.74599018107492e-06,
+      "loss": 0.7243,
+      "step": 800
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.724827834820736e-06,
+      "loss": 0.6767,
+      "step": 801
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.703666721774403e-06,
+      "loss": 0.6239,
+      "step": 802
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.682506936771284e-06,
+      "loss": 0.6875,
+      "step": 803
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.66134857464079e-06,
+      "loss": 0.7074,
+      "step": 804
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.64019173020594e-06,
+      "loss": 0.751,
+      "step": 805
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.619036498282968e-06,
+      "loss": 0.739,
+      "step": 806
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.597882973680877e-06,
+      "loss": 0.7522,
+      "step": 807
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.57673125120101e-06,
+      "loss": 0.6697,
+      "step": 808
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.555581425636648e-06,
+      "loss": 0.7064,
+      "step": 809
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.534433591772562e-06,
+      "loss": 0.741,
+      "step": 810
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.513287844384594e-06,
+      "loss": 0.727,
+      "step": 811
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.492144278239244e-06,
+      "loss": 0.7944,
+      "step": 812
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.47100298809323e-06,
+      "loss": 0.6636,
+      "step": 813
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.449864068693072e-06,
+      "loss": 0.652,
+      "step": 814
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.428727614774667e-06,
+      "loss": 0.6426,
+      "step": 815
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.407593721062858e-06,
+      "loss": 0.687,
+      "step": 816
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.38646248227102e-06,
+      "loss": 0.7082,
+      "step": 817
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.365333993100628e-06,
+      "loss": 0.6977,
+      "step": 818
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.34420834824083e-06,
+      "loss": 0.688,
+      "step": 819
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.323085642368032e-06,
+      "loss": 0.6521,
+      "step": 820
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.301965970145472e-06,
+      "loss": 0.6808,
+      "step": 821
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.280849426222778e-06,
+      "loss": 0.7322,
+      "step": 822
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.259736105235572e-06,
+      "loss": 0.6768,
+      "step": 823
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.23862610180503e-06,
+      "loss": 0.694,
+      "step": 824
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.217519510537455e-06,
+      "loss": 0.6836,
+      "step": 825
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.196416426023868e-06,
+      "loss": 0.6455,
+      "step": 826
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.175316942839557e-06,
+      "loss": 0.7085,
+      "step": 827
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.154221155543684e-06,
+      "loss": 0.7161,
+      "step": 828
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.133129158678848e-06,
+      "loss": 0.7367,
+      "step": 829
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.112041046770653e-06,
+      "loss": 0.7377,
+      "step": 830
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.090956914327297e-06,
+      "loss": 0.7225,
+      "step": 831
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.069876855839148e-06,
+      "loss": 0.7198,
+      "step": 832
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.048800965778304e-06,
+      "loss": 0.7302,
+      "step": 833
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.027729338598188e-06,
+      "loss": 0.7122,
+      "step": 834
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.006662068733128e-06,
+      "loss": 0.6355,
+      "step": 835
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.985599250597906e-06,
+      "loss": 0.6603,
+      "step": 836
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.96454097858737e-06,
+      "loss": 0.7309,
+      "step": 837
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.943487347075988e-06,
+      "loss": 0.7053,
+      "step": 838
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.922438450417424e-06,
+      "loss": 0.6807,
+      "step": 839
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.901394382944135e-06,
+      "loss": 0.6972,
+      "step": 840
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.880355238966923e-06,
+      "loss": 0.6499,
+      "step": 841
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.859321112774535e-06,
+      "loss": 0.6798,
+      "step": 842
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.838292098633229e-06,
+      "loss": 0.7812,
+      "step": 843
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.817268290786343e-06,
+      "loss": 0.6775,
+      "step": 844
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.796249783453897e-06,
+      "loss": 0.6989,
+      "step": 845
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.775236670832146e-06,
+      "loss": 0.713,
+      "step": 846
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.754229047093173e-06,
+      "loss": 0.6839,
+      "step": 847
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.733227006384459e-06,
+      "loss": 0.7053,
+      "step": 848
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.71223064282847e-06,
+      "loss": 0.6516,
+      "step": 849
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.691240050522215e-06,
+      "loss": 0.7223,
+      "step": 850
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.670255323536858e-06,
+      "loss": 0.7051,
+      "step": 851
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.649276555917267e-06,
+      "loss": 0.7047,
+      "step": 852
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.6283038416816e-06,
+      "loss": 0.6676,
+      "step": 853
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.607337274820888e-06,
+      "loss": 0.7052,
+      "step": 854
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.586376949298622e-06,
+      "loss": 0.732,
+      "step": 855
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.565422959050303e-06,
+      "loss": 0.6585,
+      "step": 856
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.54447539798306e-06,
+      "loss": 0.7222,
+      "step": 857
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.52353435997519e-06,
+      "loss": 0.7249,
+      "step": 858
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.50259993887577e-06,
+      "loss": 0.7258,
+      "step": 859
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.481672228504224e-06,
+      "loss": 0.6656,
+      "step": 860
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.460751322649887e-06,
+      "loss": 0.6192,
+      "step": 861
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.439837315071612e-06,
+      "loss": 0.7369,
+      "step": 862
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.418930299497335e-06,
+      "loss": 0.6739,
+      "step": 863
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.39803036962365e-06,
+      "loss": 0.6251,
+      "step": 864
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.377137619115401e-06,
+      "loss": 0.7312,
+      "step": 865
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.35625214160526e-06,
+      "loss": 0.6744,
+      "step": 866
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.335374030693295e-06,
+      "loss": 0.7199,
+      "step": 867
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.314503379946569e-06,
+      "loss": 0.6968,
+      "step": 868
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.293640282898709e-06,
+      "loss": 0.6898,
+      "step": 869
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.272784833049485e-06,
+      "loss": 0.653,
+      "step": 870
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.251937123864405e-06,
+      "loss": 0.7179,
+      "step": 871
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.231097248774273e-06,
+      "loss": 0.6474,
+      "step": 872
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.210265301174796e-06,
+      "loss": 0.6794,
+      "step": 873
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.18944137442615e-06,
+      "loss": 0.7214,
+      "step": 874
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.16862556185256e-06,
+      "loss": 0.6832,
+      "step": 875
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.14781795674189e-06,
+      "loss": 0.6768,
+      "step": 876
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.12701865234523e-06,
+      "loss": 0.7436,
+      "step": 877
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.106227741876447e-06,
+      "loss": 0.7026,
+      "step": 878
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.085445318511814e-06,
+      "loss": 0.6797,
+      "step": 879
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.064671475389562e-06,
+      "loss": 0.6304,
+      "step": 880
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.043906305609454e-06,
+      "loss": 0.6498,
+      "step": 881
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.023149902232404e-06,
+      "loss": 0.7137,
+      "step": 882
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.002402358280028e-06,
+      "loss": 0.7033,
+      "step": 883
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.981663766734228e-06,
+      "loss": 0.6866,
+      "step": 884
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.960934220536812e-06,
+      "loss": 0.6916,
+      "step": 885
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.940213812589018e-06,
+      "loss": 0.6677,
+      "step": 886
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.919502635751154e-06,
+      "loss": 0.6322,
+      "step": 887
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.898800782842153e-06,
+      "loss": 0.6705,
+      "step": 888
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.878108346639152e-06,
+      "loss": 0.6155,
+      "step": 889
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.857425419877097e-06,
+      "loss": 0.6893,
+      "step": 890
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.836752095248318e-06,
+      "loss": 0.6839,
+      "step": 891
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.816088465402097e-06,
+      "loss": 0.7049,
+      "step": 892
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.795434622944282e-06,
+      "loss": 0.7028,
+      "step": 893
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.774790660436857e-06,
+      "loss": 0.685,
+      "step": 894
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.754156670397522e-06,
+      "loss": 0.6787,
+      "step": 895
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.73353274529929e-06,
+      "loss": 0.6646,
+      "step": 896
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.712918977570067e-06,
+      "loss": 0.6078,
+      "step": 897
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.69231545959223e-06,
+      "loss": 0.6826,
+      "step": 898
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.671722283702227e-06,
+      "loss": 0.6507,
+      "step": 899
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.651139542190164e-06,
+      "loss": 0.6907,
+      "step": 900
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.63056732729937e-06,
+      "loss": 0.7883,
+      "step": 901
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.610005731226009e-06,
+      "loss": 0.6495,
+      "step": 902
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.589454846118648e-06,
+      "loss": 0.6702,
+      "step": 903
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.568914764077857e-06,
+      "loss": 0.7069,
+      "step": 904
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.548385577155795e-06,
+      "loss": 0.6726,
+      "step": 905
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.52786737735578e-06,
+      "loss": 0.7035,
+      "step": 906
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5073602566319045e-06,
+      "loss": 0.6734,
+      "step": 907
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.486864306888608e-06,
+      "loss": 0.693,
+      "step": 908
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.466379619980251e-06,
+      "loss": 0.6852,
+      "step": 909
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.445906287710733e-06,
+      "loss": 0.6243,
+      "step": 910
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.4254444018330685e-06,
+      "loss": 0.7032,
+      "step": 911
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.404994054048962e-06,
+      "loss": 0.7084,
+      "step": 912
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.384555336008419e-06,
+      "loss": 0.7557,
+      "step": 913
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.364128339309326e-06,
+      "loss": 0.7065,
+      "step": 914
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3437131554970254e-06,
+      "loss": 0.7313,
+      "step": 915
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.323309876063936e-06,
+      "loss": 0.6442,
+      "step": 916
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.302918592449116e-06,
+      "loss": 0.717,
+      "step": 917
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.282539396037868e-06,
+      "loss": 0.6825,
+      "step": 918
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.262172378161326e-06,
+      "loss": 0.724,
+      "step": 919
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.241817630096035e-06,
+      "loss": 0.6666,
+      "step": 920
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2214752430635625e-06,
+      "loss": 0.7376,
+      "step": 921
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.201145308230075e-06,
+      "loss": 0.6905,
+      "step": 922
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.180827916705935e-06,
+      "loss": 0.6808,
+      "step": 923
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.160523159545285e-06,
+      "loss": 0.6739,
+      "step": 924
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.140231127745661e-06,
+      "loss": 0.7001,
+      "step": 925
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.119951912247545e-06,
+      "loss": 0.6609,
+      "step": 926
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.099685603934e-06,
+      "loss": 0.6754,
+      "step": 927
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.079432293630244e-06,
+      "loss": 0.7045,
+      "step": 928
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0591920721032314e-06,
+      "loss": 0.711,
+      "step": 929
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.038965030061273e-06,
+      "loss": 0.6554,
+      "step": 930
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.018751258153599e-06,
+      "loss": 0.653,
+      "step": 931
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.998550846969981e-06,
+      "loss": 0.6283,
+      "step": 932
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.978363887040308e-06,
+      "loss": 0.7227,
+      "step": 933
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9581904688341854e-06,
+      "loss": 0.6729,
+      "step": 934
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.938030682760532e-06,
+      "loss": 0.6725,
+      "step": 935
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.917884619167177e-06,
+      "loss": 0.6944,
+      "step": 936
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.897752368340438e-06,
+      "loss": 0.674,
+      "step": 937
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8776340205047446e-06,
+      "loss": 0.6894,
+      "step": 938
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.857529665822211e-06,
+      "loss": 0.7036,
+      "step": 939
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.837439394392238e-06,
+      "loss": 0.7079,
+      "step": 940
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.817363296251119e-06,
+      "loss": 0.7317,
+      "step": 941
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.797301461371626e-06,
+      "loss": 0.6547,
+      "step": 942
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7772539796625995e-06,
+      "loss": 0.66,
+      "step": 943
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.757220940968567e-06,
+      "loss": 0.7137,
+      "step": 944
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.737202435069323e-06,
+      "loss": 0.7429,
+      "step": 945
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7171985516795315e-06,
+      "loss": 0.7124,
+      "step": 946
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.697209380448333e-06,
+      "loss": 0.6986,
+      "step": 947
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.677235010958916e-06,
+      "loss": 0.7411,
+      "step": 948
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6572755327281506e-06,
+      "loss": 0.7564,
+      "step": 949
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.637331035206166e-06,
+      "loss": 0.6841,
+      "step": 950
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.617401607775948e-06,
+      "loss": 0.6151,
+      "step": 951
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.597487339752953e-06,
+      "loss": 0.6966,
+      "step": 952
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.577588320384698e-06,
+      "loss": 0.6928,
+      "step": 953
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.557704638850352e-06,
+      "loss": 0.6826,
+      "step": 954
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.537836384260356e-06,
+      "loss": 0.6567,
+      "step": 955
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.517983645656014e-06,
+      "loss": 0.6756,
+      "step": 956
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.498146512009087e-06,
+      "loss": 0.7215,
+      "step": 957
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4783250722214066e-06,
+      "loss": 0.6939,
+      "step": 958
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.458519415124473e-06,
+      "loss": 0.6799,
+      "step": 959
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.438729629479045e-06,
+      "loss": 0.6584,
+      "step": 960
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.418955803974763e-06,
+      "loss": 0.7723,
+      "step": 961
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.399198027229732e-06,
+      "loss": 0.6232,
+      "step": 962
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3794563877901385e-06,
+      "loss": 0.6624,
+      "step": 963
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.359730974129852e-06,
+      "loss": 0.7008,
+      "step": 964
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.340021874650013e-06,
+      "loss": 0.6804,
+      "step": 965
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.320329177678656e-06,
+      "loss": 0.6662,
+      "step": 966
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.300652971470309e-06,
+      "loss": 0.7117,
+      "step": 967
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.28099334420559e-06,
+      "loss": 0.6513,
+      "step": 968
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.261350383990816e-06,
+      "loss": 0.6429,
+      "step": 969
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.241724178857621e-06,
+      "loss": 0.6465,
+      "step": 970
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.22211481676253e-06,
+      "loss": 0.6981,
+      "step": 971
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.202522385586595e-06,
+      "loss": 0.7445,
+      "step": 972
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.182946973135004e-06,
+      "loss": 0.6609,
+      "step": 973
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.163388667136646e-06,
+      "loss": 0.6613,
+      "step": 974
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.14384755524377e-06,
+      "loss": 0.6668,
+      "step": 975
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.12432372503155e-06,
+      "loss": 0.7201,
+      "step": 976
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.104817263997727e-06,
+      "loss": 0.6761,
+      "step": 977
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.085328259562195e-06,
+      "loss": 0.6539,
+      "step": 978
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.065856799066604e-06,
+      "loss": 0.7445,
+      "step": 979
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.046402969773992e-06,
+      "loss": 0.6962,
+      "step": 980
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0269668588683815e-06,
+      "loss": 0.7402,
+      "step": 981
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.007548553454379e-06,
+      "loss": 0.6802,
+      "step": 982
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9881481405568005e-06,
+      "loss": 0.6804,
+      "step": 983
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.96876570712028e-06,
+      "loss": 0.6454,
+      "step": 984
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.94940134000886e-06,
+      "loss": 0.6647,
+      "step": 985
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.93005512600563e-06,
+      "loss": 0.5945,
+      "step": 986
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.910727151812324e-06,
+      "loss": 0.6822,
+      "step": 987
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.891417504048926e-06,
+      "loss": 0.6842,
+      "step": 988
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.872126269253298e-06,
+      "loss": 0.6594,
+      "step": 989
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.852853533880768e-06,
+      "loss": 0.6616,
+      "step": 990
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8335993843037695e-06,
+      "loss": 0.7044,
+      "step": 991
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.8143639068114465e-06,
+      "loss": 0.7138,
+      "step": 992
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7951471876092445e-06,
+      "loss": 0.7043,
+      "step": 993
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7759493128185584e-06,
+      "loss": 0.6993,
+      "step": 994
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.756770368476326e-06,
+      "loss": 0.6703,
+      "step": 995
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7376104405346376e-06,
+      "loss": 0.6855,
+      "step": 996
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7184696148603725e-06,
+      "loss": 0.663,
+      "step": 997
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.699347977234799e-06,
+      "loss": 0.6208,
+      "step": 998
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.680245613353181e-06,
+      "loss": 0.6145,
+      "step": 999
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.66116260882442e-06,
+      "loss": 0.6527,
+      "step": 1000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.642099049170651e-06,
+      "loss": 0.697,
+      "step": 1001
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.623055019826862e-06,
+      "loss": 0.7267,
+      "step": 1002
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.604030606140522e-06,
+      "loss": 0.7182,
+      "step": 1003
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.585025893371187e-06,
+      "loss": 0.6791,
+      "step": 1004
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.566040966690115e-06,
+      "loss": 0.6514,
+      "step": 1005
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.547075911179902e-06,
+      "loss": 0.6766,
+      "step": 1006
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.528130811834081e-06,
+      "loss": 0.6695,
+      "step": 1007
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.50920575355675e-06,
+      "loss": 0.6441,
+      "step": 1008
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.490300821162204e-06,
+      "loss": 0.7135,
+      "step": 1009
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.471416099374525e-06,
+      "loss": 0.6637,
+      "step": 1010
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.452551672827227e-06,
+      "loss": 0.7493,
+      "step": 1011
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.43370762606287e-06,
+      "loss": 0.6633,
+      "step": 1012
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.41488404353268e-06,
+      "loss": 0.6803,
+      "step": 1013
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3960810095961705e-06,
+      "loss": 0.7045,
+      "step": 1014
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.377298608520767e-06,
+      "loss": 0.6783,
+      "step": 1015
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.358536924481416e-06,
+      "loss": 0.7126,
+      "step": 1016
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.339796041560233e-06,
+      "loss": 0.6235,
+      "step": 1017
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.321076043746108e-06,
+      "loss": 0.6988,
+      "step": 1018
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.302377014934322e-06,
+      "loss": 0.6892,
+      "step": 1019
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.283699038926199e-06,
+      "loss": 0.6588,
+      "step": 1020
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.265042199428693e-06,
+      "loss": 0.5018,
+      "step": 1021
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.246406580054051e-06,
+      "loss": 0.5078,
+      "step": 1022
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.227792264319411e-06,
+      "loss": 0.4795,
+      "step": 1023
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.20919933564644e-06,
+      "loss": 0.468,
+      "step": 1024
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.190627877360953e-06,
+      "loss": 0.4694,
+      "step": 1025
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.172077972692553e-06,
+      "loss": 0.5019,
+      "step": 1026
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.153549704774232e-06,
+      "loss": 0.4863,
+      "step": 1027
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.1350431566420326e-06,
+      "loss": 0.4599,
+      "step": 1028
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.116558411234651e-06,
+      "loss": 0.4987,
+      "step": 1029
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.098095551393066e-06,
+      "loss": 0.492,
+      "step": 1030
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.079654659860183e-06,
+      "loss": 0.5082,
+      "step": 1031
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.061235819280456e-06,
+      "loss": 0.5039,
+      "step": 1032
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.042839112199509e-06,
+      "loss": 0.4771,
+      "step": 1033
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.024464621063773e-06,
+      "loss": 0.4692,
+      "step": 1034
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.006112428220124e-06,
+      "loss": 0.5044,
+      "step": 1035
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9877826159154995e-06,
+      "loss": 0.5087,
+      "step": 1036
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9694752662965415e-06,
+      "loss": 0.4798,
+      "step": 1037
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.951190461409214e-06,
+      "loss": 0.5208,
+      "step": 1038
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.932928283198454e-06,
+      "loss": 0.4939,
+      "step": 1039
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.914688813507798e-06,
+      "loss": 0.4257,
+      "step": 1040
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.896472134078998e-06,
+      "loss": 0.4926,
+      "step": 1041
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.878278326551682e-06,
+      "loss": 0.4741,
+      "step": 1042
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.860107472462976e-06,
+      "loss": 0.4516,
+      "step": 1043
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8419596532471266e-06,
+      "loss": 0.4315,
+      "step": 1044
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.823834950235158e-06,
+      "loss": 0.4943,
+      "step": 1045
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.805733444654496e-06,
+      "loss": 0.5163,
+      "step": 1046
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.7876552176286005e-06,
+      "loss": 0.4338,
+      "step": 1047
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.7696003501766155e-06,
+      "loss": 0.4369,
+      "step": 1048
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.751568923212983e-06,
+      "loss": 0.4749,
+      "step": 1049
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.733561017547104e-06,
+      "loss": 0.4832,
+      "step": 1050
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.715576713882971e-06,
+      "loss": 0.4645,
+      "step": 1051
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.697616092818789e-06,
+      "loss": 0.4244,
+      "step": 1052
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.679679234846636e-06,
+      "loss": 0.462,
+      "step": 1053
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.661766220352098e-06,
+      "loss": 0.4808,
+      "step": 1054
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.643877129613889e-06,
+      "loss": 0.4819,
+      "step": 1055
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.626012042803521e-06,
+      "loss": 0.437,
+      "step": 1056
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.608171039984928e-06,
+      "loss": 0.4952,
+      "step": 1057
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.590354201114103e-06,
+      "loss": 0.4745,
+      "step": 1058
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5725616060387525e-06,
+      "loss": 0.502,
+      "step": 1059
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.554793334497933e-06,
+      "loss": 0.4569,
+      "step": 1060
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5370494661216835e-06,
+      "loss": 0.4782,
+      "step": 1061
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.519330080430687e-06,
+      "loss": 0.5223,
+      "step": 1062
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.501635256835908e-06,
+      "loss": 0.4394,
+      "step": 1063
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.483965074638219e-06,
+      "loss": 0.412,
+      "step": 1064
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.46631961302808e-06,
+      "loss": 0.4732,
+      "step": 1065
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.448698951085143e-06,
+      "loss": 0.4879,
+      "step": 1066
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.43110316777793e-06,
+      "loss": 0.4409,
+      "step": 1067
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.413532341963477e-06,
+      "loss": 0.4812,
+      "step": 1068
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.395986552386947e-06,
+      "loss": 0.4723,
+      "step": 1069
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.378465877681317e-06,
+      "loss": 0.4799,
+      "step": 1070
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.360970396367012e-06,
+      "loss": 0.4629,
+      "step": 1071
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.343500186851534e-06,
+      "loss": 0.4704,
+      "step": 1072
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.326055327429143e-06,
+      "loss": 0.5018,
+      "step": 1073
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.3086358962804885e-06,
+      "loss": 0.5217,
+      "step": 1074
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.29124197147225e-06,
+      "loss": 0.5263,
+      "step": 1075
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.273873630956811e-06,
+      "loss": 0.4815,
+      "step": 1076
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.256530952571889e-06,
+      "loss": 0.5306,
+      "step": 1077
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2392140140401996e-06,
+      "loss": 0.476,
+      "step": 1078
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.221922892969103e-06,
+      "loss": 0.4927,
+      "step": 1079
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.204657666850245e-06,
+      "loss": 0.4636,
+      "step": 1080
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.187418413059234e-06,
+      "loss": 0.4632,
+      "step": 1081
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.170205208855281e-06,
+      "loss": 0.4789,
+      "step": 1082
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.153018131380843e-06,
+      "loss": 0.4737,
+      "step": 1083
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.135857257661295e-06,
+      "loss": 0.4863,
+      "step": 1084
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.118722664604582e-06,
+      "loss": 0.4595,
+      "step": 1085
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.101614429000857e-06,
+      "loss": 0.4512,
+      "step": 1086
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.08453262752216e-06,
+      "loss": 0.4231,
+      "step": 1087
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.067477336722063e-06,
+      "loss": 0.4239,
+      "step": 1088
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.050448633035326e-06,
+      "loss": 0.4831,
+      "step": 1089
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.033446592777558e-06,
+      "loss": 0.4813,
+      "step": 1090
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.016471292144875e-06,
+      "loss": 0.4585,
+      "step": 1091
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.999522807213549e-06,
+      "loss": 0.5003,
+      "step": 1092
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.982601213939686e-06,
+      "loss": 0.5151,
+      "step": 1093
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.965706588158865e-06,
+      "loss": 0.4675,
+      "step": 1094
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.948839005585815e-06,
+      "loss": 0.4688,
+      "step": 1095
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.931998541814069e-06,
+      "loss": 0.446,
+      "step": 1096
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.915185272315616e-06,
+      "loss": 0.4461,
+      "step": 1097
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.89839927244058e-06,
+      "loss": 0.4324,
+      "step": 1098
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.8816406174168695e-06,
+      "loss": 0.511,
+      "step": 1099
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.86490938234985e-06,
+      "loss": 0.4729,
+      "step": 1100
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.848205642221995e-06,
+      "loss": 0.4753,
+      "step": 1101
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8315294718925656e-06,
+      "loss": 0.4349,
+      "step": 1102
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.814880946097253e-06,
+      "loss": 0.431,
+      "step": 1103
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.79826013944787e-06,
+      "loss": 0.4125,
+      "step": 1104
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7816671264320026e-06,
+      "loss": 0.449,
+      "step": 1105
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7651019814126656e-06,
+      "loss": 0.4941,
+      "step": 1106
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.748564778627993e-06,
+      "loss": 0.4352,
+      "step": 1107
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.732055592190893e-06,
+      "loss": 0.5243,
+      "step": 1108
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7155744960887054e-06,
+      "loss": 0.4816,
+      "step": 1109
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6991215641828903e-06,
+      "loss": 0.474,
+      "step": 1110
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.682696870208685e-06,
+      "loss": 0.4866,
+      "step": 1111
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6663004877747765e-06,
+      "loss": 0.4757,
+      "step": 1112
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.649932490362974e-06,
+      "loss": 0.4751,
+      "step": 1113
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6335929513278667e-06,
+      "loss": 0.4621,
+      "step": 1114
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6172819438965157e-06,
+      "loss": 0.4522,
+      "step": 1115
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.600999541168113e-06,
+      "loss": 0.4517,
+      "step": 1116
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5847458161136485e-06,
+      "loss": 0.5072,
+      "step": 1117
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.568520841575601e-06,
+      "loss": 0.4187,
+      "step": 1118
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5523246902676e-06,
+      "loss": 0.5158,
+      "step": 1119
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5361574347740912e-06,
+      "loss": 0.5174,
+      "step": 1120
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5200191475500313e-06,
+      "loss": 0.5027,
+      "step": 1121
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5039099009205503e-06,
+      "loss": 0.5001,
+      "step": 1122
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4878297670806303e-06,
+      "loss": 0.4519,
+      "step": 1123
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4717788180947855e-06,
+      "loss": 0.4456,
+      "step": 1124
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.455757125896725e-06,
+      "loss": 0.4357,
+      "step": 1125
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.439764762289051e-06,
+      "loss": 0.5258,
+      "step": 1126
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.423801798942926e-06,
+      "loss": 0.485,
+      "step": 1127
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.407868307397747e-06,
+      "loss": 0.4596,
+      "step": 1128
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3919643590608355e-06,
+      "loss": 0.4605,
+      "step": 1129
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.376090025207115e-06,
+      "loss": 0.527,
+      "step": 1130
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.360245376978779e-06,
+      "loss": 0.5222,
+      "step": 1131
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3444304853849918e-06,
+      "loss": 0.4802,
+      "step": 1132
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3286454213015582e-06,
+      "loss": 0.4119,
+      "step": 1133
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.312890255470609e-06,
+      "loss": 0.4704,
+      "step": 1134
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.29716505850028e-06,
+      "loss": 0.4777,
+      "step": 1135
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2814699008644068e-06,
+      "loss": 0.5015,
+      "step": 1136
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.265804852902188e-06,
+      "loss": 0.505,
+      "step": 1137
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.250169984817897e-06,
+      "loss": 0.4987,
+      "step": 1138
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.234565366680542e-06,
+      "loss": 0.5072,
+      "step": 1139
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2189910684235705e-06,
+      "loss": 0.4286,
+      "step": 1140
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2034471598445505e-06,
+      "loss": 0.4485,
+      "step": 1141
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.187933710604847e-06,
+      "loss": 0.4342,
+      "step": 1142
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1724507902293267e-06,
+      "loss": 0.4417,
+      "step": 1143
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1569984681060375e-06,
+      "loss": 0.4898,
+      "step": 1144
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1415768134858947e-06,
+      "loss": 0.4993,
+      "step": 1145
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1261858954823798e-06,
+      "loss": 0.4508,
+      "step": 1146
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1108257830712253e-06,
+      "loss": 0.4708,
+      "step": 1147
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0954965450900963e-06,
+      "loss": 0.4894,
+      "step": 1148
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.080198250238302e-06,
+      "loss": 0.4292,
+      "step": 1149
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.064930967076477e-06,
+      "loss": 0.4891,
+      "step": 1150
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0496947640262655e-06,
+      "loss": 0.4543,
+      "step": 1151
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0344897093700333e-06,
+      "loss": 0.5044,
+      "step": 1152
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.019315871250543e-06,
+      "loss": 0.4832,
+      "step": 1153
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0041733176706668e-06,
+      "loss": 0.5226,
+      "step": 1154
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9890621164930656e-06,
+      "loss": 0.4874,
+      "step": 1155
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9739823354398958e-06,
+      "loss": 0.4382,
+      "step": 1156
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9589340420924983e-06,
+      "loss": 0.4743,
+      "step": 1157
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.943917303891107e-06,
+      "loss": 0.4667,
+      "step": 1158
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9289321881345257e-06,
+      "loss": 0.465,
+      "step": 1159
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.913978761979849e-06,
+      "loss": 0.4455,
+      "step": 1160
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8990570924421526e-06,
+      "loss": 0.4802,
+      "step": 1161
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8841672463941827e-06,
+      "loss": 0.4512,
+      "step": 1162
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8693092905660757e-06,
+      "loss": 0.4306,
+      "step": 1163
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.854483291545046e-06,
+      "loss": 0.5132,
+      "step": 1164
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8396893157750893e-06,
+      "loss": 0.513,
+      "step": 1165
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8249274295566863e-06,
+      "loss": 0.427,
+      "step": 1166
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.810197699046513e-06,
+      "loss": 0.545,
+      "step": 1167
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.795500190257122e-06,
+      "loss": 0.4858,
+      "step": 1168
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.780834969056676e-06,
+      "loss": 0.4732,
+      "step": 1169
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.766202101168628e-06,
+      "loss": 0.5097,
+      "step": 1170
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7516016521714417e-06,
+      "loss": 0.5365,
+      "step": 1171
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7370336874982952e-06,
+      "loss": 0.478,
+      "step": 1172
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7224982724367775e-06,
+      "loss": 0.5304,
+      "step": 1173
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7079954721286108e-06,
+      "loss": 0.4352,
+      "step": 1174
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.693525351569347e-06,
+      "loss": 0.4965,
+      "step": 1175
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6790879756080845e-06,
+      "loss": 0.468,
+      "step": 1176
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.664683408947173e-06,
+      "loss": 0.4592,
+      "step": 1177
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6503117161419246e-06,
+      "loss": 0.4686,
+      "step": 1178
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6359729616003184e-06,
+      "loss": 0.4987,
+      "step": 1179
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6216672095827267e-06,
+      "loss": 0.4414,
+      "step": 1180
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.607394524201616e-06,
+      "loss": 0.5126,
+      "step": 1181
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5931549694212545e-06,
+      "loss": 0.479,
+      "step": 1182
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.578948609057446e-06,
+      "loss": 0.495,
+      "step": 1183
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.564775506777214e-06,
+      "loss": 0.4846,
+      "step": 1184
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.550635726098547e-06,
+      "loss": 0.4961,
+      "step": 1185
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.536529330390095e-06,
+      "loss": 0.4479,
+      "step": 1186
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5224563828708904e-06,
+      "loss": 0.4847,
+      "step": 1187
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5084169466100626e-06,
+      "loss": 0.4822,
+      "step": 1188
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4944110845265646e-06,
+      "loss": 0.4491,
+      "step": 1189
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.480438859388873e-06,
+      "loss": 0.3964,
+      "step": 1190
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.466500333814723e-06,
+      "loss": 0.4545,
+      "step": 1191
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.452595570270828e-06,
+      "loss": 0.5149,
+      "step": 1192
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.438724631072581e-06,
+      "loss": 0.4858,
+      "step": 1193
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.424887578383799e-06,
+      "loss": 0.4764,
+      "step": 1194
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.411084474216434e-06,
+      "loss": 0.4876,
+      "step": 1195
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.397315380430284e-06,
+      "loss": 0.4595,
+      "step": 1196
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3835803587327365e-06,
+      "loss": 0.4792,
+      "step": 1197
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.36987947067848e-06,
+      "loss": 0.447,
+      "step": 1198
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.356212777669229e-06,
+      "loss": 0.4934,
+      "step": 1199
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.342580340953451e-06,
+      "loss": 0.4825,
+      "step": 1200
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.328982221626087e-06,
+      "loss": 0.4789,
+      "step": 1201
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3154184806282863e-06,
+      "loss": 0.4874,
+      "step": 1202
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3018891787471308e-06,
+      "loss": 0.5086,
+      "step": 1203
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2883943766153503e-06,
+      "loss": 0.4656,
+      "step": 1204
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2749341347110697e-06,
+      "loss": 0.4276,
+      "step": 1205
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.261508513357532e-06,
+      "loss": 0.4495,
+      "step": 1206
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2481175727228142e-06,
+      "loss": 0.4449,
+      "step": 1207
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.234761372819577e-06,
+      "loss": 0.4749,
+      "step": 1208
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2214399735047856e-06,
+      "loss": 0.4182,
+      "step": 1209
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.208153434479442e-06,
+      "loss": 0.4718,
+      "step": 1210
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.194901815288324e-06,
+      "loss": 0.4782,
+      "step": 1211
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1816851753197023e-06,
+      "loss": 0.4644,
+      "step": 1212
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.168503573805096e-06,
+      "loss": 0.4377,
+      "step": 1213
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.155357069818995e-06,
+      "loss": 0.4816,
+      "step": 1214
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1422457222785875e-06,
+      "loss": 0.4573,
+      "step": 1215
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.129169589943514e-06,
+      "loss": 0.4495,
+      "step": 1216
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1161287314155955e-06,
+      "loss": 0.4526,
+      "step": 1217
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1031232051385606e-06,
+      "loss": 0.5039,
+      "step": 1218
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0901530693978055e-06,
+      "loss": 0.4778,
+      "step": 1219
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.077218382320113e-06,
+      "loss": 0.4648,
+      "step": 1220
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0643192018734027e-06,
+      "loss": 0.4543,
+      "step": 1221
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0514555858664663e-06,
+      "loss": 0.4902,
+      "step": 1222
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.038627591948715e-06,
+      "loss": 0.4463,
+      "step": 1223
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.025835277609908e-06,
+      "loss": 0.4224,
+      "step": 1224
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0130787001799103e-06,
+      "loss": 0.4838,
+      "step": 1225
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.000357916828428e-06,
+      "loss": 0.448,
+      "step": 1226
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.987672984564749e-06,
+      "loss": 0.4837,
+      "step": 1227
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.975023960237499e-06,
+      "loss": 0.4941,
+      "step": 1228
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.962410900534367e-06,
+      "loss": 0.4567,
+      "step": 1229
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.949833861981877e-06,
+      "loss": 0.4767,
+      "step": 1230
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9372929009451115e-06,
+      "loss": 0.48,
+      "step": 1231
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.924788073627474e-06,
+      "loss": 0.481,
+      "step": 1232
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9123194360704288e-06,
+      "loss": 0.4632,
+      "step": 1233
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8998870441532569e-06,
+      "loss": 0.4621,
+      "step": 1234
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.887490953592791e-06,
+      "loss": 0.4372,
+      "step": 1235
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.875131219943187e-06,
+      "loss": 0.4829,
+      "step": 1236
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8628078985956622e-06,
+      "loss": 0.4917,
+      "step": 1237
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8505210447782418e-06,
+      "loss": 0.4202,
+      "step": 1238
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8382707135555255e-06,
+      "loss": 0.5258,
+      "step": 1239
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8260569598284361e-06,
+      "loss": 0.4524,
+      "step": 1240
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8138798383339618e-06,
+      "loss": 0.504,
+      "step": 1241
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8017394036449276e-06,
+      "loss": 0.4862,
+      "step": 1242
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7896357101697405e-06,
+      "loss": 0.5511,
+      "step": 1243
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7775688121521518e-06,
+      "loss": 0.5067,
+      "step": 1244
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7655387636710097e-06,
+      "loss": 0.4571,
+      "step": 1245
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7535456186400123e-06,
+      "loss": 0.4469,
+      "step": 1246
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7415894308074777e-06,
+      "loss": 0.457,
+      "step": 1247
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7296702537560994e-06,
+      "loss": 0.5135,
+      "step": 1248
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7177881409026954e-06,
+      "loss": 0.4279,
+      "step": 1249
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7059431454979825e-06,
+      "loss": 0.4481,
+      "step": 1250
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6941353206263366e-06,
+      "loss": 0.4676,
+      "step": 1251
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.682364719205537e-06,
+      "loss": 0.4786,
+      "step": 1252
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6706313939865604e-06,
+      "loss": 0.4974,
+      "step": 1253
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6589353975533174e-06,
+      "loss": 0.4899,
+      "step": 1254
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6472767823224235e-06,
+      "loss": 0.4866,
+      "step": 1255
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6356556005429758e-06,
+      "loss": 0.4745,
+      "step": 1256
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6240719042963004e-06,
+      "loss": 0.4887,
+      "step": 1257
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6125257454957365e-06,
+      "loss": 0.4655,
+      "step": 1258
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6010171758863958e-06,
+      "loss": 0.4603,
+      "step": 1259
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.589546247044923e-06,
+      "loss": 0.5032,
+      "step": 1260
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5781130103792797e-06,
+      "loss": 0.4366,
+      "step": 1261
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5667175171285054e-06,
+      "loss": 0.4643,
+      "step": 1262
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5553598183624885e-06,
+      "loss": 0.4734,
+      "step": 1263
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5440399649817384e-06,
+      "loss": 0.4551,
+      "step": 1264
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5327580077171589e-06,
+      "loss": 0.4418,
+      "step": 1265
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5215139971298131e-06,
+      "loss": 0.4429,
+      "step": 1266
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5103079836107082e-06,
+      "loss": 0.4758,
+      "step": 1267
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.499140017380566e-06,
+      "loss": 0.5256,
+      "step": 1268
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4880101484895882e-06,
+      "loss": 0.4653,
+      "step": 1269
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4769184268172465e-06,
+      "loss": 0.5014,
+      "step": 1270
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4658649020720539e-06,
+      "loss": 0.4249,
+      "step": 1271
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4548496237913335e-06,
+      "loss": 0.4499,
+      "step": 1272
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4438726413410077e-06,
+      "loss": 0.4787,
+      "step": 1273
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4329340039153738e-06,
+      "loss": 0.4617,
+      "step": 1274
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4220337605368816e-06,
+      "loss": 0.433,
+      "step": 1275
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4111719600559159e-06,
+      "loss": 0.4376,
+      "step": 1276
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4003486511505715e-06,
+      "loss": 0.5173,
+      "step": 1277
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3895638823264447e-06,
+      "loss": 0.4466,
+      "step": 1278
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3788177019164117e-06,
+      "loss": 0.4571,
+      "step": 1279
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3681101580804057e-06,
+      "loss": 0.403,
+      "step": 1280
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3574412988052132e-06,
+      "loss": 0.4925,
+      "step": 1281
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3468111719042497e-06,
+      "loss": 0.5047,
+      "step": 1282
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3362198250173454e-06,
+      "loss": 0.4438,
+      "step": 1283
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3256673056105385e-06,
+      "loss": 0.4347,
+      "step": 1284
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3151536609758586e-06,
+      "loss": 0.458,
+      "step": 1285
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3046789382311132e-06,
+      "loss": 0.4496,
+      "step": 1286
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.294243184319679e-06,
+      "loss": 0.4666,
+      "step": 1287
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2838464460102862e-06,
+      "loss": 0.4909,
+      "step": 1288
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.273488769896819e-06,
+      "loss": 0.5224,
+      "step": 1289
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2631702023980997e-06,
+      "loss": 0.4497,
+      "step": 1290
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.252890789757678e-06,
+      "loss": 0.429,
+      "step": 1291
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2426505780436326e-06,
+      "loss": 0.5406,
+      "step": 1292
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2324496131483588e-06,
+      "loss": 0.509,
+      "step": 1293
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2222879407883592e-06,
+      "loss": 0.4685,
+      "step": 1294
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2121656065040498e-06,
+      "loss": 0.4794,
+      "step": 1295
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2020826556595477e-06,
+      "loss": 0.4763,
+      "step": 1296
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1920391334424675e-06,
+      "loss": 0.4752,
+      "step": 1297
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.182035084863724e-06,
+      "loss": 0.457,
+      "step": 1298
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1720705547573263e-06,
+      "loss": 0.4407,
+      "step": 1299
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1621455877801757e-06,
+      "loss": 0.478,
+      "step": 1300
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1522602284118733e-06,
+      "loss": 0.447,
+      "step": 1301
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1424145209545079e-06,
+      "loss": 0.4999,
+      "step": 1302
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1326085095324712e-06,
+      "loss": 0.4919,
+      "step": 1303
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1228422380922533e-06,
+      "loss": 0.5095,
+      "step": 1304
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1131157504022394e-06,
+      "loss": 0.438,
+      "step": 1305
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1034290900525279e-06,
+      "loss": 0.4226,
+      "step": 1306
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0937823004547254e-06,
+      "loss": 0.5428,
+      "step": 1307
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0841754248417535e-06,
+      "loss": 0.4432,
+      "step": 1308
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0746085062676548e-06,
+      "loss": 0.4787,
+      "step": 1309
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.065081587607406e-06,
+      "loss": 0.4662,
+      "step": 1310
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0555947115567123e-06,
+      "loss": 0.4909,
+      "step": 1311
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0461479206318303e-06,
+      "loss": 0.4662,
+      "step": 1312
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0367412571693746e-06,
+      "loss": 0.477,
+      "step": 1313
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0273747633261144e-06,
+      "loss": 0.4496,
+      "step": 1314
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0180484810788082e-06,
+      "loss": 0.4565,
+      "step": 1315
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0087624522239902e-06,
+      "loss": 0.4446,
+      "step": 1316
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.995167183778055e-07,
+      "loss": 0.4533,
+      "step": 1317
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.903113209758098e-07,
+      "loss": 0.4226,
+      "step": 1318
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.811463012727863e-07,
+      "loss": 0.5086,
+      "step": 1319
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.720217003425648e-07,
+      "loss": 0.5016,
+      "step": 1320
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.629375590778344e-07,
+      "loss": 0.4919,
+      "step": 1321
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.538939181899565e-07,
+      "loss": 0.5209,
+      "step": 1322
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.448908182087913e-07,
+      "loss": 0.4744,
+      "step": 1323
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.359282994825125e-07,
+      "loss": 0.4767,
+      "step": 1324
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.270064021774194e-07,
+      "loss": 0.438,
+      "step": 1325
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.181251662777668e-07,
+      "loss": 0.4188,
+      "step": 1326
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.092846315855841e-07,
+      "loss": 0.5006,
+      "step": 1327
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.004848377204878e-07,
+      "loss": 0.487,
+      "step": 1328
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.917258241195181e-07,
+      "loss": 0.4957,
+      "step": 1329
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.830076300369517e-07,
+      "loss": 0.4685,
+      "step": 1330
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.74330294544129e-07,
+      "loss": 0.5051,
+      "step": 1331
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.656938565292805e-07,
+      "loss": 0.4391,
+      "step": 1332
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.570983546973466e-07,
+      "loss": 0.4274,
+      "step": 1333
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.485438275698154e-07,
+      "loss": 0.4839,
+      "step": 1334
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.4003031348454e-07,
+      "loss": 0.5098,
+      "step": 1335
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.315578505955679e-07,
+      "loss": 0.4367,
+      "step": 1336
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.231264768729753e-07,
+      "loss": 0.426,
+      "step": 1337
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.14736230102694e-07,
+      "loss": 0.4414,
+      "step": 1338
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.063871478863383e-07,
+      "loss": 0.4452,
+      "step": 1339
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.980792676410409e-07,
+      "loss": 0.438,
+      "step": 1340
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.898126265992912e-07,
+      "loss": 0.4272,
+      "step": 1341
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.815872618087506e-07,
+      "loss": 0.4749,
+      "step": 1342
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.734032101321054e-07,
+      "loss": 0.4283,
+      "step": 1343
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.652605082468912e-07,
+      "loss": 0.4709,
+      "step": 1344
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.571591926453292e-07,
+      "loss": 0.4376,
+      "step": 1345
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.490992996341662e-07,
+      "loss": 0.4278,
+      "step": 1346
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.410808653345092e-07,
+      "loss": 0.533,
+      "step": 1347
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.331039256816664e-07,
+      "loss": 0.4748,
+      "step": 1348
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.251685164249823e-07,
+      "loss": 0.4575,
+      "step": 1349
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.17274673127677e-07,
+      "loss": 0.4213,
+      "step": 1350
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.094224311666953e-07,
+      "loss": 0.4878,
+      "step": 1351
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.016118257325399e-07,
+      "loss": 0.4468,
+      "step": 1352
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.938428918291107e-07,
+      "loss": 0.5133,
+      "step": 1353
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.861156642735578e-07,
+      "loss": 0.4635,
+      "step": 1354
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.784301776961222e-07,
+      "loss": 0.5525,
+      "step": 1355
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.707864665399733e-07,
+      "loss": 0.4544,
+      "step": 1356
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.631845650610647e-07,
+      "loss": 0.4622,
+      "step": 1357
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.556245073279777e-07,
+      "loss": 0.4354,
+      "step": 1358
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.481063272217603e-07,
+      "loss": 0.4639,
+      "step": 1359
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.406300584357905e-07,
+      "loss": 0.4796,
+      "step": 1360
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.331957344756123e-07,
+      "loss": 0.4981,
+      "step": 1361
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.258033886587911e-07,
+      "loss": 0.4776,
+      "step": 1362
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.184530541147649e-07,
+      "loss": 0.4705,
+      "step": 1363
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.111447637846923e-07,
+      "loss": 0.4115,
+      "step": 1364
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.038785504213096e-07,
+      "loss": 0.3974,
+      "step": 1365
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.966544465887803e-07,
+      "loss": 0.47,
+      "step": 1366
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.894724846625499e-07,
+      "loss": 0.501,
+      "step": 1367
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.823326968292009e-07,
+      "loss": 0.5106,
+      "step": 1368
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.752351150863133e-07,
+      "loss": 0.4754,
+      "step": 1369
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.681797712423099e-07,
+      "loss": 0.4708,
+      "step": 1370
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.611666969163243e-07,
+      "loss": 0.4429,
+      "step": 1371
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.541959235380578e-07,
+      "loss": 0.4753,
+      "step": 1372
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.472674823476343e-07,
+      "loss": 0.4835,
+      "step": 1373
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.403814043954592e-07,
+      "loss": 0.4537,
+      "step": 1374
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.335377205420911e-07,
+      "loss": 0.4342,
+      "step": 1375
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.267364614580861e-07,
+      "loss": 0.4493,
+      "step": 1376
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.199776576238791e-07,
+      "loss": 0.4988,
+      "step": 1377
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.132613393296293e-07,
+      "loss": 0.3887,
+      "step": 1378
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.065875366751016e-07,
+      "loss": 0.4718,
+      "step": 1379
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.999562795695201e-07,
+      "loss": 0.515,
+      "step": 1380
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.933675977314334e-07,
+      "loss": 0.4398,
+      "step": 1381
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.868215206885918e-07,
+      "loss": 0.477,
+      "step": 1382
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.80318077777805e-07,
+      "loss": 0.4767,
+      "step": 1383
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.738572981448153e-07,
+      "loss": 0.4699,
+      "step": 1384
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6743921074416387e-07,
+      "loss": 0.5079,
+      "step": 1385
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.61063844339068e-07,
+      "loss": 0.497,
+      "step": 1386
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.547312275012794e-07,
+      "loss": 0.4395,
+      "step": 1387
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4844138861096954e-07,
+      "loss": 0.4369,
+      "step": 1388
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.421943558565966e-07,
+      "loss": 0.4174,
+      "step": 1389
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.359901572347758e-07,
+      "loss": 0.4954,
+      "step": 1390
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.298288205501633e-07,
+      "loss": 0.4736,
+      "step": 1391
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2371037341531805e-07,
+      "loss": 0.4572,
+      "step": 1392
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.176348432505917e-07,
+      "loss": 0.5145,
+      "step": 1393
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.116022572839984e-07,
+      "loss": 0.3836,
+      "step": 1394
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0561264255109424e-07,
+      "loss": 0.4447,
+      "step": 1395
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.996660258948548e-07,
+      "loss": 0.4725,
+      "step": 1396
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.937624339655599e-07,
+      "loss": 0.4865,
+      "step": 1397
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.879018932206624e-07,
+      "loss": 0.4848,
+      "step": 1398
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.820844299246851e-07,
+      "loss": 0.5038,
+      "step": 1399
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7631007014909294e-07,
+      "loss": 0.4953,
+      "step": 1400
+    }
+  ],
+  "max_steps": 1530,
+  "num_train_epochs": 3,
+  "total_flos": 3.048437414847578e+18,
+  "trial_name": null,
+  "trial_params": null
+}