diff --git "a/checkpoint-1000/trainer_state.json" "b/checkpoint-1000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-1000/trainer_state.json"
@@ -0,0 +1,6016 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.6534879921581441,
+  "global_step": 1000,
+  "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": 1.658,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.695652173913044e-07,
+      "loss": 1.5847,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3043478260869566e-06,
+      "loss": 1.6487,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7391304347826088e-06,
+      "loss": 1.6376,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.173913043478261e-06,
+      "loss": 1.5046,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6086956521739132e-06,
+      "loss": 1.439,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.043478260869566e-06,
+      "loss": 1.4233,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4782608695652175e-06,
+      "loss": 1.4303,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.91304347826087e-06,
+      "loss": 1.4826,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.347826086956522e-06,
+      "loss": 1.3716,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.782608695652174e-06,
+      "loss": 1.3658,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.2173913043478265e-06,
+      "loss": 1.3963,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.652173913043479e-06,
+      "loss": 1.4064,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.086956521739132e-06,
+      "loss": 1.3534,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.521739130434783e-06,
+      "loss": 1.3417,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.956521739130435e-06,
+      "loss": 1.3738,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.391304347826087e-06,
+      "loss": 1.2103,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.82608695652174e-06,
+      "loss": 1.2787,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.260869565217392e-06,
+      "loss": 1.4267,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.695652173913044e-06,
+      "loss": 1.3208,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.130434782608697e-06,
+      "loss": 1.2949,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.565217391304349e-06,
+      "loss": 1.3164,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 1.3789,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0434782608695653e-05,
+      "loss": 1.3058,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0869565217391305e-05,
+      "loss": 1.3274,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1304347826086957e-05,
+      "loss": 1.3245,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1739130434782611e-05,
+      "loss": 1.3816,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2173913043478263e-05,
+      "loss": 1.3398,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2608695652173915e-05,
+      "loss": 1.3014,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3043478260869566e-05,
+      "loss": 1.2833,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3478260869565218e-05,
+      "loss": 1.3021,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.391304347826087e-05,
+      "loss": 1.2483,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4347826086956522e-05,
+      "loss": 1.197,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4782608695652174e-05,
+      "loss": 1.2796,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5217391304347828e-05,
+      "loss": 1.2436,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.565217391304348e-05,
+      "loss": 1.3143,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6086956521739132e-05,
+      "loss": 1.2572,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6521739130434785e-05,
+      "loss": 1.2133,
+      "step": 38
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6956521739130437e-05,
+      "loss": 1.1562,
+      "step": 39
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.739130434782609e-05,
+      "loss": 1.2607,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.782608695652174e-05,
+      "loss": 1.2708,
+      "step": 41
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8260869565217393e-05,
+      "loss": 1.2855,
+      "step": 42
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8695652173913045e-05,
+      "loss": 1.2714,
+      "step": 43
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9130434782608697e-05,
+      "loss": 1.2747,
+      "step": 44
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.956521739130435e-05,
+      "loss": 1.2935,
+      "step": 45
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 1.3802,
+      "step": 46
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999977592067603e-05,
+      "loss": 1.3119,
+      "step": 47
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999910368370826e-05,
+      "loss": 1.3561,
+      "step": 48
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999798329210938e-05,
+      "loss": 1.3027,
+      "step": 49
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999964147509006e-05,
+      "loss": 1.3295,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999439806711137e-05,
+      "loss": 1.2899,
+      "step": 51
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999193324977974e-05,
+      "loss": 1.3297,
+      "step": 52
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999890203099519e-05,
+      "loss": 1.3059,
+      "step": 53
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998565926068253e-05,
+      "loss": 1.29,
+      "step": 54
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999818501170344e-05,
+      "loss": 1.2265,
+      "step": 55
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997759289607854e-05,
+      "loss": 1.2652,
+      "step": 56
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997288761689404e-05,
+      "loss": 1.3858,
+      "step": 57
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996773430056806e-05,
+      "loss": 1.3069,
+      "step": 58
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996213297019556e-05,
+      "loss": 1.2414,
+      "step": 59
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995608365087945e-05,
+      "loss": 1.3432,
+      "step": 60
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994958636973024e-05,
+      "loss": 1.4153,
+      "step": 61
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999426411558661e-05,
+      "loss": 1.3386,
+      "step": 62
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993524804041255e-05,
+      "loss": 1.2682,
+      "step": 63
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992740705650252e-05,
+      "loss": 1.3186,
+      "step": 64
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991911823927607e-05,
+      "loss": 1.4159,
+      "step": 65
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991038162588018e-05,
+      "loss": 1.2346,
+      "step": 66
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999011972554688e-05,
+      "loss": 1.3291,
+      "step": 67
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989156516920248e-05,
+      "loss": 1.2836,
+      "step": 68
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998814854102482e-05,
+      "loss": 1.2737,
+      "step": 69
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987095802377933e-05,
+      "loss": 1.3478,
+      "step": 70
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998599830569752e-05,
+      "loss": 1.391,
+      "step": 71
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998485605590211e-05,
+      "loss": 1.2899,
+      "step": 72
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99836690581108e-05,
+      "loss": 1.3001,
+      "step": 73
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982437317643218e-05,
+      "loss": 1.2358,
+      "step": 74
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981160840019513e-05,
+      "loss": 1.2875,
+      "step": 75
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979839630960333e-05,
+      "loss": 1.2769,
+      "step": 76
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978473696386787e-05,
+      "loss": 1.3458,
+      "step": 77
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977063042420438e-05,
+      "loss": 1.3433,
+      "step": 78
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975607675383243e-05,
+      "loss": 1.3246,
+      "step": 79
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997410760179756e-05,
+      "loss": 1.2984,
+      "step": 80
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99725628283861e-05,
+      "loss": 1.2665,
+      "step": 81
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99709733620719e-05,
+      "loss": 1.3427,
+      "step": 82
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969339209978287e-05,
+      "loss": 1.3314,
+      "step": 83
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967660379428855e-05,
+      "loss": 1.2411,
+      "step": 84
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996593687794743e-05,
+      "loss": 1.212,
+      "step": 85
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996416871325803e-05,
+      "loss": 1.3132,
+      "step": 86
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996235589328484e-05,
+      "loss": 1.2823,
+      "step": 87
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996049842615217e-05,
+      "loss": 1.4153,
+      "step": 88
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995859632018442e-05,
+      "loss": 1.3212,
+      "step": 89
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995664958390604e-05,
+      "loss": 1.1837,
+      "step": 90
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99546582260415e-05,
+      "loss": 1.3189,
+      "step": 91
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995262225551524e-05,
+      "loss": 1.2477,
+      "step": 92
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950541681451644e-05,
+      "loss": 1.248,
+      "step": 93
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948416513174976e-05,
+      "loss": 1.2963,
+      "step": 94
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994624676020936e-05,
+      "loss": 1.2874,
+      "step": 95
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944032432278743e-05,
+      "loss": 1.2947,
+      "step": 96
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941773539306818e-05,
+      "loss": 1.309,
+      "step": 97
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939470091417012e-05,
+      "loss": 1.3141,
+      "step": 98
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937122098932428e-05,
+      "loss": 1.2828,
+      "step": 99
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934729572375792e-05,
+      "loss": 1.3323,
+      "step": 100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932292522469424e-05,
+      "loss": 1.2671,
+      "step": 101
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992981096013517e-05,
+      "loss": 1.3129,
+      "step": 102
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992728489649437e-05,
+      "loss": 1.2566,
+      "step": 103
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992471434286779e-05,
+      "loss": 1.279,
+      "step": 104
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99220993107756e-05,
+      "loss": 1.2325,
+      "step": 105
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919439811937283e-05,
+      "loss": 1.3848,
+      "step": 106
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916735858271615e-05,
+      "loss": 1.2736,
+      "step": 107
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913987461896597e-05,
+      "loss": 1.2252,
+      "step": 108
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991119463512941e-05,
+      "loss": 1.2248,
+      "step": 109
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908357390486342e-05,
+      "loss": 1.253,
+      "step": 110
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905475740682756e-05,
+      "loss": 1.256,
+      "step": 111
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990254969863301e-05,
+      "loss": 1.3436,
+      "step": 112
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899579277450417e-05,
+      "loss": 1.1671,
+      "step": 113
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989656449044718e-05,
+      "loss": 1.3457,
+      "step": 114
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893505351134322e-05,
+      "loss": 1.315,
+      "step": 115
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890401873221642e-05,
+      "loss": 1.2173,
+      "step": 116
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887254070617644e-05,
+      "loss": 1.3156,
+      "step": 117
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988406195742948e-05,
+      "loss": 1.3153,
+      "step": 118
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880825547962882e-05,
+      "loss": 1.2818,
+      "step": 119
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877544856722094e-05,
+      "loss": 1.3481,
+      "step": 120
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987421989840982e-05,
+      "loss": 1.3723,
+      "step": 121
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987085068792715e-05,
+      "loss": 1.2784,
+      "step": 122
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986743724037349e-05,
+      "loss": 1.2764,
+      "step": 123
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98639795710465e-05,
+      "loss": 1.2896,
+      "step": 124
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986047769544203e-05,
+      "loss": 1.3305,
+      "step": 125
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856931629254032e-05,
+      "loss": 1.2695,
+      "step": 126
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853341388374504e-05,
+      "loss": 1.2534,
+      "step": 127
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849706988893433e-05,
+      "loss": 1.2319,
+      "step": 128
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846028447098687e-05,
+      "loss": 1.2202,
+      "step": 129
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984230577947597e-05,
+      "loss": 1.3401,
+      "step": 130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983853900270874e-05,
+      "loss": 1.3309,
+      "step": 131
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983472813367813e-05,
+      "loss": 1.3011,
+      "step": 132
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830873189462886e-05,
+      "loss": 1.3471,
+      "step": 133
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826974187339267e-05,
+      "loss": 1.2037,
+      "step": 134
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982303114478099e-05,
+      "loss": 1.2936,
+      "step": 135
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819044079459142e-05,
+      "loss": 1.1976,
+      "step": 136
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815013009242103e-05,
+      "loss": 1.248,
+      "step": 137
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981093795219546e-05,
+      "loss": 1.3496,
+      "step": 138
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980681892658193e-05,
+      "loss": 1.2167,
+      "step": 139
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980265595086129e-05,
+      "loss": 1.2875,
+      "step": 140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798449043690272e-05,
+      "loss": 1.2559,
+      "step": 141
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9794198223922496e-05,
+      "loss": 1.313,
+      "step": 142
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789903510608374e-05,
+      "loss": 1.2399,
+      "step": 143
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785564922995042e-05,
+      "loss": 1.2255,
+      "step": 144
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978118248052625e-05,
+      "loss": 1.3483,
+      "step": 145
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9776756202842297e-05,
+      "loss": 1.2305,
+      "step": 146
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977228610977992e-05,
+      "loss": 1.2741,
+      "step": 147
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976777222137224e-05,
+      "loss": 1.2095,
+      "step": 148
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9763214557848634e-05,
+      "loss": 1.2364,
+      "step": 149
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9758613139634662e-05,
+      "loss": 1.3913,
+      "step": 150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975396798735198e-05,
+      "loss": 1.2574,
+      "step": 151
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749279121818235e-05,
+      "loss": 1.2894,
+      "step": 152
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744546564046995e-05,
+      "loss": 1.3109,
+      "step": 153
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9739770335247616e-05,
+      "loss": 1.3483,
+      "step": 154
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9734950456825187e-05,
+      "loss": 1.3476,
+      "step": 155
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9730086950380404e-05,
+      "loss": 1.2305,
+      "step": 156
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9725179837709494e-05,
+      "loss": 1.2062,
+      "step": 157
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972022914080411e-05,
+      "loss": 1.2074,
+      "step": 158
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715234881851226e-05,
+      "loss": 1.2691,
+      "step": 159
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9710197083233044e-05,
+      "loss": 1.2515,
+      "step": 160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9705115767526894e-05,
+      "loss": 1.2018,
+      "step": 161
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9699990957505136e-05,
+      "loss": 1.1156,
+      "step": 162
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9694822676135045e-05,
+      "loss": 1.2945,
+      "step": 163
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9689610946578725e-05,
+      "loss": 1.2831,
+      "step": 164
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968435579219299e-05,
+      "loss": 1.2824,
+      "step": 165
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9679057236529266e-05,
+      "loss": 1.2378,
+      "step": 166
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9673715303333495e-05,
+      "loss": 1.2443,
+      "step": 167
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668330016546004e-05,
+      "loss": 1.3786,
+      "step": 168
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9662901400301432e-05,
+      "loss": 1.2761,
+      "step": 169
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965742947892858e-05,
+      "loss": 1.2731,
+      "step": 170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965191427695035e-05,
+      "loss": 1.2325,
+      "step": 171
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964635581908359e-05,
+      "loss": 1.4167,
+      "step": 172
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9640754130239004e-05,
+      "loss": 1.2157,
+      "step": 173
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9635109235521057e-05,
+      "loss": 1.2924,
+      "step": 174
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9629421160227825e-05,
+      "loss": 1.3156,
+      "step": 175
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962368992985091e-05,
+      "loss": 1.3434,
+      "step": 176
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9617915570075317e-05,
+      "loss": 1.3234,
+      "step": 177
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961209810677934e-05,
+      "loss": 1.2945,
+      "step": 178
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9606237566034443e-05,
+      "loss": 1.277,
+      "step": 179
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9600333974105145e-05,
+      "loss": 1.2148,
+      "step": 180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9594387357448908e-05,
+      "loss": 1.2183,
+      "step": 181
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9588397742716004e-05,
+      "loss": 1.2559,
+      "step": 182
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958236515674941e-05,
+      "loss": 1.3053,
+      "step": 183
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9576289626584685e-05,
+      "loss": 1.286,
+      "step": 184
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957017117944984e-05,
+      "loss": 1.242,
+      "step": 185
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9564009842765225e-05,
+      "loss": 1.1833,
+      "step": 186
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9557805644143408e-05,
+      "loss": 1.2265,
+      "step": 187
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955155861138903e-05,
+      "loss": 1.2032,
+      "step": 188
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9545268772498723e-05,
+      "loss": 1.3044,
+      "step": 189
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9538936155660934e-05,
+      "loss": 1.3028,
+      "step": 190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9532560789255838e-05,
+      "loss": 1.3328,
+      "step": 191
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9526142701855187e-05,
+      "loss": 1.2071,
+      "step": 192
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9519681922222195e-05,
+      "loss": 1.3534,
+      "step": 193
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951317847931141e-05,
+      "loss": 1.2878,
+      "step": 194
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950663240226857e-05,
+      "loss": 1.2647,
+      "step": 195
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9500043720430484e-05,
+      "loss": 1.2338,
+      "step": 196
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9493412463324903e-05,
+      "loss": 1.3765,
+      "step": 197
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9486738660670373e-05,
+      "loss": 1.1946,
+      "step": 198
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9480022342376125e-05,
+      "loss": 1.2972,
+      "step": 199
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9473263538541916e-05,
+      "loss": 1.2961,
+      "step": 200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946646227945791e-05,
+      "loss": 1.2324,
+      "step": 201
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945961859560454e-05,
+      "loss": 1.1474,
+      "step": 202
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9452732517652366e-05,
+      "loss": 1.359,
+      "step": 203
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9445804076461946e-05,
+      "loss": 1.3617,
+      "step": 204
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9438833303083677e-05,
+      "loss": 1.3684,
+      "step": 205
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943182022875769e-05,
+      "loss": 1.2624,
+      "step": 206
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942476488491369e-05,
+      "loss": 1.2837,
+      "step": 207
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9417667303170803e-05,
+      "loss": 1.2425,
+      "step": 208
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941052751533745e-05,
+      "loss": 1.3908,
+      "step": 209
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940334555341122e-05,
+      "loss": 1.1629,
+      "step": 210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9396121449578694e-05,
+      "loss": 1.225,
+      "step": 211
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938885523621531e-05,
+      "loss": 1.2935,
+      "step": 212
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9381546945885236e-05,
+      "loss": 1.2493,
+      "step": 213
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9374196611341212e-05,
+      "loss": 1.1924,
+      "step": 214
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9366804265524388e-05,
+      "loss": 1.1538,
+      "step": 215
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935936994156421e-05,
+      "loss": 1.3094,
+      "step": 216
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935189367277824e-05,
+      "loss": 1.2147,
+      "step": 217
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9344375492672024e-05,
+      "loss": 1.1849,
+      "step": 218
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9336815434938935e-05,
+      "loss": 1.2968,
+      "step": 219
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932921353346003e-05,
+      "loss": 1.2963,
+      "step": 220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932156982230388e-05,
+      "loss": 1.279,
+      "step": 221
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9313884335726443e-05,
+      "loss": 1.194,
+      "step": 222
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9306157108170892e-05,
+      "loss": 1.2834,
+      "step": 223
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9298388174267463e-05,
+      "loss": 1.1519,
+      "step": 224
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9290577568833305e-05,
+      "loss": 1.3101,
+      "step": 225
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9282725326872324e-05,
+      "loss": 1.2542,
+      "step": 226
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927483148357502e-05,
+      "loss": 1.3028,
+      "step": 227
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9266896074318335e-05,
+      "loss": 1.2972,
+      "step": 228
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9258919134665494e-05,
+      "loss": 1.1933,
+      "step": 229
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9250900700365837e-05,
+      "loss": 1.3796,
+      "step": 230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9242840807354676e-05,
+      "loss": 1.271,
+      "step": 231
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923473949175311e-05,
+      "loss": 1.192,
+      "step": 232
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9226596789867895e-05,
+      "loss": 1.3255,
+      "step": 233
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921841273819125e-05,
+      "loss": 1.2359,
+      "step": 234
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921018737340071e-05,
+      "loss": 1.3459,
+      "step": 235
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920192073235896e-05,
+      "loss": 1.1901,
+      "step": 236
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9193612852113666e-05,
+      "loss": 1.1438,
+      "step": 237
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918526376989731e-05,
+      "loss": 1.2968,
+      "step": 238
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9176873523127025e-05,
+      "loss": 1.2323,
+      "step": 239
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9168442149404438e-05,
+      "loss": 1.3393,
+      "step": 240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9159969686515462e-05,
+      "loss": 1.2142,
+      "step": 241
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9151456172430186e-05,
+      "loss": 1.1371,
+      "step": 242
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9142901645302654e-05,
+      "loss": 1.2626,
+      "step": 243
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9134306143470722e-05,
+      "loss": 1.253,
+      "step": 244
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912566970545587e-05,
+      "loss": 1.204,
+      "step": 245
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911699236996305e-05,
+      "loss": 1.2175,
+      "step": 246
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9108274175880483e-05,
+      "loss": 1.3093,
+      "step": 247
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9099515162279515e-05,
+      "loss": 1.2699,
+      "step": 248
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909071536841442e-05,
+      "loss": 1.307,
+      "step": 249
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9081874833722234e-05,
+      "loss": 1.2227,
+      "step": 250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9072993597822586e-05,
+      "loss": 1.1945,
+      "step": 251
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906407170051749e-05,
+      "loss": 1.2266,
+      "step": 252
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905510918179121e-05,
+      "loss": 1.2669,
+      "step": 253
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9046106081810047e-05,
+      "loss": 1.2756,
+      "step": 254
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903706244092217e-05,
+      "loss": 1.2515,
+      "step": 255
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9027978299657436e-05,
+      "loss": 1.2266,
+      "step": 256
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9018853698727216e-05,
+      "loss": 1.3235,
+      "step": 257
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900968867902419e-05,
+      "loss": 1.2487,
+      "step": 258
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9000483281622198e-05,
+      "loss": 1.2412,
+      "step": 259
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8991237547776014e-05,
+      "loss": 1.1866,
+      "step": 260
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8981951518921194e-05,
+      "loss": 1.2566,
+      "step": 261
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8972625236673887e-05,
+      "loss": 1.2875,
+      "step": 262
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896325874283063e-05,
+      "loss": 1.1893,
+      "step": 263
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895385207936817e-05,
+      "loss": 1.3017,
+      "step": 264
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894440528844329e-05,
+      "loss": 1.2351,
+      "step": 265
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8934918412392596e-05,
+      "loss": 1.3018,
+      "step": 266
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8925391493732346e-05,
+      "loss": 1.3364,
+      "step": 267
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891582457515825e-05,
+      "loss": 1.2788,
+      "step": 268
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890621769954528e-05,
+      "loss": 1.2523,
+      "step": 269
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8896570909947477e-05,
+      "loss": 1.2419,
+      "step": 270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8886884249597763e-05,
+      "loss": 1.1993,
+      "step": 271
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887715776190775e-05,
+      "loss": 1.2581,
+      "step": 272
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8867391490467534e-05,
+      "loss": 1.1852,
+      "step": 273
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8857585479045493e-05,
+      "loss": 1.2443,
+      "step": 274
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884773977158813e-05,
+      "loss": 1.2524,
+      "step": 275
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8837854412219828e-05,
+      "loss": 1.1538,
+      "step": 276
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8827929445242678e-05,
+      "loss": 1.2405,
+      "step": 277
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8817964915136277e-05,
+      "loss": 1.1908,
+      "step": 278
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8807960866557535e-05,
+      "loss": 1.2412,
+      "step": 279
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8797917344340453e-05,
+      "loss": 1.2212,
+      "step": 280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8787834393495952e-05,
+      "loss": 1.2132,
+      "step": 281
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8777712059211643e-05,
+      "loss": 1.2642,
+      "step": 282
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8767550386851646e-05,
+      "loss": 1.2226,
+      "step": 283
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875734942195637e-05,
+      "loss": 1.2801,
+      "step": 284
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8747109210242326e-05,
+      "loss": 1.3122,
+      "step": 285
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8736829797601903e-05,
+      "loss": 1.2728,
+      "step": 286
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8726511230103183e-05,
+      "loss": 1.2868,
+      "step": 287
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8716153553989716e-05,
+      "loss": 1.1839,
+      "step": 288
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8705756815680324e-05,
+      "loss": 1.1724,
+      "step": 289
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8695321061768886e-05,
+      "loss": 1.1994,
+      "step": 290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8684846339024145e-05,
+      "loss": 1.2004,
+      "step": 291
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8674332694389464e-05,
+      "loss": 1.1944,
+      "step": 292
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8663780174982656e-05,
+      "loss": 1.2524,
+      "step": 293
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8653188828095754e-05,
+      "loss": 1.3951,
+      "step": 294
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8642558701194787e-05,
+      "loss": 1.3361,
+      "step": 295
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8631889841919596e-05,
+      "loss": 1.198,
+      "step": 296
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862118229808359e-05,
+      "loss": 1.2669,
+      "step": 297
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8610436117673557e-05,
+      "loss": 1.2139,
+      "step": 298
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859965134884943e-05,
+      "loss": 1.3054,
+      "step": 299
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8588828039944086e-05,
+      "loss": 1.239,
+      "step": 300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857796623946312e-05,
+      "loss": 1.2179,
+      "step": 301
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8567065996084628e-05,
+      "loss": 1.1934,
+      "step": 302
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8556127358658995e-05,
+      "loss": 1.2408,
+      "step": 303
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854515037620867e-05,
+      "loss": 1.3218,
+      "step": 304
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853413509792795e-05,
+      "loss": 1.2498,
+      "step": 305
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8523081573182754e-05,
+      "loss": 1.1693,
+      "step": 306
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8511989851510416e-05,
+      "loss": 1.1536,
+      "step": 307
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8500859982619438e-05,
+      "loss": 1.275,
+      "step": 308
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848969201638929e-05,
+      "loss": 1.2712,
+      "step": 309
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847848600287019e-05,
+      "loss": 1.266,
+      "step": 310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8467241992282842e-05,
+      "loss": 1.2122,
+      "step": 311
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845596003501826e-05,
+      "loss": 1.2051,
+      "step": 312
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8444640181637514e-05,
+      "loss": 1.2724,
+      "step": 313
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8433282482871497e-05,
+      "loss": 1.2526,
+      "step": 314
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8421886989620724e-05,
+      "loss": 1.1622,
+      "step": 315
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841045375295508e-05,
+      "loss": 1.2903,
+      "step": 316
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839898282411361e-05,
+      "loss": 1.2329,
+      "step": 317
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8387474254504265e-05,
+      "loss": 1.2136,
+      "step": 318
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8375928095703703e-05,
+      "loss": 1.2846,
+      "step": 319
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8364344399457028e-05,
+      "loss": 1.2696,
+      "step": 320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835272321767758e-05,
+      "loss": 1.187,
+      "step": 321
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8341064602446686e-05,
+      "loss": 1.2331,
+      "step": 322
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832936860601344e-05,
+      "loss": 1.2724,
+      "step": 323
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8317635280794466e-05,
+      "loss": 1.2181,
+      "step": 324
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8305864679373667e-05,
+      "loss": 1.1992,
+      "step": 325
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829405685450202e-05,
+      "loss": 1.1572,
+      "step": 326
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8282211859097308e-05,
+      "loss": 1.1881,
+      "step": 327
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8270329746243903e-05,
+      "loss": 1.2213,
+      "step": 328
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8258410569192526e-05,
+      "loss": 1.2199,
+      "step": 329
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824645438135999e-05,
+      "loss": 1.2589,
+      "step": 330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8234461236328992e-05,
+      "loss": 1.1814,
+      "step": 331
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822243118784785e-05,
+      "loss": 1.1442,
+      "step": 332
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821036428983026e-05,
+      "loss": 1.26,
+      "step": 333
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8198260596355077e-05,
+      "loss": 1.268,
+      "step": 334
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818612016166604e-05,
+      "loss": 1.131,
+      "step": 335
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8173943040171567e-05,
+      "loss": 1.2818,
+      "step": 336
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8161729286444474e-05,
+      "loss": 1.1387,
+      "step": 337
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814947895522176e-05,
+      "loss": 1.2065,
+      "step": 338
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8137192101404343e-05,
+      "loss": 1.2687,
+      "step": 339
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8124868780056814e-05,
+      "loss": 1.2522,
+      "step": 340
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811250904640721e-05,
+      "loss": 1.1396,
+      "step": 341
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8100112955846746e-05,
+      "loss": 1.2635,
+      "step": 342
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8087680563929574e-05,
+      "loss": 1.2622,
+      "step": 343
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807521192637253e-05,
+      "loss": 1.0812,
+      "step": 344
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806270709905489e-05,
+      "loss": 1.1504,
+      "step": 345
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805016613801813e-05,
+      "loss": 1.246,
+      "step": 346
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8037589099465637e-05,
+      "loss": 1.2893,
+      "step": 347
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8024976039762507e-05,
+      "loss": 1.228,
+      "step": 348
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8012327015435255e-05,
+      "loss": 1.1992,
+      "step": 349
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7999642083171576e-05,
+      "loss": 1.2022,
+      "step": 350
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798692129982009e-05,
+      "loss": 1.1628,
+      "step": 351
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7974164722390094e-05,
+      "loss": 1.3536,
+      "step": 352
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796137240805129e-05,
+      "loss": 1.228,
+      "step": 353
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7948544414133534e-05,
+      "loss": 1.2561,
+      "step": 354
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7935680798126597e-05,
+      "loss": 1.1878,
+      "step": 355
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792278161767989e-05,
+      "loss": 1.3139,
+      "step": 356
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79098469306022e-05,
+      "loss": 1.2356,
+      "step": 357
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7896876794861443e-05,
+      "loss": 1.2904,
+      "step": 358
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788387126858441e-05,
+      "loss": 1.3259,
+      "step": 359
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7870830410056487e-05,
+      "loss": 1.2614,
+      "step": 360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7857754277721417e-05,
+      "loss": 1.2127,
+      "step": 361
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7844642930181008e-05,
+      "loss": 1.2234,
+      "step": 362
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7831496426194906e-05,
+      "loss": 1.3165,
+      "step": 363
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78183148246803e-05,
+      "loss": 1.1426,
+      "step": 364
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780509818471168e-05,
+      "loss": 1.2525,
+      "step": 365
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779184656552056e-05,
+      "loss": 1.2378,
+      "step": 366
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7778560026495218e-05,
+      "loss": 1.2249,
+      "step": 367
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7765238627180424e-05,
+      "loss": 1.1897,
+      "step": 368
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775188242727719e-05,
+      "loss": 1.2608,
+      "step": 369
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773849148664247e-05,
+      "loss": 1.2953,
+      "step": 370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772506586528893e-05,
+      "loss": 1.2901,
+      "step": 371
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7711605623384653e-05,
+      "loss": 1.2712,
+      "step": 372
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7698110821252873e-05,
+      "loss": 1.2286,
+      "step": 373
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7684581519371714e-05,
+      "loss": 1.2791,
+      "step": 374
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7671017778373914e-05,
+      "loss": 1.2633,
+      "step": 375
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7657419659046555e-05,
+      "loss": 1.14,
+      "step": 376
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7643787222330773e-05,
+      "loss": 1.2696,
+      "step": 377
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7630120529321518e-05,
+      "loss": 1.1512,
+      "step": 378
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7616419641267267e-05,
+      "loss": 1.1531,
+      "step": 379
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760268461956972e-05,
+      "loss": 1.1738,
+      "step": 380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758891552578357e-05,
+      "loss": 1.2082,
+      "step": 381
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7575112421616203e-05,
+      "loss": 1.1992,
+      "step": 382
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7561275368927422e-05,
+      "loss": 1.2195,
+      "step": 383
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7547404429729174e-05,
+      "loss": 1.1806,
+      "step": 384
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753349966618528e-05,
+      "loss": 1.1613,
+      "step": 385
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751956114061113e-05,
+      "loss": 1.2528,
+      "step": 386
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7505588915473437e-05,
+      "loss": 1.2632,
+      "step": 387
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7491583053389937e-05,
+      "loss": 1.216,
+      "step": 388
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747754361712911e-05,
+      "loss": 1.1966,
+      "step": 389
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7463470669609907e-05,
+      "loss": 1.2612,
+      "step": 390
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7449364273901457e-05,
+      "loss": 1.1057,
+      "step": 391
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743522449322279e-05,
+      "loss": 1.1335,
+      "step": 392
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742105139094256e-05,
+      "loss": 1.2405,
+      "step": 393
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7406845030578747e-05,
+      "loss": 1.2579,
+      "step": 394
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7392605475798387e-05,
+      "loss": 1.0973,
+      "step": 395
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7378332790417275e-05,
+      "loss": 1.1534,
+      "step": 396
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7364027038399682e-05,
+      "loss": 1.216,
+      "step": 397
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734968828385808e-05,
+      "loss": 1.2669,
+      "step": 398
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7335316591052828e-05,
+      "loss": 1.1461,
+      "step": 399
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7320912024391915e-05,
+      "loss": 1.2305,
+      "step": 400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7306474648430653e-05,
+      "loss": 1.2888,
+      "step": 401
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729200452787139e-05,
+      "loss": 1.3148,
+      "step": 402
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7277501727563227e-05,
+      "loss": 1.1997,
+      "step": 403
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726296631250171e-05,
+      "loss": 1.208,
+      "step": 404
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724839834782856e-05,
+      "loss": 1.2427,
+      "step": 405
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7233797898831376e-05,
+      "loss": 1.2214,
+      "step": 406
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7219165030943327e-05,
+      "loss": 1.1986,
+      "step": 407
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720449980974288e-05,
+      "loss": 1.2567,
+      "step": 408
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7189802300953488e-05,
+      "loss": 1.1098,
+      "step": 409
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717507257044331e-05,
+      "loss": 1.1329,
+      "step": 410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716031068422491e-05,
+      "loss": 1.1693,
+      "step": 411
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7145516708454953e-05,
+      "loss": 1.1304,
+      "step": 412
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7130690709433927e-05,
+      "loss": 1.2287,
+      "step": 413
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711583275360582e-05,
+      "loss": 1.2817,
+      "step": 414
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710094290755785e-05,
+      "loss": 1.2225,
+      "step": 415
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7086021238020154e-05,
+      "loss": 1.1936,
+      "step": 416
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 1.2964,
+      "step": 417
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7056082696108896e-05,
+      "loss": 1.2376,
+      "step": 418
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70410659579075e-05,
+      "loss": 1.1612,
+      "step": 419
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7026017664560108e-05,
+      "loss": 1.3043,
+      "step": 420
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7010937883506936e-05,
+      "loss": 1.2882,
+      "step": 421
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699582668232934e-05,
+      "loss": 1.2565,
+      "step": 422
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698068412874946e-05,
+      "loss": 1.1343,
+      "step": 423
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6965510290629973e-05,
+      "loss": 1.2255,
+      "step": 424
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695030523597374e-05,
+      "loss": 1.2678,
+      "step": 425
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6935069032923525e-05,
+      "loss": 1.1335,
+      "step": 426
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6919801749761696e-05,
+      "loss": 1.2225,
+      "step": 427
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6904503454909905e-05,
+      "loss": 1.1734,
+      "step": 428
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6889174216928776e-05,
+      "loss": 1.1718,
+      "step": 429
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6873814104517617e-05,
+      "loss": 1.1448,
+      "step": 430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6858423186514108e-05,
+      "loss": 1.1668,
+      "step": 431
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6843001531893967e-05,
+      "loss": 1.1733,
+      "step": 432
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6827549209770676e-05,
+      "loss": 1.2871,
+      "step": 433
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6812066289395157e-05,
+      "loss": 1.2404,
+      "step": 434
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6796552840155453e-05,
+      "loss": 1.186,
+      "step": 435
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6781008931576433e-05,
+      "loss": 1.3174,
+      "step": 436
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676543463331946e-05,
+      "loss": 1.1726,
+      "step": 437
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6749830015182106e-05,
+      "loss": 1.2188,
+      "step": 438
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6734195147097815e-05,
+      "loss": 1.2124,
+      "step": 439
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6718530099135596e-05,
+      "loss": 1.1971,
+      "step": 440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6702834941499718e-05,
+      "loss": 1.1957,
+      "step": 441
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6687109744529394e-05,
+      "loss": 1.2282,
+      "step": 442
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6671354578698444e-05,
+      "loss": 1.2366,
+      "step": 443
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665556951461501e-05,
+      "loss": 1.1922,
+      "step": 444
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6639754623021227e-05,
+      "loss": 1.257,
+      "step": 445
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6623909974792888e-05,
+      "loss": 1.2759,
+      "step": 446
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6608035640939165e-05,
+      "loss": 1.2557,
+      "step": 447
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6592131692602257e-05,
+      "loss": 1.1747,
+      "step": 448
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657619820105708e-05,
+      "loss": 1.2136,
+      "step": 449
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.656023523771095e-05,
+      "loss": 1.3245,
+      "step": 450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6544242874103278e-05,
+      "loss": 1.1963,
+      "step": 451
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6528221181905217e-05,
+      "loss": 1.2523,
+      "step": 452
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6512170232919372e-05,
+      "loss": 1.2043,
+      "step": 453
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6496090099079452e-05,
+      "loss": 1.2063,
+      "step": 454
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6479980852449974e-05,
+      "loss": 1.0912,
+      "step": 455
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6463842565225914e-05,
+      "loss": 1.1231,
+      "step": 456
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6447675309732403e-05,
+      "loss": 1.2476,
+      "step": 457
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64314791584244e-05,
+      "loss": 1.1912,
+      "step": 458
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6415254183886356e-05,
+      "loss": 1.2297,
+      "step": 459
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6399000458831893e-05,
+      "loss": 1.1678,
+      "step": 460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6382718056103486e-05,
+      "loss": 1.227,
+      "step": 461
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6366407048672135e-05,
+      "loss": 1.1352,
+      "step": 462
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6350067509637028e-05,
+      "loss": 1.2545,
+      "step": 463
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6333699512225224e-05,
+      "loss": 1.1916,
+      "step": 464
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6317303129791315e-05,
+      "loss": 1.1412,
+      "step": 465
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6300878435817115e-05,
+      "loss": 1.2384,
+      "step": 466
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.62844255039113e-05,
+      "loss": 1.2379,
+      "step": 467
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626794440780911e-05,
+      "loss": 1.1708,
+      "step": 468
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6251435221372007e-05,
+      "loss": 1.1741,
+      "step": 469
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6234898018587336e-05,
+      "loss": 1.14,
+      "step": 470
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6218332873568e-05,
+      "loss": 1.1982,
+      "step": 471
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620173986055213e-05,
+      "loss": 1.1246,
+      "step": 472
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6185119053902748e-05,
+      "loss": 1.1792,
+      "step": 473
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616847052810744e-05,
+      "loss": 1.2365,
+      "step": 474
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6151794357778006e-05,
+      "loss": 1.1503,
+      "step": 475
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613509061765015e-05,
+      "loss": 1.1668,
+      "step": 476
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.611835938258313e-05,
+      "loss": 1.1859,
+      "step": 477
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6101600727559423e-05,
+      "loss": 1.1515,
+      "step": 478
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6084814727684386e-05,
+      "loss": 1.1114,
+      "step": 479
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6068001458185934e-05,
+      "loss": 1.1375,
+      "step": 480
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6051160994414187e-05,
+      "loss": 1.2151,
+      "step": 481
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.603429341184114e-05,
+      "loss": 1.2092,
+      "step": 482
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601739878606032e-05,
+      "loss": 1.1854,
+      "step": 483
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600047719278645e-05,
+      "loss": 1.1809,
+      "step": 484
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5983528707855128e-05,
+      "loss": 1.1968,
+      "step": 485
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596655340722244e-05,
+      "loss": 1.1921,
+      "step": 486
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5949551366964675e-05,
+      "loss": 1.1509,
+      "step": 487
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593252266327794e-05,
+      "loss": 1.2754,
+      "step": 488
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5915467372477843e-05,
+      "loss": 1.1864,
+      "step": 489
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5898385570999146e-05,
+      "loss": 1.2278,
+      "step": 490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5881277335395425e-05,
+      "loss": 1.2407,
+      "step": 491
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5864142742338707e-05,
+      "loss": 1.2947,
+      "step": 492
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584698186861916e-05,
+      "loss": 1.1966,
+      "step": 493
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5829794791144723e-05,
+      "loss": 1.1708,
+      "step": 494
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5812581586940767e-05,
+      "loss": 1.2184,
+      "step": 495
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5795342333149757e-05,
+      "loss": 1.2204,
+      "step": 496
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.57780771070309e-05,
+      "loss": 1.2833,
+      "step": 497
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.57607859859598e-05,
+      "loss": 1.1883,
+      "step": 498
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5743469047428113e-05,
+      "loss": 1.2125,
+      "step": 499
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572612636904319e-05,
+      "loss": 1.1894,
+      "step": 500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5708758028527754e-05,
+      "loss": 1.1902,
+      "step": 501
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5691364103719515e-05,
+      "loss": 1.2049,
+      "step": 502
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5673944672570858e-05,
+      "loss": 1.2759,
+      "step": 503
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565649981314847e-05,
+      "loss": 1.1998,
+      "step": 504
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5639029603632995e-05,
+      "loss": 1.1195,
+      "step": 505
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5621534122318682e-05,
+      "loss": 1.1845,
+      "step": 506
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5604013447613057e-05,
+      "loss": 1.2123,
+      "step": 507
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5586467658036526e-05,
+      "loss": 1.0993,
+      "step": 508
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556889683222207e-05,
+      "loss": 1.2664,
+      "step": 509
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5551301048914863e-05,
+      "loss": 1.1448,
+      "step": 510
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5533680386971927e-05,
+      "loss": 1.2405,
+      "step": 511
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5516034925361785e-05,
+      "loss": 1.2208,
+      "step": 512
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5498364743164098e-05,
+      "loss": 1.2041,
+      "step": 513
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5480669919569313e-05,
+      "loss": 1.2938,
+      "step": 514
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5462950533878318e-05,
+      "loss": 1.0969,
+      "step": 515
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544520666550207e-05,
+      "loss": 1.2332,
+      "step": 516
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5427438393961246e-05,
+      "loss": 1.1775,
+      "step": 517
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.54096457988859e-05,
+      "loss": 1.1766,
+      "step": 518
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5391828960015074e-05,
+      "loss": 1.1892,
+      "step": 519
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.537398795719648e-05,
+      "loss": 1.0989,
+      "step": 520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5356122870386117e-05,
+      "loss": 1.173,
+      "step": 521
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533823377964791e-05,
+      "loss": 1.2476,
+      "step": 522
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5320320765153367e-05,
+      "loss": 1.1974,
+      "step": 523
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5302383907181212e-05,
+      "loss": 1.1981,
+      "step": 524
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5284423286117035e-05,
+      "loss": 1.2087,
+      "step": 525
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.52664389824529e-05,
+      "loss": 1.1894,
+      "step": 526
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.524843107678702e-05,
+      "loss": 1.1454,
+      "step": 527
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5230399649823389e-05,
+      "loss": 1.1228,
+      "step": 528
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.52123447823714e-05,
+      "loss": 1.2225,
+      "step": 529
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5194266555345505e-05,
+      "loss": 1.1987,
+      "step": 530
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5176165049764846e-05,
+      "loss": 1.1753,
+      "step": 531
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5158040346752878e-05,
+      "loss": 1.2879,
+      "step": 532
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.513989252753703e-05,
+      "loss": 1.198,
+      "step": 533
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5121721673448319e-05,
+      "loss": 1.1642,
+      "step": 534
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5103527865921005e-05,
+      "loss": 1.1747,
+      "step": 535
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5085311186492206e-05,
+      "loss": 1.2151,
+      "step": 536
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5067071716801546e-05,
+      "loss": 1.2433,
+      "step": 537
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5048809538590789e-05,
+      "loss": 1.1899,
+      "step": 538
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5030524733703464e-05,
+      "loss": 1.1452,
+      "step": 539
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5012217384084501e-05,
+      "loss": 1.27,
+      "step": 540
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4993887571779877e-05,
+      "loss": 1.2202,
+      "step": 541
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4975535378936228e-05,
+      "loss": 1.2659,
+      "step": 542
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4957160887800494e-05,
+      "loss": 1.1568,
+      "step": 543
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4938764180719548e-05,
+      "loss": 1.1203,
+      "step": 544
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4920345340139819e-05,
+      "loss": 1.2977,
+      "step": 545
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.490190444860694e-05,
+      "loss": 1.2189,
+      "step": 546
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4883441588765355e-05,
+      "loss": 1.1423,
+      "step": 547
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4864956843357967e-05,
+      "loss": 1.1703,
+      "step": 548
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4846450295225769e-05,
+      "loss": 1.156,
+      "step": 549
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.482792202730745e-05,
+      "loss": 1.1884,
+      "step": 550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4809372122639047e-05,
+      "loss": 1.0674,
+      "step": 551
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4790800664353562e-05,
+      "loss": 1.2705,
+      "step": 552
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.477220773568059e-05,
+      "loss": 1.1745,
+      "step": 553
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.475359341994595e-05,
+      "loss": 1.1846,
+      "step": 554
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.473495780057131e-05,
+      "loss": 1.1643,
+      "step": 555
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4716300961073808e-05,
+      "loss": 1.1691,
+      "step": 556
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.469762298506568e-05,
+      "loss": 1.246,
+      "step": 557
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4678923956253894e-05,
+      "loss": 1.18,
+      "step": 558
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4660203958439768e-05,
+      "loss": 1.1507,
+      "step": 559
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4641463075518587e-05,
+      "loss": 1.0105,
+      "step": 560
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4622701391479237e-05,
+      "loss": 1.2385,
+      "step": 561
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.460391899040383e-05,
+      "loss": 1.1939,
+      "step": 562
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4585115956467322e-05,
+      "loss": 1.1724,
+      "step": 563
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4566292373937133e-05,
+      "loss": 1.117,
+      "step": 564
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4547448327172778e-05,
+      "loss": 1.2078,
+      "step": 565
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4528583900625481e-05,
+      "loss": 1.2424,
+      "step": 566
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.45096991788378e-05,
+      "loss": 1.1431,
+      "step": 567
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4490794246443249e-05,
+      "loss": 1.1496,
+      "step": 568
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4471869188165923e-05,
+      "loss": 1.0905,
+      "step": 569
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4452924088820101e-05,
+      "loss": 1.2019,
+      "step": 570
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4433959033309888e-05,
+      "loss": 1.1296,
+      "step": 571
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4414974106628816e-05,
+      "loss": 1.1794,
+      "step": 572
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4395969393859478e-05,
+      "loss": 1.1849,
+      "step": 573
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4376944980173138e-05,
+      "loss": 1.2448,
+      "step": 574
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.435790095082935e-05,
+      "loss": 1.1362,
+      "step": 575
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4338837391175582e-05,
+      "loss": 1.24,
+      "step": 576
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4319754386646824e-05,
+      "loss": 1.1808,
+      "step": 577
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4300652022765207e-05,
+      "loss": 1.2099,
+      "step": 578
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4281530385139627e-05,
+      "loss": 1.3174,
+      "step": 579
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4262389559465364e-05,
+      "loss": 1.2025,
+      "step": 580
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4243229631523678e-05,
+      "loss": 1.1764,
+      "step": 581
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4224050687181442e-05,
+      "loss": 1.1645,
+      "step": 582
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4204852812390757e-05,
+      "loss": 1.094,
+      "step": 583
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4185636093188558e-05,
+      "loss": 1.1105,
+      "step": 584
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4166400615696232e-05,
+      "loss": 1.1215,
+      "step": 585
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4147146466119235e-05,
+      "loss": 1.1906,
+      "step": 586
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4127873730746706e-05,
+      "loss": 1.2395,
+      "step": 587
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4108582495951077e-05,
+      "loss": 1.1686,
+      "step": 588
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4089272848187678e-05,
+      "loss": 1.075,
+      "step": 589
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.406994487399437e-05,
+      "loss": 1.1651,
+      "step": 590
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4050598659991144e-05,
+      "loss": 1.1707,
+      "step": 591
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4031234292879726e-05,
+      "loss": 1.1978,
+      "step": 592
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.40118518594432e-05,
+      "loss": 1.2424,
+      "step": 593
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3992451446545624e-05,
+      "loss": 1.2487,
+      "step": 594
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.397303314113162e-05,
+      "loss": 1.2713,
+      "step": 595
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3953597030226006e-05,
+      "loss": 1.1485,
+      "step": 596
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3934143200933398e-05,
+      "loss": 1.19,
+      "step": 597
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3914671740437811e-05,
+      "loss": 1.1636,
+      "step": 598
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3895182736002276e-05,
+      "loss": 1.2896,
+      "step": 599
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3875676274968453e-05,
+      "loss": 1.2035,
+      "step": 600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3856152444756235e-05,
+      "loss": 1.1361,
+      "step": 601
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3836611332863356e-05,
+      "loss": 1.1512,
+      "step": 602
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3817053026865e-05,
+      "loss": 1.1591,
+      "step": 603
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3797477614413404e-05,
+      "loss": 1.0688,
+      "step": 604
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3777885183237473e-05,
+      "loss": 1.1533,
+      "step": 605
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3758275821142382e-05,
+      "loss": 1.2396,
+      "step": 606
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3738649616009183e-05,
+      "loss": 1.0865,
+      "step": 607
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3719006655794414e-05,
+      "loss": 1.1845,
+      "step": 608
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3699347028529694e-05,
+      "loss": 1.1594,
+      "step": 609
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3679670822321347e-05,
+      "loss": 1.1641,
+      "step": 610
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3659978125349991e-05,
+      "loss": 1.177,
+      "step": 611
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3640269025870152e-05,
+      "loss": 1.1944,
+      "step": 612
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3620543612209861e-05,
+      "loss": 1.1914,
+      "step": 613
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3600801972770272e-05,
+      "loss": 1.1376,
+      "step": 614
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3581044196025243e-05,
+      "loss": 1.1693,
+      "step": 615
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3561270370520957e-05,
+      "loss": 1.2032,
+      "step": 616
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3541480584875531e-05,
+      "loss": 1.2004,
+      "step": 617
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3521674927778594e-05,
+      "loss": 1.2486,
+      "step": 618
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3501853487990917e-05,
+      "loss": 1.1432,
+      "step": 619
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.348201635434399e-05,
+      "loss": 1.2211,
+      "step": 620
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3462163615739647e-05,
+      "loss": 1.2249,
+      "step": 621
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3442295361149651e-05,
+      "loss": 1.1785,
+      "step": 622
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3422411679615305e-05,
+      "loss": 1.1642,
+      "step": 623
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3402512660247046e-05,
+      "loss": 1.2141,
+      "step": 624
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3382598392224053e-05,
+      "loss": 1.1936,
+      "step": 625
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.336266896479384e-05,
+      "loss": 1.1162,
+      "step": 626
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3342724467271852e-05,
+      "loss": 1.2254,
+      "step": 627
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3322764989041086e-05,
+      "loss": 1.2393,
+      "step": 628
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3302790619551673e-05,
+      "loss": 1.0991,
+      "step": 629
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.328280144832047e-05,
+      "loss": 1.2194,
+      "step": 630
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.326279756493068e-05,
+      "loss": 1.1406,
+      "step": 631
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3242779059031436e-05,
+      "loss": 1.1808,
+      "step": 632
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3222746020337406e-05,
+      "loss": 1.1802,
+      "step": 633
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3202698538628376e-05,
+      "loss": 1.2555,
+      "step": 634
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3182636703748881e-05,
+      "loss": 1.1126,
+      "step": 635
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3162560605607763e-05,
+      "loss": 1.1475,
+      "step": 636
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3142470334177794e-05,
+      "loss": 1.1588,
+      "step": 637
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3122365979495259e-05,
+      "loss": 1.103,
+      "step": 638
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3102247631659564e-05,
+      "loss": 1.1317,
+      "step": 639
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3082115380832829e-05,
+      "loss": 1.1694,
+      "step": 640
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.306196931723947e-05,
+      "loss": 1.1437,
+      "step": 641
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3041809531165819e-05,
+      "loss": 1.1556,
+      "step": 642
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3021636112959696e-05,
+      "loss": 1.1503,
+      "step": 643
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3001449153030024e-05,
+      "loss": 1.1503,
+      "step": 644
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2981248741846404e-05,
+      "loss": 1.1144,
+      "step": 645
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2961034969938732e-05,
+      "loss": 1.2132,
+      "step": 646
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2940807927896773e-05,
+      "loss": 1.182,
+      "step": 647
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.292056770636976e-05,
+      "loss": 1.1225,
+      "step": 648
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2900314396066e-05,
+      "loss": 1.0938,
+      "step": 649
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.288004808775246e-05,
+      "loss": 1.0935,
+      "step": 650
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2859768872254344e-05,
+      "loss": 1.2288,
+      "step": 651
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2839476840454714e-05,
+      "loss": 1.23,
+      "step": 652
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2819172083294068e-05,
+      "loss": 1.1665,
+      "step": 653
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2798854691769927e-05,
+      "loss": 1.207,
+      "step": 654
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2778524756936438e-05,
+      "loss": 1.1237,
+      "step": 655
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2758182369903969e-05,
+      "loss": 1.0907,
+      "step": 656
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2737827621838678e-05,
+      "loss": 1.2161,
+      "step": 657
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2717460603962132e-05,
+      "loss": 1.12,
+      "step": 658
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2697081407550886e-05,
+      "loss": 1.1566,
+      "step": 659
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2676690123936066e-05,
+      "loss": 1.2275,
+      "step": 660
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2656286844502978e-05,
+      "loss": 1.1194,
+      "step": 661
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2635871660690677e-05,
+      "loss": 1.1361,
+      "step": 662
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.261544466399158e-05,
+      "loss": 1.1951,
+      "step": 663
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.259500594595104e-05,
+      "loss": 1.143,
+      "step": 664
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2574555598166932e-05,
+      "loss": 1.0943,
+      "step": 665
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2554093712289267e-05,
+      "loss": 1.2183,
+      "step": 666
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2533620380019752e-05,
+      "loss": 1.1799,
+      "step": 667
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2513135693111399e-05,
+      "loss": 1.1475,
+      "step": 668
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2492639743368096e-05,
+      "loss": 1.1951,
+      "step": 669
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2472132622644222e-05,
+      "loss": 1.2485,
+      "step": 670
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2451614422844208e-05,
+      "loss": 1.1272,
+      "step": 671
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2431085235922143e-05,
+      "loss": 1.2868,
+      "step": 672
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2410545153881355e-05,
+      "loss": 1.2332,
+      "step": 673
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2389994268773995e-05,
+      "loss": 1.277,
+      "step": 674
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2369432672700634e-05,
+      "loss": 1.1355,
+      "step": 675
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.234886045780984e-05,
+      "loss": 1.1973,
+      "step": 676
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2328277716297771e-05,
+      "loss": 1.2391,
+      "step": 677
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2307684540407775e-05,
+      "loss": 1.2523,
+      "step": 678
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2287081022429935e-05,
+      "loss": 1.1896,
+      "step": 679
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2266467254700712e-05,
+      "loss": 1.1751,
+      "step": 680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.224584332960248e-05,
+      "loss": 1.1977,
+      "step": 681
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2225209339563144e-05,
+      "loss": 1.1177,
+      "step": 682
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.220456537705572e-05,
+      "loss": 1.0793,
+      "step": 683
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2183911534597908e-05,
+      "loss": 1.2298,
+      "step": 684
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2163247904751688e-05,
+      "loss": 1.1423,
+      "step": 685
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2142574580122903e-05,
+      "loss": 1.1471,
+      "step": 686
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2121891653360851e-05,
+      "loss": 1.2614,
+      "step": 687
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.210119921715785e-05,
+      "loss": 1.182,
+      "step": 688
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2080497364248849e-05,
+      "loss": 1.1426,
+      "step": 689
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2059786187410984e-05,
+      "loss": 1.1813,
+      "step": 690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2039065779463195e-05,
+      "loss": 1.0127,
+      "step": 691
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2018336233265775e-05,
+      "loss": 1.1333,
+      "step": 692
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1997597641719976e-05,
+      "loss": 1.1618,
+      "step": 693
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1976850097767598e-05,
+      "loss": 1.1588,
+      "step": 694
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1956093694390547e-05,
+      "loss": 1.2159,
+      "step": 695
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1935328524610443e-05,
+      "loss": 1.148,
+      "step": 696
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1914554681488188e-05,
+      "loss": 1.1849,
+      "step": 697
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1893772258123554e-05,
+      "loss": 1.1439,
+      "step": 698
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1872981347654778e-05,
+      "loss": 1.1562,
+      "step": 699
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1852182043258112e-05,
+      "loss": 1.0943,
+      "step": 700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1831374438147445e-05,
+      "loss": 1.0538,
+      "step": 701
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1810558625573856e-05,
+      "loss": 1.1594,
+      "step": 702
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1789734698825209e-05,
+      "loss": 1.1912,
+      "step": 703
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.176890275122573e-05,
+      "loss": 1.1694,
+      "step": 704
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.17480628761356e-05,
+      "loss": 1.1504,
+      "step": 705
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1727215166950519e-05,
+      "loss": 1.2278,
+      "step": 706
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1706359717101295e-05,
+      "loss": 1.1468,
+      "step": 707
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1685496620053434e-05,
+      "loss": 1.2667,
+      "step": 708
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1664625969306707e-05,
+      "loss": 1.0349,
+      "step": 709
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1643747858394743e-05,
+      "loss": 1.1696,
+      "step": 710
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.16228623808846e-05,
+      "loss": 1.1717,
+      "step": 711
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1601969630376353e-05,
+      "loss": 1.1546,
+      "step": 712
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.158106970050267e-05,
+      "loss": 1.1034,
+      "step": 713
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.156016268492839e-05,
+      "loss": 1.1305,
+      "step": 714
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1539248677350114e-05,
+      "loss": 1.1802,
+      "step": 715
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.151832777149578e-05,
+      "loss": 1.2072,
+      "step": 716
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1497400061124231e-05,
+      "loss": 1.1901,
+      "step": 717
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1476465640024814e-05,
+      "loss": 1.1725,
+      "step": 718
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1455524602016946e-05,
+      "loss": 1.1871,
+      "step": 719
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1434577040949699e-05,
+      "loss": 1.2033,
+      "step": 720
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1413623050701381e-05,
+      "loss": 1.1906,
+      "step": 721
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1392662725179114e-05,
+      "loss": 1.2073,
+      "step": 722
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1371696158318404e-05,
+      "loss": 1.135,
+      "step": 723
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1350723444082736e-05,
+      "loss": 1.1548,
+      "step": 724
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1329744676463144e-05,
+      "loss": 1.1163,
+      "step": 725
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1308759949477786e-05,
+      "loss": 1.1286,
+      "step": 726
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1287769357171536e-05,
+      "loss": 1.1423,
+      "step": 727
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1266772993615543e-05,
+      "loss": 1.1635,
+      "step": 728
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1245770952906829e-05,
+      "loss": 1.1653,
+      "step": 729
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1224763329167859e-05,
+      "loss": 1.1401,
+      "step": 730
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1203750216546105e-05,
+      "loss": 1.1561,
+      "step": 731
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1182731709213658e-05,
+      "loss": 1.1302,
+      "step": 732
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1161707901366776e-05,
+      "loss": 1.1773,
+      "step": 733
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1140678887225468e-05,
+      "loss": 1.1528,
+      "step": 734
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1119644761033079e-05,
+      "loss": 1.0764,
+      "step": 735
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1098605617055871e-05,
+      "loss": 1.0947,
+      "step": 736
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1077561549582578e-05,
+      "loss": 1.1801,
+      "step": 737
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1056512652924014e-05,
+      "loss": 1.1466,
+      "step": 738
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1035459021412631e-05,
+      "loss": 1.1361,
+      "step": 739
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1014400749402096e-05,
+      "loss": 1.1022,
+      "step": 740
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0993337931266879e-05,
+      "loss": 1.0657,
+      "step": 741
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0972270661401812e-05,
+      "loss": 1.1162,
+      "step": 742
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.09511990342217e-05,
+      "loss": 1.1556,
+      "step": 743
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0930123144160857e-05,
+      "loss": 1.1398,
+      "step": 744
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0909043085672703e-05,
+      "loss": 1.1445,
+      "step": 745
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0887958953229349e-05,
+      "loss": 1.1551,
+      "step": 746
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0866870841321156e-05,
+      "loss": 1.2261,
+      "step": 747
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0845778844456319e-05,
+      "loss": 1.1775,
+      "step": 748
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0824683057160445e-05,
+      "loss": 1.3162,
+      "step": 749
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0803583573976137e-05,
+      "loss": 1.2561,
+      "step": 750
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0782480489462546e-05,
+      "loss": 1.1989,
+      "step": 751
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0761373898194972e-05,
+      "loss": 1.0945,
+      "step": 752
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.074026389476443e-05,
+      "loss": 1.1633,
+      "step": 753
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0719150573777226e-05,
+      "loss": 1.2924,
+      "step": 754
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0698034029854533e-05,
+      "loss": 1.1764,
+      "step": 755
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0676914357631968e-05,
+      "loss": 1.2571,
+      "step": 756
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0655791651759174e-05,
+      "loss": 1.1202,
+      "step": 757
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0634666006899375e-05,
+      "loss": 1.2447,
+      "step": 758
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.061353751772898e-05,
+      "loss": 1.1389,
+      "step": 759
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0592406278937143e-05,
+      "loss": 1.2367,
+      "step": 760
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0571272385225338e-05,
+      "loss": 1.0912,
+      "step": 761
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.055013593130693e-05,
+      "loss": 1.0857,
+      "step": 762
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0528997011906775e-05,
+      "loss": 1.0864,
+      "step": 763
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0507855721760757e-05,
+      "loss": 1.1322,
+      "step": 764
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0486712155615408e-05,
+      "loss": 1.0751,
+      "step": 765
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.046556640822744e-05,
+      "loss": 1.1866,
+      "step": 766
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0444418574363354e-05,
+      "loss": 1.0581,
+      "step": 767
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0423268748798992e-05,
+      "loss": 1.2088,
+      "step": 768
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0402117026319128e-05,
+      "loss": 1.154,
+      "step": 769
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0380963501717034e-05,
+      "loss": 1.1151,
+      "step": 770
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0359808269794063e-05,
+      "loss": 1.0969,
+      "step": 771
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0338651425359217e-05,
+      "loss": 1.0839,
+      "step": 772
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0317493063228716e-05,
+      "loss": 1.1155,
+      "step": 773
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0296333278225599e-05,
+      "loss": 1.0979,
+      "step": 774
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0275172165179268e-05,
+      "loss": 1.1974,
+      "step": 775
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0254009818925082e-05,
+      "loss": 1.1419,
+      "step": 776
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0232846334303925e-05,
+      "loss": 1.0999,
+      "step": 777
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0211681806161787e-05,
+      "loss": 1.1071,
+      "step": 778
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0190516329349328e-05,
+      "loss": 1.1991,
+      "step": 779
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0169349998721459e-05,
+      "loss": 1.1723,
+      "step": 780
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0148182909136928e-05,
+      "loss": 1.0594,
+      "step": 781
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0127015155457875e-05,
+      "loss": 1.1058,
+      "step": 782
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0105846832549417e-05,
+      "loss": 1.1558,
+      "step": 783
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0084678035279227e-05,
+      "loss": 0.9988,
+      "step": 784
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.00635088585171e-05,
+      "loss": 1.1597,
+      "step": 785
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0042339397134528e-05,
+      "loss": 1.2476,
+      "step": 786
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0021169746004287e-05,
+      "loss": 1.0788,
+      "step": 787
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1e-05,
+      "loss": 1.1749,
+      "step": 788
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.978830253995713e-06,
+      "loss": 1.2155,
+      "step": 789
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.957660602865477e-06,
+      "loss": 1.0134,
+      "step": 790
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.936491141482902e-06,
+      "loss": 1.1299,
+      "step": 791
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.915321964720773e-06,
+      "loss": 1.2221,
+      "step": 792
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.894153167450586e-06,
+      "loss": 1.1356,
+      "step": 793
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.872984844542128e-06,
+      "loss": 1.161,
+      "step": 794
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.851817090863073e-06,
+      "loss": 1.1427,
+      "step": 795
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.830650001278543e-06,
+      "loss": 1.1405,
+      "step": 796
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.809483670650676e-06,
+      "loss": 1.0917,
+      "step": 797
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.788318193838218e-06,
+      "loss": 1.1756,
+      "step": 798
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.767153665696077e-06,
+      "loss": 1.1769,
+      "step": 799
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.74599018107492e-06,
+      "loss": 1.0569,
+      "step": 800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.724827834820736e-06,
+      "loss": 1.1441,
+      "step": 801
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.703666721774403e-06,
+      "loss": 1.0548,
+      "step": 802
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.682506936771284e-06,
+      "loss": 1.116,
+      "step": 803
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.66134857464079e-06,
+      "loss": 1.1596,
+      "step": 804
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.64019173020594e-06,
+      "loss": 1.1394,
+      "step": 805
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.619036498282968e-06,
+      "loss": 1.0937,
+      "step": 806
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.597882973680877e-06,
+      "loss": 1.1612,
+      "step": 807
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.57673125120101e-06,
+      "loss": 1.1387,
+      "step": 808
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.555581425636648e-06,
+      "loss": 1.1419,
+      "step": 809
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.534433591772562e-06,
+      "loss": 1.1038,
+      "step": 810
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.513287844384594e-06,
+      "loss": 1.1451,
+      "step": 811
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.492144278239244e-06,
+      "loss": 1.1563,
+      "step": 812
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.47100298809323e-06,
+      "loss": 1.115,
+      "step": 813
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.449864068693072e-06,
+      "loss": 1.0893,
+      "step": 814
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.428727614774667e-06,
+      "loss": 1.1369,
+      "step": 815
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.407593721062858e-06,
+      "loss": 1.1946,
+      "step": 816
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.38646248227102e-06,
+      "loss": 1.1654,
+      "step": 817
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.365333993100628e-06,
+      "loss": 1.1397,
+      "step": 818
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.34420834824083e-06,
+      "loss": 1.2131,
+      "step": 819
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.323085642368032e-06,
+      "loss": 1.1222,
+      "step": 820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.301965970145472e-06,
+      "loss": 1.22,
+      "step": 821
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.280849426222778e-06,
+      "loss": 1.2111,
+      "step": 822
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.259736105235572e-06,
+      "loss": 1.123,
+      "step": 823
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.23862610180503e-06,
+      "loss": 1.1623,
+      "step": 824
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.217519510537455e-06,
+      "loss": 1.1052,
+      "step": 825
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.196416426023868e-06,
+      "loss": 1.1576,
+      "step": 826
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.175316942839557e-06,
+      "loss": 1.0609,
+      "step": 827
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.154221155543684e-06,
+      "loss": 1.1462,
+      "step": 828
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.133129158678848e-06,
+      "loss": 1.1307,
+      "step": 829
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.112041046770653e-06,
+      "loss": 1.1345,
+      "step": 830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.090956914327297e-06,
+      "loss": 1.1544,
+      "step": 831
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.069876855839148e-06,
+      "loss": 1.1459,
+      "step": 832
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.048800965778304e-06,
+      "loss": 1.0818,
+      "step": 833
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.027729338598188e-06,
+      "loss": 1.1615,
+      "step": 834
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.006662068733128e-06,
+      "loss": 1.0819,
+      "step": 835
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.985599250597906e-06,
+      "loss": 1.0893,
+      "step": 836
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.96454097858737e-06,
+      "loss": 1.0602,
+      "step": 837
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.943487347075988e-06,
+      "loss": 1.1368,
+      "step": 838
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.922438450417424e-06,
+      "loss": 1.1827,
+      "step": 839
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.901394382944135e-06,
+      "loss": 1.0157,
+      "step": 840
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.880355238966923e-06,
+      "loss": 1.094,
+      "step": 841
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.859321112774535e-06,
+      "loss": 1.0768,
+      "step": 842
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.838292098633229e-06,
+      "loss": 1.112,
+      "step": 843
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.817268290786343e-06,
+      "loss": 1.1027,
+      "step": 844
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.796249783453897e-06,
+      "loss": 1.067,
+      "step": 845
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.775236670832146e-06,
+      "loss": 1.1478,
+      "step": 846
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.754229047093173e-06,
+      "loss": 1.1377,
+      "step": 847
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.733227006384459e-06,
+      "loss": 1.1327,
+      "step": 848
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.71223064282847e-06,
+      "loss": 1.2245,
+      "step": 849
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.691240050522215e-06,
+      "loss": 1.23,
+      "step": 850
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.670255323536858e-06,
+      "loss": 1.0874,
+      "step": 851
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.649276555917267e-06,
+      "loss": 1.0935,
+      "step": 852
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.6283038416816e-06,
+      "loss": 1.0869,
+      "step": 853
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.607337274820888e-06,
+      "loss": 1.1394,
+      "step": 854
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.586376949298622e-06,
+      "loss": 1.246,
+      "step": 855
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.565422959050303e-06,
+      "loss": 1.1507,
+      "step": 856
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.54447539798306e-06,
+      "loss": 1.0539,
+      "step": 857
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.52353435997519e-06,
+      "loss": 1.1779,
+      "step": 858
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.50259993887577e-06,
+      "loss": 1.1256,
+      "step": 859
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.481672228504224e-06,
+      "loss": 1.0908,
+      "step": 860
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.460751322649887e-06,
+      "loss": 1.15,
+      "step": 861
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.439837315071612e-06,
+      "loss": 1.1486,
+      "step": 862
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.418930299497335e-06,
+      "loss": 1.1543,
+      "step": 863
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.39803036962365e-06,
+      "loss": 1.1491,
+      "step": 864
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.377137619115401e-06,
+      "loss": 1.1429,
+      "step": 865
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.35625214160526e-06,
+      "loss": 1.1171,
+      "step": 866
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.335374030693295e-06,
+      "loss": 1.0617,
+      "step": 867
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.314503379946569e-06,
+      "loss": 1.0737,
+      "step": 868
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.293640282898709e-06,
+      "loss": 1.1907,
+      "step": 869
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.272784833049485e-06,
+      "loss": 1.165,
+      "step": 870
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.251937123864405e-06,
+      "loss": 1.132,
+      "step": 871
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.231097248774273e-06,
+      "loss": 1.1441,
+      "step": 872
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.210265301174796e-06,
+      "loss": 1.1223,
+      "step": 873
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.18944137442615e-06,
+      "loss": 1.2506,
+      "step": 874
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.16862556185256e-06,
+      "loss": 1.1089,
+      "step": 875
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.14781795674189e-06,
+      "loss": 1.1689,
+      "step": 876
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.12701865234523e-06,
+      "loss": 1.1346,
+      "step": 877
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.106227741876447e-06,
+      "loss": 1.0872,
+      "step": 878
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.085445318511814e-06,
+      "loss": 1.1677,
+      "step": 879
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.064671475389562e-06,
+      "loss": 1.223,
+      "step": 880
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.043906305609454e-06,
+      "loss": 1.0917,
+      "step": 881
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.023149902232404e-06,
+      "loss": 1.1277,
+      "step": 882
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.002402358280028e-06,
+      "loss": 1.0845,
+      "step": 883
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.981663766734228e-06,
+      "loss": 1.1018,
+      "step": 884
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.960934220536812e-06,
+      "loss": 1.1171,
+      "step": 885
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.940213812589018e-06,
+      "loss": 1.1786,
+      "step": 886
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.919502635751154e-06,
+      "loss": 1.1101,
+      "step": 887
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.898800782842153e-06,
+      "loss": 1.1189,
+      "step": 888
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.878108346639152e-06,
+      "loss": 1.0901,
+      "step": 889
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.857425419877097e-06,
+      "loss": 1.1288,
+      "step": 890
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.836752095248318e-06,
+      "loss": 1.2,
+      "step": 891
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.816088465402097e-06,
+      "loss": 1.1284,
+      "step": 892
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.795434622944282e-06,
+      "loss": 1.1544,
+      "step": 893
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.774790660436857e-06,
+      "loss": 1.1624,
+      "step": 894
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.754156670397522e-06,
+      "loss": 1.0516,
+      "step": 895
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.73353274529929e-06,
+      "loss": 1.1343,
+      "step": 896
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.712918977570067e-06,
+      "loss": 1.1037,
+      "step": 897
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.69231545959223e-06,
+      "loss": 1.1554,
+      "step": 898
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.671722283702227e-06,
+      "loss": 1.0409,
+      "step": 899
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.651139542190164e-06,
+      "loss": 1.1872,
+      "step": 900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.63056732729937e-06,
+      "loss": 1.0072,
+      "step": 901
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.610005731226009e-06,
+      "loss": 1.1388,
+      "step": 902
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.589454846118648e-06,
+      "loss": 1.0879,
+      "step": 903
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.568914764077857e-06,
+      "loss": 1.1612,
+      "step": 904
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.548385577155795e-06,
+      "loss": 1.1668,
+      "step": 905
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.52786737735578e-06,
+      "loss": 1.0886,
+      "step": 906
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5073602566319045e-06,
+      "loss": 1.1748,
+      "step": 907
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.486864306888608e-06,
+      "loss": 0.9568,
+      "step": 908
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.466379619980251e-06,
+      "loss": 1.0956,
+      "step": 909
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.445906287710733e-06,
+      "loss": 1.1372,
+      "step": 910
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4254444018330685e-06,
+      "loss": 1.1351,
+      "step": 911
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.404994054048962e-06,
+      "loss": 1.0904,
+      "step": 912
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.384555336008419e-06,
+      "loss": 1.1904,
+      "step": 913
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.364128339309326e-06,
+      "loss": 1.0676,
+      "step": 914
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3437131554970254e-06,
+      "loss": 1.1349,
+      "step": 915
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.323309876063936e-06,
+      "loss": 1.0683,
+      "step": 916
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.302918592449116e-06,
+      "loss": 1.117,
+      "step": 917
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.282539396037868e-06,
+      "loss": 1.1506,
+      "step": 918
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.262172378161326e-06,
+      "loss": 1.1416,
+      "step": 919
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.241817630096035e-06,
+      "loss": 1.1772,
+      "step": 920
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2214752430635625e-06,
+      "loss": 1.1078,
+      "step": 921
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.201145308230075e-06,
+      "loss": 1.1346,
+      "step": 922
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.180827916705935e-06,
+      "loss": 1.1915,
+      "step": 923
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.160523159545285e-06,
+      "loss": 1.1097,
+      "step": 924
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.140231127745661e-06,
+      "loss": 1.146,
+      "step": 925
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.119951912247545e-06,
+      "loss": 1.1206,
+      "step": 926
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.099685603934e-06,
+      "loss": 1.0996,
+      "step": 927
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.079432293630244e-06,
+      "loss": 1.1719,
+      "step": 928
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0591920721032314e-06,
+      "loss": 1.1842,
+      "step": 929
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.038965030061273e-06,
+      "loss": 1.1304,
+      "step": 930
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.018751258153599e-06,
+      "loss": 1.1575,
+      "step": 931
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.998550846969981e-06,
+      "loss": 1.1115,
+      "step": 932
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.978363887040308e-06,
+      "loss": 1.1116,
+      "step": 933
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9581904688341854e-06,
+      "loss": 1.1036,
+      "step": 934
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.938030682760532e-06,
+      "loss": 1.1774,
+      "step": 935
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.917884619167177e-06,
+      "loss": 1.0543,
+      "step": 936
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.897752368340438e-06,
+      "loss": 1.1196,
+      "step": 937
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8776340205047446e-06,
+      "loss": 1.0891,
+      "step": 938
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.857529665822211e-06,
+      "loss": 1.1248,
+      "step": 939
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.837439394392238e-06,
+      "loss": 1.1502,
+      "step": 940
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.817363296251119e-06,
+      "loss": 1.1279,
+      "step": 941
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.797301461371626e-06,
+      "loss": 1.0646,
+      "step": 942
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7772539796625995e-06,
+      "loss": 1.0768,
+      "step": 943
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.757220940968567e-06,
+      "loss": 1.1074,
+      "step": 944
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.737202435069323e-06,
+      "loss": 1.1293,
+      "step": 945
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7171985516795315e-06,
+      "loss": 1.1457,
+      "step": 946
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.697209380448333e-06,
+      "loss": 1.1421,
+      "step": 947
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.677235010958916e-06,
+      "loss": 1.0734,
+      "step": 948
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6572755327281506e-06,
+      "loss": 1.0765,
+      "step": 949
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.637331035206166e-06,
+      "loss": 1.0684,
+      "step": 950
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.617401607775948e-06,
+      "loss": 1.1704,
+      "step": 951
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.597487339752953e-06,
+      "loss": 1.0845,
+      "step": 952
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.577588320384698e-06,
+      "loss": 1.0984,
+      "step": 953
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.557704638850352e-06,
+      "loss": 1.0336,
+      "step": 954
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.537836384260356e-06,
+      "loss": 1.1241,
+      "step": 955
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.517983645656014e-06,
+      "loss": 1.1529,
+      "step": 956
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.498146512009087e-06,
+      "loss": 1.1041,
+      "step": 957
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4783250722214066e-06,
+      "loss": 1.0797,
+      "step": 958
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.458519415124473e-06,
+      "loss": 1.1976,
+      "step": 959
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.438729629479045e-06,
+      "loss": 1.164,
+      "step": 960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.418955803974763e-06,
+      "loss": 1.1359,
+      "step": 961
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.399198027229732e-06,
+      "loss": 1.0448,
+      "step": 962
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3794563877901385e-06,
+      "loss": 1.174,
+      "step": 963
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.359730974129852e-06,
+      "loss": 1.1929,
+      "step": 964
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.340021874650013e-06,
+      "loss": 1.0819,
+      "step": 965
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.320329177678656e-06,
+      "loss": 1.0704,
+      "step": 966
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.300652971470309e-06,
+      "loss": 1.1605,
+      "step": 967
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.28099334420559e-06,
+      "loss": 1.0867,
+      "step": 968
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.261350383990816e-06,
+      "loss": 1.1154,
+      "step": 969
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.241724178857621e-06,
+      "loss": 1.1594,
+      "step": 970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.22211481676253e-06,
+      "loss": 1.2517,
+      "step": 971
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.202522385586595e-06,
+      "loss": 1.0812,
+      "step": 972
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.182946973135004e-06,
+      "loss": 1.1563,
+      "step": 973
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.163388667136646e-06,
+      "loss": 1.0737,
+      "step": 974
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.14384755524377e-06,
+      "loss": 1.1234,
+      "step": 975
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.12432372503155e-06,
+      "loss": 1.1489,
+      "step": 976
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.104817263997727e-06,
+      "loss": 1.1966,
+      "step": 977
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.085328259562195e-06,
+      "loss": 1.1246,
+      "step": 978
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.065856799066604e-06,
+      "loss": 1.1106,
+      "step": 979
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.046402969773992e-06,
+      "loss": 1.1674,
+      "step": 980
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0269668588683815e-06,
+      "loss": 1.1928,
+      "step": 981
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.007548553454379e-06,
+      "loss": 1.1612,
+      "step": 982
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9881481405568005e-06,
+      "loss": 1.0279,
+      "step": 983
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.96876570712028e-06,
+      "loss": 1.0752,
+      "step": 984
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.94940134000886e-06,
+      "loss": 1.1377,
+      "step": 985
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.93005512600563e-06,
+      "loss": 1.1203,
+      "step": 986
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.910727151812324e-06,
+      "loss": 1.211,
+      "step": 987
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.891417504048926e-06,
+      "loss": 1.1416,
+      "step": 988
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.872126269253298e-06,
+      "loss": 1.0657,
+      "step": 989
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.852853533880768e-06,
+      "loss": 1.1387,
+      "step": 990
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8335993843037695e-06,
+      "loss": 1.0362,
+      "step": 991
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8143639068114465e-06,
+      "loss": 1.1358,
+      "step": 992
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7951471876092445e-06,
+      "loss": 1.0629,
+      "step": 993
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7759493128185584e-06,
+      "loss": 1.1024,
+      "step": 994
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.756770368476326e-06,
+      "loss": 1.1495,
+      "step": 995
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7376104405346376e-06,
+      "loss": 1.0856,
+      "step": 996
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7184696148603725e-06,
+      "loss": 1.0878,
+      "step": 997
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.699347977234799e-06,
+      "loss": 1.1247,
+      "step": 998
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.680245613353181e-06,
+      "loss": 1.1405,
+      "step": 999
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.66116260882442e-06,
+      "loss": 1.0988,
+      "step": 1000
+    }
+  ],
+  "max_steps": 1530,
+  "num_train_epochs": 1,
+  "total_flos": 1.339688795242496e+17,
+  "trial_name": null,
+  "trial_params": null
+}