diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,18016 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.5879317189360858,
+  "global_step": 3000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.042253521126761e-08,
+      "loss": 0.9991,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4084507042253522e-07,
+      "loss": 0.982,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1126760563380284e-07,
+      "loss": 1.0135,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8169014084507043e-07,
+      "loss": 0.9558,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.521126760563381e-07,
+      "loss": 1.0749,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.225352112676057e-07,
+      "loss": 1.0304,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.929577464788733e-07,
+      "loss": 0.9809,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.633802816901409e-07,
+      "loss": 0.9271,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.338028169014085e-07,
+      "loss": 0.911,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.042253521126762e-07,
+      "loss": 0.8964,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.746478873239437e-07,
+      "loss": 0.9058,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.450704225352114e-07,
+      "loss": 0.9289,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.154929577464789e-07,
+      "loss": 0.7715,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.859154929577465e-07,
+      "loss": 0.745,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0563380281690142e-06,
+      "loss": 0.7159,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1267605633802817e-06,
+      "loss": 0.698,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1971830985915492e-06,
+      "loss": 0.6938,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.267605633802817e-06,
+      "loss": 0.5836,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3380281690140844e-06,
+      "loss": 0.674,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4084507042253523e-06,
+      "loss": 0.6405,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4788732394366198e-06,
+      "loss": 0.5486,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5492957746478873e-06,
+      "loss": 0.5166,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6197183098591552e-06,
+      "loss": 0.338,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6901408450704227e-06,
+      "loss": 0.5036,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7605633802816902e-06,
+      "loss": 0.4492,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8309859154929579e-06,
+      "loss": 0.3996,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9014084507042254e-06,
+      "loss": 0.5513,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.971830985915493e-06,
+      "loss": 0.3885,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0422535211267608e-06,
+      "loss": 0.3533,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1126760563380285e-06,
+      "loss": 0.4806,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1830985915492958e-06,
+      "loss": 0.3584,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2535211267605635e-06,
+      "loss": 0.3971,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.323943661971831e-06,
+      "loss": 0.3297,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3943661971830984e-06,
+      "loss": 0.3274,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4647887323943666e-06,
+      "loss": 0.3345,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.535211267605634e-06,
+      "loss": 0.4462,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6056338028169015e-06,
+      "loss": 0.4242,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.676056338028169e-06,
+      "loss": 0.3258,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.746478873239437e-06,
+      "loss": 0.3861,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8169014084507046e-06,
+      "loss": 0.5329,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.887323943661972e-06,
+      "loss": 0.5445,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9577464788732396e-06,
+      "loss": 0.3269,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0281690140845073e-06,
+      "loss": 0.485,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0985915492957746e-06,
+      "loss": 0.2517,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1690140845070427e-06,
+      "loss": 0.3704,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2394366197183104e-06,
+      "loss": 0.2578,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3098591549295777e-06,
+      "loss": 0.2914,
+      "step": 47
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.3802816901408454e-06,
+      "loss": 0.3726,
+      "step": 48
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4507042253521127e-06,
+      "loss": 0.268,
+      "step": 49
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.5211267605633804e-06,
+      "loss": 0.5147,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.5915492957746485e-06,
+      "loss": 0.4073,
+      "step": 51
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6619718309859158e-06,
+      "loss": 0.4051,
+      "step": 52
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7323943661971835e-06,
+      "loss": 0.3266,
+      "step": 53
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.8028169014084508e-06,
+      "loss": 0.3473,
+      "step": 54
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.873239436619718e-06,
+      "loss": 0.3463,
+      "step": 55
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.943661971830986e-06,
+      "loss": 0.2735,
+      "step": 56
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.014084507042254e-06,
+      "loss": 0.3144,
+      "step": 57
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.0845070422535216e-06,
+      "loss": 0.3105,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.154929577464789e-06,
+      "loss": 0.4126,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.225352112676057e-06,
+      "loss": 0.3294,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.295774647887324e-06,
+      "loss": 0.2541,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3661971830985915e-06,
+      "loss": 0.2963,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.43661971830986e-06,
+      "loss": 0.2646,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.507042253521127e-06,
+      "loss": 0.3772,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.577464788732395e-06,
+      "loss": 0.4184,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.647887323943662e-06,
+      "loss": 0.2736,
+      "step": 66
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.71830985915493e-06,
+      "loss": 0.3726,
+      "step": 67
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.788732394366197e-06,
+      "loss": 0.3049,
+      "step": 68
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.859154929577465e-06,
+      "loss": 0.3652,
+      "step": 69
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.929577464788733e-06,
+      "loss": 0.4649,
+      "step": 70
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5e-06,
+      "loss": 0.4129,
+      "step": 71
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.070422535211268e-06,
+      "loss": 0.3197,
+      "step": 72
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.140845070422536e-06,
+      "loss": 0.2889,
+      "step": 73
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.211267605633803e-06,
+      "loss": 0.3439,
+      "step": 74
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.28169014084507e-06,
+      "loss": 0.3384,
+      "step": 75
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.352112676056338e-06,
+      "loss": 0.2421,
+      "step": 76
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.422535211267607e-06,
+      "loss": 0.2434,
+      "step": 77
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.492957746478874e-06,
+      "loss": 0.3097,
+      "step": 78
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.563380281690142e-06,
+      "loss": 0.253,
+      "step": 79
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.633802816901409e-06,
+      "loss": 0.3247,
+      "step": 80
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.7042253521126766e-06,
+      "loss": 0.3341,
+      "step": 81
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.774647887323944e-06,
+      "loss": 0.2774,
+      "step": 82
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.845070422535212e-06,
+      "loss": 0.3111,
+      "step": 83
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.915492957746479e-06,
+      "loss": 0.3988,
+      "step": 84
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9859154929577465e-06,
+      "loss": 0.3226,
+      "step": 85
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.056338028169015e-06,
+      "loss": 0.2715,
+      "step": 86
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.126760563380282e-06,
+      "loss": 0.3528,
+      "step": 87
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.197183098591549e-06,
+      "loss": 0.3711,
+      "step": 88
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.267605633802818e-06,
+      "loss": 0.3835,
+      "step": 89
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.3380281690140855e-06,
+      "loss": 0.2041,
+      "step": 90
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.408450704225353e-06,
+      "loss": 0.2666,
+      "step": 91
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.478873239436621e-06,
+      "loss": 0.3453,
+      "step": 92
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.549295774647888e-06,
+      "loss": 0.2053,
+      "step": 93
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.619718309859155e-06,
+      "loss": 0.4689,
+      "step": 94
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.690140845070423e-06,
+      "loss": 0.2694,
+      "step": 95
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.760563380281691e-06,
+      "loss": 0.2822,
+      "step": 96
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.830985915492958e-06,
+      "loss": 0.3153,
+      "step": 97
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.901408450704225e-06,
+      "loss": 0.2786,
+      "step": 98
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.9718309859154935e-06,
+      "loss": 0.3277,
+      "step": 99
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.042253521126761e-06,
+      "loss": 0.2813,
+      "step": 100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.112676056338029e-06,
+      "loss": 0.3475,
+      "step": 101
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.183098591549297e-06,
+      "loss": 0.2379,
+      "step": 102
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.253521126760564e-06,
+      "loss": 0.3561,
+      "step": 103
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.3239436619718316e-06,
+      "loss": 0.2051,
+      "step": 104
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.3943661971831e-06,
+      "loss": 0.3269,
+      "step": 105
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.464788732394367e-06,
+      "loss": 0.3184,
+      "step": 106
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.535211267605634e-06,
+      "loss": 0.5486,
+      "step": 107
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.6056338028169015e-06,
+      "loss": 0.2629,
+      "step": 108
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.67605633802817e-06,
+      "loss": 0.3492,
+      "step": 109
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.746478873239436e-06,
+      "loss": 0.217,
+      "step": 110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.816901408450704e-06,
+      "loss": 0.2889,
+      "step": 111
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.887323943661972e-06,
+      "loss": 0.2451,
+      "step": 112
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.95774647887324e-06,
+      "loss": 0.2728,
+      "step": 113
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.028169014084509e-06,
+      "loss": 0.2953,
+      "step": 114
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.098591549295775e-06,
+      "loss": 0.2847,
+      "step": 115
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.169014084507043e-06,
+      "loss": 0.2623,
+      "step": 116
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.239436619718311e-06,
+      "loss": 0.3969,
+      "step": 117
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.309859154929578e-06,
+      "loss": 0.2579,
+      "step": 118
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.380281690140846e-06,
+      "loss": 0.245,
+      "step": 119
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.450704225352114e-06,
+      "loss": 0.2743,
+      "step": 120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.52112676056338e-06,
+      "loss": 0.2273,
+      "step": 121
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.591549295774648e-06,
+      "loss": 0.3096,
+      "step": 122
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.661971830985915e-06,
+      "loss": 0.415,
+      "step": 123
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.732394366197183e-06,
+      "loss": 0.2351,
+      "step": 124
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.802816901408451e-06,
+      "loss": 0.237,
+      "step": 125
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.87323943661972e-06,
+      "loss": 0.3535,
+      "step": 126
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.943661971830987e-06,
+      "loss": 0.4072,
+      "step": 127
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.014084507042254e-06,
+      "loss": 0.3245,
+      "step": 128
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.084507042253522e-06,
+      "loss": 0.2921,
+      "step": 129
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.15492957746479e-06,
+      "loss": 0.1987,
+      "step": 130
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.225352112676057e-06,
+      "loss": 0.2812,
+      "step": 131
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.295774647887325e-06,
+      "loss": 0.2603,
+      "step": 132
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.366197183098593e-06,
+      "loss": 0.4538,
+      "step": 133
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.43661971830986e-06,
+      "loss": 0.2754,
+      "step": 134
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.507042253521127e-06,
+      "loss": 0.3688,
+      "step": 135
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.577464788732394e-06,
+      "loss": 0.378,
+      "step": 136
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.647887323943664e-06,
+      "loss": 0.4352,
+      "step": 137
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.71830985915493e-06,
+      "loss": 0.1091,
+      "step": 138
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.788732394366198e-06,
+      "loss": 0.3099,
+      "step": 139
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.859154929577466e-06,
+      "loss": 0.2694,
+      "step": 140
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.929577464788733e-06,
+      "loss": 0.4208,
+      "step": 141
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1e-05,
+      "loss": 0.2774,
+      "step": 142
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0070422535211269e-05,
+      "loss": 0.2744,
+      "step": 143
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0140845070422535e-05,
+      "loss": 0.3598,
+      "step": 144
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0211267605633803e-05,
+      "loss": 0.3036,
+      "step": 145
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0281690140845072e-05,
+      "loss": 0.1622,
+      "step": 146
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0352112676056338e-05,
+      "loss": 0.2657,
+      "step": 147
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0422535211267606e-05,
+      "loss": 0.2632,
+      "step": 148
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0492957746478873e-05,
+      "loss": 0.1815,
+      "step": 149
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.056338028169014e-05,
+      "loss": 0.2577,
+      "step": 150
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0633802816901409e-05,
+      "loss": 0.2063,
+      "step": 151
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0704225352112675e-05,
+      "loss": 0.2859,
+      "step": 152
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0774647887323943e-05,
+      "loss": 0.406,
+      "step": 153
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0845070422535213e-05,
+      "loss": 0.4142,
+      "step": 154
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0915492957746481e-05,
+      "loss": 0.1504,
+      "step": 155
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0985915492957748e-05,
+      "loss": 0.1746,
+      "step": 156
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1056338028169016e-05,
+      "loss": 0.1952,
+      "step": 157
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1126760563380284e-05,
+      "loss": 0.2084,
+      "step": 158
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.119718309859155e-05,
+      "loss": 0.2945,
+      "step": 159
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1267605633802819e-05,
+      "loss": 0.247,
+      "step": 160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1338028169014087e-05,
+      "loss": 0.1968,
+      "step": 161
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1408450704225353e-05,
+      "loss": 0.1935,
+      "step": 162
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1478873239436621e-05,
+      "loss": 0.3113,
+      "step": 163
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1549295774647888e-05,
+      "loss": 0.1949,
+      "step": 164
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1619718309859156e-05,
+      "loss": 0.3254,
+      "step": 165
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1690140845070424e-05,
+      "loss": 0.2308,
+      "step": 166
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.176056338028169e-05,
+      "loss": 0.2349,
+      "step": 167
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1830985915492958e-05,
+      "loss": 0.37,
+      "step": 168
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1901408450704227e-05,
+      "loss": 0.2988,
+      "step": 169
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1971830985915493e-05,
+      "loss": 0.3991,
+      "step": 170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2042253521126761e-05,
+      "loss": 0.2378,
+      "step": 171
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.211267605633803e-05,
+      "loss": 0.3736,
+      "step": 172
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2183098591549296e-05,
+      "loss": 0.3531,
+      "step": 173
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2253521126760564e-05,
+      "loss": 0.2288,
+      "step": 174
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.232394366197183e-05,
+      "loss": 0.1799,
+      "step": 175
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2394366197183098e-05,
+      "loss": 0.3801,
+      "step": 176
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2464788732394367e-05,
+      "loss": 0.3145,
+      "step": 177
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2535211267605636e-05,
+      "loss": 0.2927,
+      "step": 178
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2605633802816903e-05,
+      "loss": 0.3182,
+      "step": 179
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2676056338028171e-05,
+      "loss": 0.27,
+      "step": 180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2746478873239439e-05,
+      "loss": 0.2261,
+      "step": 181
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2816901408450705e-05,
+      "loss": 0.2645,
+      "step": 182
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2887323943661974e-05,
+      "loss": 0.2815,
+      "step": 183
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2957746478873242e-05,
+      "loss": 0.2574,
+      "step": 184
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3028169014084508e-05,
+      "loss": 0.3003,
+      "step": 185
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3098591549295776e-05,
+      "loss": 0.2545,
+      "step": 186
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3169014084507044e-05,
+      "loss": 0.2763,
+      "step": 187
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.323943661971831e-05,
+      "loss": 0.36,
+      "step": 188
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3309859154929579e-05,
+      "loss": 0.1868,
+      "step": 189
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3380281690140845e-05,
+      "loss": 0.4571,
+      "step": 190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3450704225352114e-05,
+      "loss": 0.2419,
+      "step": 191
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3521126760563382e-05,
+      "loss": 0.2845,
+      "step": 192
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3591549295774648e-05,
+      "loss": 0.3147,
+      "step": 193
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3661971830985916e-05,
+      "loss": 0.419,
+      "step": 194
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3732394366197184e-05,
+      "loss": 0.2945,
+      "step": 195
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.380281690140845e-05,
+      "loss": 0.1917,
+      "step": 196
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3873239436619719e-05,
+      "loss": 0.2909,
+      "step": 197
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3943661971830987e-05,
+      "loss": 0.1775,
+      "step": 198
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4014084507042253e-05,
+      "loss": 0.2548,
+      "step": 199
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4084507042253522e-05,
+      "loss": 0.2051,
+      "step": 200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4154929577464788e-05,
+      "loss": 0.3303,
+      "step": 201
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4225352112676058e-05,
+      "loss": 0.2845,
+      "step": 202
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4295774647887326e-05,
+      "loss": 0.2016,
+      "step": 203
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4366197183098594e-05,
+      "loss": 0.1657,
+      "step": 204
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.443661971830986e-05,
+      "loss": 0.338,
+      "step": 205
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4507042253521129e-05,
+      "loss": 0.197,
+      "step": 206
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4577464788732397e-05,
+      "loss": 0.2388,
+      "step": 207
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4647887323943663e-05,
+      "loss": 0.1975,
+      "step": 208
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4718309859154931e-05,
+      "loss": 0.2463,
+      "step": 209
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.47887323943662e-05,
+      "loss": 0.3282,
+      "step": 210
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4859154929577466e-05,
+      "loss": 0.3105,
+      "step": 211
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4929577464788734e-05,
+      "loss": 0.2413,
+      "step": 212
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.3187,
+      "step": 213
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5070422535211269e-05,
+      "loss": 0.3096,
+      "step": 214
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5140845070422537e-05,
+      "loss": 0.1883,
+      "step": 215
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5211267605633803e-05,
+      "loss": 0.1222,
+      "step": 216
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.528169014084507e-05,
+      "loss": 0.254,
+      "step": 217
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.535211267605634e-05,
+      "loss": 0.297,
+      "step": 218
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5422535211267607e-05,
+      "loss": 0.3285,
+      "step": 219
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5492957746478872e-05,
+      "loss": 0.2907,
+      "step": 220
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.556338028169014e-05,
+      "loss": 0.3265,
+      "step": 221
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.563380281690141e-05,
+      "loss": 0.2819,
+      "step": 222
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5704225352112677e-05,
+      "loss": 0.3837,
+      "step": 223
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5774647887323945e-05,
+      "loss": 0.3022,
+      "step": 224
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5845070422535213e-05,
+      "loss": 0.2491,
+      "step": 225
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.591549295774648e-05,
+      "loss": 0.2202,
+      "step": 226
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.598591549295775e-05,
+      "loss": 0.2314,
+      "step": 227
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6056338028169017e-05,
+      "loss": 0.3388,
+      "step": 228
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6126760563380285e-05,
+      "loss": 0.3595,
+      "step": 229
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.619718309859155e-05,
+      "loss": 0.2253,
+      "step": 230
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6267605633802818e-05,
+      "loss": 0.4379,
+      "step": 231
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6338028169014086e-05,
+      "loss": 0.2713,
+      "step": 232
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6408450704225354e-05,
+      "loss": 0.209,
+      "step": 233
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6478873239436623e-05,
+      "loss": 0.3046,
+      "step": 234
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6549295774647887e-05,
+      "loss": 0.2628,
+      "step": 235
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6619718309859155e-05,
+      "loss": 0.2569,
+      "step": 236
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6690140845070424e-05,
+      "loss": 0.2466,
+      "step": 237
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.676056338028169e-05,
+      "loss": 0.2371,
+      "step": 238
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.683098591549296e-05,
+      "loss": 0.1569,
+      "step": 239
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6901408450704228e-05,
+      "loss": 0.2176,
+      "step": 240
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6971830985915493e-05,
+      "loss": 0.264,
+      "step": 241
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.704225352112676e-05,
+      "loss": 0.25,
+      "step": 242
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.711267605633803e-05,
+      "loss": 0.2626,
+      "step": 243
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7183098591549297e-05,
+      "loss": 0.2664,
+      "step": 244
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7253521126760565e-05,
+      "loss": 0.3253,
+      "step": 245
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.732394366197183e-05,
+      "loss": 0.3211,
+      "step": 246
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7394366197183098e-05,
+      "loss": 0.2281,
+      "step": 247
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7464788732394366e-05,
+      "loss": 0.2939,
+      "step": 248
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7535211267605638e-05,
+      "loss": 0.2551,
+      "step": 249
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7605633802816902e-05,
+      "loss": 0.2686,
+      "step": 250
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.767605633802817e-05,
+      "loss": 0.2199,
+      "step": 251
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.774647887323944e-05,
+      "loss": 0.3774,
+      "step": 252
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7816901408450707e-05,
+      "loss": 0.3094,
+      "step": 253
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7887323943661975e-05,
+      "loss": 0.1645,
+      "step": 254
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7957746478873243e-05,
+      "loss": 0.2424,
+      "step": 255
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8028169014084508e-05,
+      "loss": 0.2802,
+      "step": 256
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8098591549295776e-05,
+      "loss": 0.3561,
+      "step": 257
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8169014084507044e-05,
+      "loss": 0.1656,
+      "step": 258
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8239436619718312e-05,
+      "loss": 0.2367,
+      "step": 259
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.830985915492958e-05,
+      "loss": 0.1392,
+      "step": 260
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8380281690140845e-05,
+      "loss": 0.3206,
+      "step": 261
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8450704225352113e-05,
+      "loss": 0.2734,
+      "step": 262
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.852112676056338e-05,
+      "loss": 0.389,
+      "step": 263
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.859154929577465e-05,
+      "loss": 0.1136,
+      "step": 264
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8661971830985917e-05,
+      "loss": 0.1316,
+      "step": 265
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8732394366197186e-05,
+      "loss": 0.3205,
+      "step": 266
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.880281690140845e-05,
+      "loss": 0.2652,
+      "step": 267
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.887323943661972e-05,
+      "loss": 0.2805,
+      "step": 268
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8943661971830987e-05,
+      "loss": 0.177,
+      "step": 269
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9014084507042255e-05,
+      "loss": 0.1931,
+      "step": 270
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9084507042253523e-05,
+      "loss": 0.3576,
+      "step": 271
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9154929577464788e-05,
+      "loss": 0.1907,
+      "step": 272
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.922535211267606e-05,
+      "loss": 0.2938,
+      "step": 273
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9295774647887327e-05,
+      "loss": 0.289,
+      "step": 274
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9366197183098595e-05,
+      "loss": 0.2156,
+      "step": 275
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.943661971830986e-05,
+      "loss": 0.4307,
+      "step": 276
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9507042253521128e-05,
+      "loss": 0.1977,
+      "step": 277
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9577464788732396e-05,
+      "loss": 0.2752,
+      "step": 278
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9647887323943664e-05,
+      "loss": 0.3232,
+      "step": 279
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9718309859154933e-05,
+      "loss": 0.3029,
+      "step": 280
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.97887323943662e-05,
+      "loss": 0.2517,
+      "step": 281
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9859154929577465e-05,
+      "loss": 0.2347,
+      "step": 282
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9929577464788734e-05,
+      "loss": 0.29,
+      "step": 283
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.1168,
+      "step": 284
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999999411990993e-05,
+      "loss": 0.2454,
+      "step": 285
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999999764796404e-05,
+      "loss": 0.2419,
+      "step": 286
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999994707919344e-05,
+      "loss": 0.1554,
+      "step": 287
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999999059185726e-05,
+      "loss": 0.262,
+      "step": 288
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999985299778262e-05,
+      "loss": 0.3607,
+      "step": 289
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999997883168298e-05,
+      "loss": 0.2037,
+      "step": 290
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999971187572172e-05,
+      "loss": 0.332,
+      "step": 291
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999962367446736e-05,
+      "loss": 0.3099,
+      "step": 292
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999952371307708e-05,
+      "loss": 0.2294,
+      "step": 293
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999941199156264e-05,
+      "loss": 0.1785,
+      "step": 294
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999928850993725e-05,
+      "loss": 0.2078,
+      "step": 295
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999915326821536e-05,
+      "loss": 0.2526,
+      "step": 296
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999900626641288e-05,
+      "loss": 0.3691,
+      "step": 297
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999988475045471e-05,
+      "loss": 0.2773,
+      "step": 298
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999986769826367e-05,
+      "loss": 0.2654,
+      "step": 299
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999849470070175e-05,
+      "loss": 0.2491,
+      "step": 300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999830065876364e-05,
+      "loss": 0.2324,
+      "step": 301
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999809485684524e-05,
+      "loss": 0.362,
+      "step": 302
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999787729497072e-05,
+      "loss": 0.1768,
+      "step": 303
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999976479731657e-05,
+      "loss": 0.206,
+      "step": 304
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999974068914571e-05,
+      "loss": 0.2639,
+      "step": 305
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999971540498733e-05,
+      "loss": 0.2274,
+      "step": 306
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999688944844404e-05,
+      "loss": 0.2431,
+      "step": 307
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999966130872004e-05,
+      "loss": 0.1362,
+      "step": 308
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99996324966175e-05,
+      "loss": 0.1769,
+      "step": 309
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999602508540153e-05,
+      "loss": 0.1815,
+      "step": 310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999957134449154e-05,
+      "loss": 0.257,
+      "step": 311
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999539004475323e-05,
+      "loss": 0.3034,
+      "step": 312
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999505488495303e-05,
+      "loss": 0.0926,
+      "step": 313
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999470796555423e-05,
+      "loss": 0.2726,
+      "step": 314
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999943492865976e-05,
+      "loss": 0.1474,
+      "step": 315
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999939788481254e-05,
+      "loss": 0.2079,
+      "step": 316
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999935966501811e-05,
+      "loss": 0.2725,
+      "step": 317
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999320269280974e-05,
+      "loss": 0.3008,
+      "step": 318
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999279697605757e-05,
+      "loss": 0.3026,
+      "step": 319
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999923794999723e-05,
+      "loss": 0.2887,
+      "step": 320
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999919502646031e-05,
+      "loss": 0.2222,
+      "step": 321
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999150927000037e-05,
+      "loss": 0.2462,
+      "step": 322
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999105651621603e-05,
+      "loss": 0.2317,
+      "step": 323
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999905920033033e-05,
+      "loss": 0.3508,
+      "step": 324
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999901157313168e-05,
+      "loss": 0.1809,
+      "step": 325
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998962770031258e-05,
+      "loss": 0.18,
+      "step": 326
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998912791034798e-05,
+      "loss": 0.1345,
+      "step": 327
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999886163614818e-05,
+      "loss": 0.2936,
+      "step": 328
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999880930537742e-05,
+      "loss": 0.341,
+      "step": 329
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998755798728673e-05,
+      "loss": 0.4943,
+      "step": 330
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999870111620823e-05,
+      "loss": 0.2253,
+      "step": 331
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998645257822523e-05,
+      "loss": 0.2927,
+      "step": 332
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999858822357812e-05,
+      "loss": 0.2199,
+      "step": 333
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999853001348173e-05,
+      "loss": 0.3093,
+      "step": 334
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998470627540195e-05,
+      "loss": 0.2779,
+      "step": 335
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998410065760503e-05,
+      "loss": 0.3728,
+      "step": 336
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998348328149775e-05,
+      "loss": 0.1684,
+      "step": 337
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999828541471527e-05,
+      "loss": 0.1665,
+      "step": 338
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998221325464385e-05,
+      "loss": 0.3486,
+      "step": 339
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998156060404665e-05,
+      "loss": 0.3234,
+      "step": 340
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998089619543775e-05,
+      "loss": 0.259,
+      "step": 341
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998022002889535e-05,
+      "loss": 0.1971,
+      "step": 342
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997953210449895e-05,
+      "loss": 0.234,
+      "step": 343
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997883242232948e-05,
+      "loss": 0.2632,
+      "step": 344
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997812098246918e-05,
+      "loss": 0.2022,
+      "step": 345
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997739778500172e-05,
+      "loss": 0.2579,
+      "step": 346
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999766628300122e-05,
+      "loss": 0.2116,
+      "step": 347
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.99975916117587e-05,
+      "loss": 0.2291,
+      "step": 348
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997515764781393e-05,
+      "loss": 0.3236,
+      "step": 349
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997438742078223e-05,
+      "loss": 0.1856,
+      "step": 350
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997360543658244e-05,
+      "loss": 0.3259,
+      "step": 351
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997281169530657e-05,
+      "loss": 0.3582,
+      "step": 352
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999720061970479e-05,
+      "loss": 0.3517,
+      "step": 353
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997118894190126e-05,
+      "loss": 0.2094,
+      "step": 354
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997035992996264e-05,
+      "loss": 0.2639,
+      "step": 355
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996951916132958e-05,
+      "loss": 0.2242,
+      "step": 356
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99968666636101e-05,
+      "loss": 0.2927,
+      "step": 357
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999678023543771e-05,
+      "loss": 0.2414,
+      "step": 358
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996692631625956e-05,
+      "loss": 0.1805,
+      "step": 359
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996603852185136e-05,
+      "loss": 0.2372,
+      "step": 360
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996513897125696e-05,
+      "loss": 0.2431,
+      "step": 361
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996422766458214e-05,
+      "loss": 0.2567,
+      "step": 362
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996330460193402e-05,
+      "loss": 0.2399,
+      "step": 363
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999623697834212e-05,
+      "loss": 0.3421,
+      "step": 364
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999614232091536e-05,
+      "loss": 0.302,
+      "step": 365
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996046487924255e-05,
+      "loss": 0.1755,
+      "step": 366
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9995949479380075e-05,
+      "loss": 0.2457,
+      "step": 367
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9995851295294227e-05,
+      "loss": 0.2446,
+      "step": 368
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995751935678257e-05,
+      "loss": 0.2594,
+      "step": 369
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999565140054385e-05,
+      "loss": 0.312,
+      "step": 370
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995549689902836e-05,
+      "loss": 0.2861,
+      "step": 371
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999544680376717e-05,
+      "loss": 0.3332,
+      "step": 372
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999534274214895e-05,
+      "loss": 0.2258,
+      "step": 373
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995237505060413e-05,
+      "loss": 0.2773,
+      "step": 374
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995131092513944e-05,
+      "loss": 0.3378,
+      "step": 375
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999502350452205e-05,
+      "loss": 0.1891,
+      "step": 376
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994914741097385e-05,
+      "loss": 0.22,
+      "step": 377
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999480480225274e-05,
+      "loss": 0.1847,
+      "step": 378
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994693688001042e-05,
+      "loss": 0.2696,
+      "step": 379
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999458139835536e-05,
+      "loss": 0.2121,
+      "step": 380
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994467933328904e-05,
+      "loss": 0.1706,
+      "step": 381
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994353292935012e-05,
+      "loss": 0.4377,
+      "step": 382
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994237477187165e-05,
+      "loss": 0.2472,
+      "step": 383
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994120486098985e-05,
+      "loss": 0.2426,
+      "step": 384
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999400231968423e-05,
+      "loss": 0.3053,
+      "step": 385
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.99938829779568e-05,
+      "loss": 0.2884,
+      "step": 386
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9993762460930726e-05,
+      "loss": 0.2118,
+      "step": 387
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993640768620184e-05,
+      "loss": 0.2006,
+      "step": 388
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999351790103948e-05,
+      "loss": 0.2924,
+      "step": 389
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999339385820307e-05,
+      "loss": 0.1175,
+      "step": 390
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993268640125536e-05,
+      "loss": 0.3781,
+      "step": 391
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993142246821608e-05,
+      "loss": 0.3718,
+      "step": 392
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993014678306147e-05,
+      "loss": 0.2057,
+      "step": 393
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999288593459416e-05,
+      "loss": 0.2265,
+      "step": 394
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999275601570078e-05,
+      "loss": 0.241,
+      "step": 395
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999262492164129e-05,
+      "loss": 0.2133,
+      "step": 396
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999249265243111e-05,
+      "loss": 0.1932,
+      "step": 397
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992359208085793e-05,
+      "loss": 0.3148,
+      "step": 398
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992224588621027e-05,
+      "loss": 0.1627,
+      "step": 399
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999208879405265e-05,
+      "loss": 0.2231,
+      "step": 400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999195182439663e-05,
+      "loss": 0.2576,
+      "step": 401
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991813679669072e-05,
+      "loss": 0.2693,
+      "step": 402
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991674359886226e-05,
+      "loss": 0.1994,
+      "step": 403
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991533865064473e-05,
+      "loss": 0.1777,
+      "step": 404
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999139219522034e-05,
+      "loss": 0.2621,
+      "step": 405
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999124935037048e-05,
+      "loss": 0.3486,
+      "step": 406
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.99911053305317e-05,
+      "loss": 0.3706,
+      "step": 407
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990960135720932e-05,
+      "loss": 0.1464,
+      "step": 408
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990813765955256e-05,
+      "loss": 0.2552,
+      "step": 409
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990666221251875e-05,
+      "loss": 0.1483,
+      "step": 410
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990517501628152e-05,
+      "loss": 0.3424,
+      "step": 411
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990367607101572e-05,
+      "loss": 0.3158,
+      "step": 412
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990216537689763e-05,
+      "loss": 0.2016,
+      "step": 413
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999006429341049e-05,
+      "loss": 0.1819,
+      "step": 414
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989910874281662e-05,
+      "loss": 0.1576,
+      "step": 415
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989756280321315e-05,
+      "loss": 0.3027,
+      "step": 416
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998960051154763e-05,
+      "loss": 0.1799,
+      "step": 417
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989443567978932e-05,
+      "loss": 0.2549,
+      "step": 418
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998928544963367e-05,
+      "loss": 0.2166,
+      "step": 419
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989126156530444e-05,
+      "loss": 0.24,
+      "step": 420
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988965688687987e-05,
+      "loss": 0.3242,
+      "step": 421
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998880404612517e-05,
+      "loss": 0.2642,
+      "step": 422
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988641228861e-05,
+      "loss": 0.2312,
+      "step": 423
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988477236914625e-05,
+      "loss": 0.2722,
+      "step": 424
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988312070305336e-05,
+      "loss": 0.3362,
+      "step": 425
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998814572905255e-05,
+      "loss": 0.1802,
+      "step": 426
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998797821317583e-05,
+      "loss": 0.1532,
+      "step": 427
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987809522694884e-05,
+      "loss": 0.3054,
+      "step": 428
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998763965762954e-05,
+      "loss": 0.2366,
+      "step": 429
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998746861799978e-05,
+      "loss": 0.2022,
+      "step": 430
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987296403825715e-05,
+      "loss": 0.2095,
+      "step": 431
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987123015127602e-05,
+      "loss": 0.2762,
+      "step": 432
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998694845192583e-05,
+      "loss": 0.2508,
+      "step": 433
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998677271424093e-05,
+      "loss": 0.2044,
+      "step": 434
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986595802093566e-05,
+      "loss": 0.257,
+      "step": 435
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986417715504542e-05,
+      "loss": 0.2175,
+      "step": 436
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986238454494805e-05,
+      "loss": 0.2883,
+      "step": 437
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998605801908543e-05,
+      "loss": 0.2902,
+      "step": 438
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998587640929765e-05,
+      "loss": 0.2407,
+      "step": 439
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985693625152807e-05,
+      "loss": 0.2437,
+      "step": 440
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998550966667241e-05,
+      "loss": 0.2797,
+      "step": 441
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985324533878082e-05,
+      "loss": 0.2519,
+      "step": 442
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.99851382267916e-05,
+      "loss": 0.3,
+      "step": 443
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984950745434875e-05,
+      "loss": 0.3161,
+      "step": 444
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984762089829953e-05,
+      "loss": 0.2461,
+      "step": 445
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998457225999902e-05,
+      "loss": 0.2453,
+      "step": 446
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.99843812559644e-05,
+      "loss": 0.3031,
+      "step": 447
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984189077748562e-05,
+      "loss": 0.2504,
+      "step": 448
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983995725374096e-05,
+      "loss": 0.3365,
+      "step": 449
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983801198863752e-05,
+      "loss": 0.1846,
+      "step": 450
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983605498240395e-05,
+      "loss": 0.1735,
+      "step": 451
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998340862352705e-05,
+      "loss": 0.3807,
+      "step": 452
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983210574746857e-05,
+      "loss": 0.3203,
+      "step": 453
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983011351923123e-05,
+      "loss": 0.3364,
+      "step": 454
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982810955079266e-05,
+      "loss": 0.4128,
+      "step": 455
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982609384238857e-05,
+      "loss": 0.1329,
+      "step": 456
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.99824066394256e-05,
+      "loss": 0.2151,
+      "step": 457
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998220272066333e-05,
+      "loss": 0.3818,
+      "step": 458
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981997627976046e-05,
+      "loss": 0.2789,
+      "step": 459
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981791361387858e-05,
+      "loss": 0.2509,
+      "step": 460
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981583920923018e-05,
+      "loss": 0.2289,
+      "step": 461
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981375306605927e-05,
+      "loss": 0.3319,
+      "step": 462
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.998116551846112e-05,
+      "loss": 0.2209,
+      "step": 463
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.998095455651327e-05,
+      "loss": 0.201,
+      "step": 464
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980742420787176e-05,
+      "loss": 0.1265,
+      "step": 465
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980529111307792e-05,
+      "loss": 0.2153,
+      "step": 466
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980314628100207e-05,
+      "loss": 0.3453,
+      "step": 467
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980098971189644e-05,
+      "loss": 0.1681,
+      "step": 468
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979882140601458e-05,
+      "loss": 0.4614,
+      "step": 469
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979664136361153e-05,
+      "loss": 0.1611,
+      "step": 470
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997944495849437e-05,
+      "loss": 0.2278,
+      "step": 471
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979224607026876e-05,
+      "loss": 0.2224,
+      "step": 472
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979003081984595e-05,
+      "loss": 0.2981,
+      "step": 473
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997878038339357e-05,
+      "loss": 0.2581,
+      "step": 474
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978556511279995e-05,
+      "loss": 0.1634,
+      "step": 475
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.99783314656702e-05,
+      "loss": 0.1611,
+      "step": 476
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978105246590645e-05,
+      "loss": 0.2558,
+      "step": 477
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997787785406794e-05,
+      "loss": 0.4443,
+      "step": 478
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997764928812882e-05,
+      "loss": 0.3482,
+      "step": 479
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997741954880017e-05,
+      "loss": 0.2938,
+      "step": 480
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977188636109007e-05,
+      "loss": 0.2157,
+      "step": 481
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976956550082486e-05,
+      "loss": 0.275,
+      "step": 482
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976723290747895e-05,
+      "loss": 0.1706,
+      "step": 483
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976488858132675e-05,
+      "loss": 0.1644,
+      "step": 484
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976253252264394e-05,
+      "loss": 0.3645,
+      "step": 485
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976016473170755e-05,
+      "loss": 0.2113,
+      "step": 486
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997577852087961e-05,
+      "loss": 0.261,
+      "step": 487
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975539395418936e-05,
+      "loss": 0.1485,
+      "step": 488
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975299096816857e-05,
+      "loss": 0.2286,
+      "step": 489
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997505762510163e-05,
+      "loss": 0.1943,
+      "step": 490
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997481498030166e-05,
+      "loss": 0.3522,
+      "step": 491
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974571162445477e-05,
+      "loss": 0.2328,
+      "step": 492
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974326171561757e-05,
+      "loss": 0.2486,
+      "step": 493
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974080007679307e-05,
+      "loss": 0.261,
+      "step": 494
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997383267082708e-05,
+      "loss": 0.1919,
+      "step": 495
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973584161034164e-05,
+      "loss": 0.3078,
+      "step": 496
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973334478329782e-05,
+      "loss": 0.183,
+      "step": 497
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973083622743295e-05,
+      "loss": 0.2306,
+      "step": 498
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997283159430421e-05,
+      "loss": 0.2703,
+      "step": 499
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9972578393042162e-05,
+      "loss": 0.2295,
+      "step": 500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997232401898693e-05,
+      "loss": 0.1347,
+      "step": 501
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9972068472168424e-05,
+      "loss": 0.4344,
+      "step": 502
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.99718117526167e-05,
+      "loss": 0.1095,
+      "step": 503
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9971553860361954e-05,
+      "loss": 0.2552,
+      "step": 504
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9971294795434505e-05,
+      "loss": 0.2313,
+      "step": 505
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9971034557864825e-05,
+      "loss": 0.2918,
+      "step": 506
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997077314768352e-05,
+      "loss": 0.2088,
+      "step": 507
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9970510564921328e-05,
+      "loss": 0.2522,
+      "step": 508
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9970246809609127e-05,
+      "loss": 0.3004,
+      "step": 509
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9969981881777946e-05,
+      "loss": 0.1378,
+      "step": 510
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996971578145893e-05,
+      "loss": 0.2116,
+      "step": 511
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9969448508683377e-05,
+      "loss": 0.1752,
+      "step": 512
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996918006348272e-05,
+      "loss": 0.1693,
+      "step": 513
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996891044588853e-05,
+      "loss": 0.2801,
+      "step": 514
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9968639655932506e-05,
+      "loss": 0.2676,
+      "step": 515
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9968367693646505e-05,
+      "loss": 0.1486,
+      "step": 516
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.99680945590625e-05,
+      "loss": 0.2472,
+      "step": 517
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996782025221262e-05,
+      "loss": 0.2849,
+      "step": 518
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9967544773129118e-05,
+      "loss": 0.3063,
+      "step": 519
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9967268121844395e-05,
+      "loss": 0.1517,
+      "step": 520
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966990298390984e-05,
+      "loss": 0.3536,
+      "step": 521
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996671130280156e-05,
+      "loss": 0.2268,
+      "step": 522
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966431135108925e-05,
+      "loss": 0.3127,
+      "step": 523
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966149795346037e-05,
+      "loss": 0.337,
+      "step": 524
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9965867283545978e-05,
+      "loss": 0.246,
+      "step": 525
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9965583599741974e-05,
+      "loss": 0.2052,
+      "step": 526
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9965298743967382e-05,
+      "loss": 0.1672,
+      "step": 527
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9965012716255704e-05,
+      "loss": 0.2795,
+      "step": 528
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996472551664058e-05,
+      "loss": 0.113,
+      "step": 529
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9964437145155785e-05,
+      "loss": 0.2437,
+      "step": 530
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9964147601835228e-05,
+      "loss": 0.1965,
+      "step": 531
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996385688671296e-05,
+      "loss": 0.3071,
+      "step": 532
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9963564999823174e-05,
+      "loss": 0.328,
+      "step": 533
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996327194120019e-05,
+      "loss": 0.2102,
+      "step": 534
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9962977710878485e-05,
+      "loss": 0.2243,
+      "step": 535
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9962682308892643e-05,
+      "loss": 0.3501,
+      "step": 536
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996238573527742e-05,
+      "loss": 0.194,
+      "step": 537
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9962087990067682e-05,
+      "loss": 0.3538,
+      "step": 538
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9961789073298453e-05,
+      "loss": 0.2567,
+      "step": 539
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.996148898500488e-05,
+      "loss": 0.3082,
+      "step": 540
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9961187725222256e-05,
+      "loss": 0.1733,
+      "step": 541
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9960885293986006e-05,
+      "loss": 0.2668,
+      "step": 542
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9960581691331705e-05,
+      "loss": 0.135,
+      "step": 543
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9960276917295052e-05,
+      "loss": 0.4188,
+      "step": 544
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9959970971911885e-05,
+      "loss": 0.1251,
+      "step": 545
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9959663855218193e-05,
+      "loss": 0.2158,
+      "step": 546
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995935556725009e-05,
+      "loss": 0.2193,
+      "step": 547
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9959046108043822e-05,
+      "loss": 0.2622,
+      "step": 548
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9958735477635795e-05,
+      "loss": 0.1161,
+      "step": 549
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9958423676062537e-05,
+      "loss": 0.2069,
+      "step": 550
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995811070336071e-05,
+      "loss": 0.2278,
+      "step": 551
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9957796559567125e-05,
+      "loss": 0.2833,
+      "step": 552
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9957481244718723e-05,
+      "loss": 0.1651,
+      "step": 553
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9957164758852593e-05,
+      "loss": 0.2054,
+      "step": 554
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9956847102005942e-05,
+      "loss": 0.2573,
+      "step": 555
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9956528274216137e-05,
+      "loss": 0.179,
+      "step": 556
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995620827552067e-05,
+      "loss": 0.1858,
+      "step": 557
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9955887105957175e-05,
+      "loss": 0.293,
+      "step": 558
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995556476556342e-05,
+      "loss": 0.4158,
+      "step": 559
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995524125437731e-05,
+      "loss": 0.2615,
+      "step": 560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995491657243689e-05,
+      "loss": 0.17,
+      "step": 561
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9954590719780355e-05,
+      "loss": 0.2396,
+      "step": 562
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9954263696446012e-05,
+      "loss": 0.1827,
+      "step": 563
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995393550247233e-05,
+      "loss": 0.233,
+      "step": 564
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9953606137897893e-05,
+      "loss": 0.2635,
+      "step": 565
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9953275602761445e-05,
+      "loss": 0.2609,
+      "step": 566
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9952943897101855e-05,
+      "loss": 0.2655,
+      "step": 567
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9952611020958136e-05,
+      "loss": 0.1995,
+      "step": 568
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9952276974369426e-05,
+      "loss": 0.2869,
+      "step": 569
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9951941757375015e-05,
+      "loss": 0.1026,
+      "step": 570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9951605370014323e-05,
+      "loss": 0.2273,
+      "step": 571
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995126781232691e-05,
+      "loss": 0.3283,
+      "step": 572
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9950929084352476e-05,
+      "loss": 0.1861,
+      "step": 573
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9950589186130857e-05,
+      "loss": 0.2848,
+      "step": 574
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9950248117702017e-05,
+      "loss": 0.2382,
+      "step": 575
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.994990587910608e-05,
+      "loss": 0.2179,
+      "step": 576
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994956247038328e-05,
+      "loss": 0.1382,
+      "step": 577
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994921789157401e-05,
+      "loss": 0.3586,
+      "step": 578
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9948872142718795e-05,
+      "loss": 0.2055,
+      "step": 579
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9948525223858288e-05,
+      "loss": 0.1407,
+      "step": 580
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9948177135033295e-05,
+      "loss": 0.2875,
+      "step": 581
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9947827876284745e-05,
+      "loss": 0.1412,
+      "step": 582
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994747744765372e-05,
+      "loss": 0.1601,
+      "step": 583
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9947125849181422e-05,
+      "loss": 0.1432,
+      "step": 584
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9946773080909207e-05,
+      "loss": 0.2344,
+      "step": 585
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9946419142878556e-05,
+      "loss": 0.2231,
+      "step": 586
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9946064035131095e-05,
+      "loss": 0.2926,
+      "step": 587
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9945707757708583e-05,
+      "loss": 0.1906,
+      "step": 588
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9945350310652922e-05,
+      "loss": 0.2445,
+      "step": 589
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9944991694006147e-05,
+      "loss": 0.2385,
+      "step": 590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9944631907810435e-05,
+      "loss": 0.1737,
+      "step": 591
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9944270952108092e-05,
+      "loss": 0.1244,
+      "step": 592
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9943908826941572e-05,
+      "loss": 0.2167,
+      "step": 593
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9943545532353458e-05,
+      "loss": 0.3113,
+      "step": 594
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9943181068386473e-05,
+      "loss": 0.153,
+      "step": 595
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9942815435083484e-05,
+      "loss": 0.2814,
+      "step": 596
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9942448632487486e-05,
+      "loss": 0.2517,
+      "step": 597
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9942080660641615e-05,
+      "loss": 0.2224,
+      "step": 598
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9941711519589152e-05,
+      "loss": 0.1844,
+      "step": 599
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.99413412093735e-05,
+      "loss": 0.3252,
+      "step": 600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.994096973003821e-05,
+      "loss": 0.2876,
+      "step": 601
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9940597081626975e-05,
+      "loss": 0.1736,
+      "step": 602
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9940223264183612e-05,
+      "loss": 0.3604,
+      "step": 603
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9939848277752085e-05,
+      "loss": 0.1095,
+      "step": 604
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9939472122376492e-05,
+      "loss": 0.3865,
+      "step": 605
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9939094798101074e-05,
+      "loss": 0.1472,
+      "step": 606
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.99387163049702e-05,
+      "loss": 0.3937,
+      "step": 607
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9938336643028383e-05,
+      "loss": 0.2002,
+      "step": 608
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9937955812320275e-05,
+      "loss": 0.2333,
+      "step": 609
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9937573812890655e-05,
+      "loss": 0.189,
+      "step": 610
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9937190644784453e-05,
+      "loss": 0.1206,
+      "step": 611
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993680630804673e-05,
+      "loss": 0.3613,
+      "step": 612
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993642080272268e-05,
+      "loss": 0.2806,
+      "step": 613
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9936034128857645e-05,
+      "loss": 0.229,
+      "step": 614
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.99356462864971e-05,
+      "loss": 0.2066,
+      "step": 615
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9935257275686647e-05,
+      "loss": 0.2905,
+      "step": 616
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993486709647204e-05,
+      "loss": 0.1671,
+      "step": 617
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9934475748899164e-05,
+      "loss": 0.1844,
+      "step": 618
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993408323301404e-05,
+      "loss": 0.2943,
+      "step": 619
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9933689548862837e-05,
+      "loss": 0.1762,
+      "step": 620
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9933294696491843e-05,
+      "loss": 0.2095,
+      "step": 621
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9932898675947496e-05,
+      "loss": 0.3674,
+      "step": 622
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9932501487276373e-05,
+      "loss": 0.2781,
+      "step": 623
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993210313052518e-05,
+      "loss": 0.1451,
+      "step": 624
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993170360574076e-05,
+      "loss": 0.2386,
+      "step": 625
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993130291297011e-05,
+      "loss": 0.1373,
+      "step": 626
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9930901052260342e-05,
+      "loss": 0.2261,
+      "step": 627
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993049802365872e-05,
+      "loss": 0.1439,
+      "step": 628
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9930093827212646e-05,
+      "loss": 0.2097,
+      "step": 629
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.992968846296964e-05,
+      "loss": 0.2284,
+      "step": 630
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9929281930977386e-05,
+      "loss": 0.235,
+      "step": 631
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9928874231283685e-05,
+      "loss": 0.2238,
+      "step": 632
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992846536393649e-05,
+      "loss": 0.1817,
+      "step": 633
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992805532898388e-05,
+      "loss": 0.3337,
+      "step": 634
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9927644126474076e-05,
+      "loss": 0.1553,
+      "step": 635
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992723175645544e-05,
+      "loss": 0.1621,
+      "step": 636
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9926818218976465e-05,
+      "loss": 0.1426,
+      "step": 637
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992640351408578e-05,
+      "loss": 0.239,
+      "step": 638
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992598764183216e-05,
+      "loss": 0.2589,
+      "step": 639
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9925570602264518e-05,
+      "loss": 0.1584,
+      "step": 640
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9925152395431882e-05,
+      "loss": 0.1688,
+      "step": 641
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9924733021383448e-05,
+      "loss": 0.2596,
+      "step": 642
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992431248016853e-05,
+      "loss": 0.1687,
+      "step": 643
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9923890771836584e-05,
+      "loss": 0.2695,
+      "step": 644
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9923467896437204e-05,
+      "loss": 0.1995,
+      "step": 645
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9923043854020125e-05,
+      "loss": 0.2822,
+      "step": 646
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992261864463521e-05,
+      "loss": 0.2606,
+      "step": 647
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9922192268332464e-05,
+      "loss": 0.2154,
+      "step": 648
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9921764725162034e-05,
+      "loss": 0.2106,
+      "step": 649
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9921336015174196e-05,
+      "loss": 0.357,
+      "step": 650
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992090613841937e-05,
+      "loss": 0.1654,
+      "step": 651
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.992047509494811e-05,
+      "loss": 0.2548,
+      "step": 652
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9920042884811105e-05,
+      "loss": 0.1341,
+      "step": 653
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9919609508059184e-05,
+      "loss": 0.234,
+      "step": 654
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9919174964743316e-05,
+      "loss": 0.2678,
+      "step": 655
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9918739254914602e-05,
+      "loss": 0.2128,
+      "step": 656
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991830237862428e-05,
+      "loss": 0.1537,
+      "step": 657
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9917864335923732e-05,
+      "loss": 0.2586,
+      "step": 658
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9917425126864467e-05,
+      "loss": 0.1207,
+      "step": 659
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9916984751498143e-05,
+      "loss": 0.2626,
+      "step": 660
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9916543209876545e-05,
+      "loss": 0.2261,
+      "step": 661
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9916100502051604e-05,
+      "loss": 0.2174,
+      "step": 662
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9915656628075374e-05,
+      "loss": 0.1861,
+      "step": 663
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9915211588000062e-05,
+      "loss": 0.3252,
+      "step": 664
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9914765381878006e-05,
+      "loss": 0.3271,
+      "step": 665
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991431800976168e-05,
+      "loss": 0.2647,
+      "step": 666
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9913869471703695e-05,
+      "loss": 0.2541,
+      "step": 667
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9913419767756796e-05,
+      "loss": 0.2386,
+      "step": 668
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9912968897973874e-05,
+      "loss": 0.2756,
+      "step": 669
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991251686240795e-05,
+      "loss": 0.27,
+      "step": 670
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9912063661112188e-05,
+      "loss": 0.2722,
+      "step": 671
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.991160929413988e-05,
+      "loss": 0.1791,
+      "step": 672
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.991115376154446e-05,
+      "loss": 0.1848,
+      "step": 673
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.991069706337951e-05,
+      "loss": 0.2836,
+      "step": 674
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.991023919969872e-05,
+      "loss": 0.2531,
+      "step": 675
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9909780170555955e-05,
+      "loss": 0.2832,
+      "step": 676
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9909319976005184e-05,
+      "loss": 0.1364,
+      "step": 677
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9908858616100534e-05,
+      "loss": 0.1761,
+      "step": 678
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9908396090896257e-05,
+      "loss": 0.257,
+      "step": 679
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990793240044675e-05,
+      "loss": 0.2609,
+      "step": 680
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9907467544806542e-05,
+      "loss": 0.2986,
+      "step": 681
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9907001524030303e-05,
+      "loss": 0.268,
+      "step": 682
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9906534338172835e-05,
+      "loss": 0.3071,
+      "step": 683
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9906065987289084e-05,
+      "loss": 0.2191,
+      "step": 684
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9905596471434124e-05,
+      "loss": 0.1482,
+      "step": 685
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9905125790663174e-05,
+      "loss": 0.3271,
+      "step": 686
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9904653945031587e-05,
+      "loss": 0.2501,
+      "step": 687
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9904180934594853e-05,
+      "loss": 0.2252,
+      "step": 688
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.99037067594086e-05,
+      "loss": 0.2543,
+      "step": 689
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9903231419528582e-05,
+      "loss": 0.1956,
+      "step": 690
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9902754915010715e-05,
+      "loss": 0.2151,
+      "step": 691
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9902277245911027e-05,
+      "loss": 0.1808,
+      "step": 692
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9901798412285695e-05,
+      "loss": 0.176,
+      "step": 693
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9901318414191032e-05,
+      "loss": 0.2241,
+      "step": 694
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9900837251683486e-05,
+      "loss": 0.2601,
+      "step": 695
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.990035492481964e-05,
+      "loss": 0.3599,
+      "step": 696
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989987143365622e-05,
+      "loss": 0.2882,
+      "step": 697
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9899386778250084e-05,
+      "loss": 0.1632,
+      "step": 698
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9898900958658226e-05,
+      "loss": 0.1783,
+      "step": 699
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9898413974937783e-05,
+      "loss": 0.3442,
+      "step": 700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9897925827146026e-05,
+      "loss": 0.2605,
+      "step": 701
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989743651534036e-05,
+      "loss": 0.2263,
+      "step": 702
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9896946039578324e-05,
+      "loss": 0.2173,
+      "step": 703
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989645439991761e-05,
+      "loss": 0.2647,
+      "step": 704
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9895961596416024e-05,
+      "loss": 0.1204,
+      "step": 705
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9895467629131532e-05,
+      "loss": 0.2769,
+      "step": 706
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9894972498122213e-05,
+      "loss": 0.2311,
+      "step": 707
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9894476203446308e-05,
+      "loss": 0.1996,
+      "step": 708
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989397874516217e-05,
+      "loss": 0.191,
+      "step": 709
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989348012332831e-05,
+      "loss": 0.3325,
+      "step": 710
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9892980338003364e-05,
+      "loss": 0.2061,
+      "step": 711
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9892479389246106e-05,
+      "loss": 0.1651,
+      "step": 712
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989197727711545e-05,
+      "loss": 0.1653,
+      "step": 713
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9891474001670447e-05,
+      "loss": 0.2746,
+      "step": 714
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989096956297028e-05,
+      "loss": 0.2196,
+      "step": 715
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9890463961074273e-05,
+      "loss": 0.3351,
+      "step": 716
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9889957196041886e-05,
+      "loss": 0.1057,
+      "step": 717
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9889449267932713e-05,
+      "loss": 0.1467,
+      "step": 718
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9888940176806494e-05,
+      "loss": 0.1975,
+      "step": 719
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988842992272309e-05,
+      "loss": 0.1672,
+      "step": 720
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9887918505742514e-05,
+      "loss": 0.2443,
+      "step": 721
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988740592592491e-05,
+      "loss": 0.2059,
+      "step": 722
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9886892183330557e-05,
+      "loss": 0.3004,
+      "step": 723
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9886377278019866e-05,
+      "loss": 0.2759,
+      "step": 724
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.98858612100534e-05,
+      "loss": 0.3338,
+      "step": 725
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9885343979491846e-05,
+      "loss": 0.3445,
+      "step": 726
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9884825586396032e-05,
+      "loss": 0.1364,
+      "step": 727
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9884306030826915e-05,
+      "loss": 0.1738,
+      "step": 728
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9883785312845605e-05,
+      "loss": 0.1064,
+      "step": 729
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9883263432513334e-05,
+      "loss": 0.2541,
+      "step": 730
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9882740389891483e-05,
+      "loss": 0.141,
+      "step": 731
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9882216185041555e-05,
+      "loss": 0.2095,
+      "step": 732
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.98816908180252e-05,
+      "loss": 0.2766,
+      "step": 733
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.98811642889042e-05,
+      "loss": 0.3181,
+      "step": 734
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.988063659774048e-05,
+      "loss": 0.3473,
+      "step": 735
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9880107744596095e-05,
+      "loss": 0.2065,
+      "step": 736
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987957772953324e-05,
+      "loss": 0.2802,
+      "step": 737
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987904655261425e-05,
+      "loss": 0.174,
+      "step": 738
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987851421390158e-05,
+      "loss": 0.2709,
+      "step": 739
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9877980713457844e-05,
+      "loss": 0.2233,
+      "step": 740
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987744605134578e-05,
+      "loss": 0.2174,
+      "step": 741
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9876910227628267e-05,
+      "loss": 0.3169,
+      "step": 742
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9876373242368318e-05,
+      "loss": 0.1722,
+      "step": 743
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987583509562908e-05,
+      "loss": 0.2045,
+      "step": 744
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9875295787473845e-05,
+      "loss": 0.188,
+      "step": 745
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9874755317966033e-05,
+      "loss": 0.214,
+      "step": 746
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9874213687169206e-05,
+      "loss": 0.263,
+      "step": 747
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987367089514706e-05,
+      "loss": 0.1311,
+      "step": 748
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987312694196343e-05,
+      "loss": 0.1629,
+      "step": 749
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987258182768228e-05,
+      "loss": 0.2826,
+      "step": 750
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9872035552367727e-05,
+      "loss": 0.2814,
+      "step": 751
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9871488116084007e-05,
+      "loss": 0.2751,
+      "step": 752
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9870939518895497e-05,
+      "loss": 0.312,
+      "step": 753
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9870389760866717e-05,
+      "loss": 0.2431,
+      "step": 754
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9869838842062322e-05,
+      "loss": 0.3343,
+      "step": 755
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9869286762547094e-05,
+      "loss": 0.118,
+      "step": 756
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9868733522385966e-05,
+      "loss": 0.3153,
+      "step": 757
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9868179121643997e-05,
+      "loss": 0.3495,
+      "step": 758
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986762356038638e-05,
+      "loss": 0.2987,
+      "step": 759
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986706683867846e-05,
+      "loss": 0.2979,
+      "step": 760
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.98665089565857e-05,
+      "loss": 0.3024,
+      "step": 761
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986594991417371e-05,
+      "loss": 0.1311,
+      "step": 762
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9865389711508243e-05,
+      "loss": 0.1167,
+      "step": 763
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986482834865517e-05,
+      "loss": 0.2428,
+      "step": 764
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9864265825680507e-05,
+      "loss": 0.1666,
+      "step": 765
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9863702142650418e-05,
+      "loss": 0.2342,
+      "step": 766
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986313729963118e-05,
+      "loss": 0.2028,
+      "step": 767
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986257129668923e-05,
+      "loss": 0.1619,
+      "step": 768
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986200413389113e-05,
+      "loss": 0.2938,
+      "step": 769
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986143581130357e-05,
+      "loss": 0.1361,
+      "step": 770
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.98608663289934e-05,
+      "loss": 0.2128,
+      "step": 771
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9860295687027583e-05,
+      "loss": 0.1582,
+      "step": 772
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9859723885473225e-05,
+      "loss": 0.2226,
+      "step": 773
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9859150924397577e-05,
+      "loss": 0.2538,
+      "step": 774
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985857680386802e-05,
+      "loss": 0.1975,
+      "step": 775
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985800152395207e-05,
+      "loss": 0.2229,
+      "step": 776
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9857425084717378e-05,
+      "loss": 0.1774,
+      "step": 777
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985684748623174e-05,
+      "loss": 0.3467,
+      "step": 778
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9856268728563078e-05,
+      "loss": 0.3318,
+      "step": 779
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9855688811779457e-05,
+      "loss": 0.2468,
+      "step": 780
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9855107735949077e-05,
+      "loss": 0.3241,
+      "step": 781
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9854525501140273e-05,
+      "loss": 0.2384,
+      "step": 782
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9853942107421516e-05,
+      "loss": 0.3909,
+      "step": 783
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9853357554861415e-05,
+      "loss": 0.1282,
+      "step": 784
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9852771843528712e-05,
+      "loss": 0.403,
+      "step": 785
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9852184973492292e-05,
+      "loss": 0.3027,
+      "step": 786
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9851596944821167e-05,
+      "loss": 0.2732,
+      "step": 787
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.98510077575845e-05,
+      "loss": 0.2696,
+      "step": 788
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9850417411851565e-05,
+      "loss": 0.1355,
+      "step": 789
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9849825907691798e-05,
+      "loss": 0.1902,
+      "step": 790
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984923324517476e-05,
+      "loss": 0.174,
+      "step": 791
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9848639424370152e-05,
+      "loss": 0.1769,
+      "step": 792
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.98480444453478e-05,
+      "loss": 0.2301,
+      "step": 793
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9847448308177678e-05,
+      "loss": 0.2371,
+      "step": 794
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.98468510129299e-05,
+      "loss": 0.1646,
+      "step": 795
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9846252559674698e-05,
+      "loss": 0.2075,
+      "step": 796
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984565294848246e-05,
+      "loss": 0.1589,
+      "step": 797
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9845052179423695e-05,
+      "loss": 0.1164,
+      "step": 798
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9844450252569056e-05,
+      "loss": 0.1871,
+      "step": 799
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9843847167989337e-05,
+      "loss": 0.2884,
+      "step": 800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9843242925755452e-05,
+      "loss": 0.2551,
+      "step": 801
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9842637525938468e-05,
+      "loss": 0.348,
+      "step": 802
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9842030968609574e-05,
+      "loss": 0.172,
+      "step": 803
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9841423253840113e-05,
+      "loss": 0.2832,
+      "step": 804
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9840814381701547e-05,
+      "loss": 0.1925,
+      "step": 805
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9840204352265476e-05,
+      "loss": 0.1623,
+      "step": 806
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983959316560365e-05,
+      "loss": 0.2191,
+      "step": 807
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983898082178794e-05,
+      "loss": 0.3226,
+      "step": 808
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983836732089036e-05,
+      "loss": 0.2034,
+      "step": 809
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9837752662983056e-05,
+      "loss": 0.2321,
+      "step": 810
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9837136848138317e-05,
+      "loss": 0.2435,
+      "step": 811
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9836519876428563e-05,
+      "loss": 0.1807,
+      "step": 812
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9835901747926352e-05,
+      "loss": 0.2673,
+      "step": 813
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9835282462704372e-05,
+      "loss": 0.2116,
+      "step": 814
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9834662020835456e-05,
+      "loss": 0.1573,
+      "step": 815
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983404042239257e-05,
+      "loss": 0.2065,
+      "step": 816
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9833417667448812e-05,
+      "loss": 0.1911,
+      "step": 817
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9832793756077423e-05,
+      "loss": 0.2327,
+      "step": 818
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9832168688351772e-05,
+      "loss": 0.3149,
+      "step": 819
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983154246434537e-05,
+      "loss": 0.3124,
+      "step": 820
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9830915084131864e-05,
+      "loss": 0.1862,
+      "step": 821
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.983028654778503e-05,
+      "loss": 0.18,
+      "step": 822
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9829656855378787e-05,
+      "loss": 0.3075,
+      "step": 823
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9829026006987192e-05,
+      "loss": 0.3959,
+      "step": 824
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9828394002684432e-05,
+      "loss": 0.2593,
+      "step": 825
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982776084254483e-05,
+      "loss": 0.128,
+      "step": 826
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9827126526642843e-05,
+      "loss": 0.174,
+      "step": 827
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9826491055053075e-05,
+      "loss": 0.2636,
+      "step": 828
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982585442785026e-05,
+      "loss": 0.1244,
+      "step": 829
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982521664510926e-05,
+      "loss": 0.1733,
+      "step": 830
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982457770690508e-05,
+      "loss": 0.2366,
+      "step": 831
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9823937613312863e-05,
+      "loss": 0.272,
+      "step": 832
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982329636440788e-05,
+      "loss": 0.1595,
+      "step": 833
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9822653960265552e-05,
+      "loss": 0.201,
+      "step": 834
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982201040096142e-05,
+      "loss": 0.2166,
+      "step": 835
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9821365686571172e-05,
+      "loss": 0.2518,
+      "step": 836
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9820719817170625e-05,
+      "loss": 0.1737,
+      "step": 837
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9820072792835738e-05,
+      "loss": 0.2317,
+      "step": 838
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.981942461364259e-05,
+      "loss": 0.2609,
+      "step": 839
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9818775279667425e-05,
+      "loss": 0.2377,
+      "step": 840
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9818124790986595e-05,
+      "loss": 0.2015,
+      "step": 841
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.98174731476766e-05,
+      "loss": 0.2761,
+      "step": 842
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9816820349814078e-05,
+      "loss": 0.1878,
+      "step": 843
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9816166397475796e-05,
+      "loss": 0.3091,
+      "step": 844
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9815511290738666e-05,
+      "loss": 0.4403,
+      "step": 845
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981485502967972e-05,
+      "loss": 0.1977,
+      "step": 846
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9814197614376137e-05,
+      "loss": 0.167,
+      "step": 847
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9813539044905238e-05,
+      "loss": 0.272,
+      "step": 848
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981287932134447e-05,
+      "loss": 0.2473,
+      "step": 849
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9812218443771412e-05,
+      "loss": 0.2149,
+      "step": 850
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9811556412263788e-05,
+      "loss": 0.2295,
+      "step": 851
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9810893226899453e-05,
+      "loss": 0.3151,
+      "step": 852
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9810228887756403e-05,
+      "loss": 0.4157,
+      "step": 853
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9809563394912757e-05,
+      "loss": 0.2262,
+      "step": 854
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9808896748446788e-05,
+      "loss": 0.1367,
+      "step": 855
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9808228948436883e-05,
+      "loss": 0.1457,
+      "step": 856
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.980755999496159e-05,
+      "loss": 0.1301,
+      "step": 857
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9806889888099573e-05,
+      "loss": 0.1416,
+      "step": 858
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9806218627929633e-05,
+      "loss": 0.3045,
+      "step": 859
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.980554621453072e-05,
+      "loss": 0.2247,
+      "step": 860
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9804872647981903e-05,
+      "loss": 0.1511,
+      "step": 861
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.98041979283624e-05,
+      "loss": 0.1647,
+      "step": 862
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9803522055751557e-05,
+      "loss": 0.2222,
+      "step": 863
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.980284503022886e-05,
+      "loss": 0.2678,
+      "step": 864
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9802166851873925e-05,
+      "loss": 0.3829,
+      "step": 865
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9801487520766513e-05,
+      "loss": 0.2994,
+      "step": 866
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9800807036986504e-05,
+      "loss": 0.2398,
+      "step": 867
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9800125400613935e-05,
+      "loss": 0.3297,
+      "step": 868
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9799442611728964e-05,
+      "loss": 0.2428,
+      "step": 869
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9798758670411886e-05,
+      "loss": 0.2451,
+      "step": 870
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9798073576743137e-05,
+      "loss": 0.2817,
+      "step": 871
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979738733080328e-05,
+      "loss": 0.2186,
+      "step": 872
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9796699932673023e-05,
+      "loss": 0.2541,
+      "step": 873
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9796011382433206e-05,
+      "loss": 0.1578,
+      "step": 874
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9795321680164798e-05,
+      "loss": 0.1974,
+      "step": 875
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979463082594892e-05,
+      "loss": 0.313,
+      "step": 876
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9793938819866806e-05,
+      "loss": 0.2081,
+      "step": 877
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9793245661999845e-05,
+      "loss": 0.1746,
+      "step": 878
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.979255135242955e-05,
+      "loss": 0.1484,
+      "step": 879
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9791855891237577e-05,
+      "loss": 0.2144,
+      "step": 880
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9791159278505706e-05,
+      "loss": 0.3924,
+      "step": 881
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9790461514315868e-05,
+      "loss": 0.1738,
+      "step": 882
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9789762598750115e-05,
+      "loss": 0.1677,
+      "step": 883
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9789062531890642e-05,
+      "loss": 0.2613,
+      "step": 884
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9788361313819783e-05,
+      "loss": 0.2109,
+      "step": 885
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9787658944619998e-05,
+      "loss": 0.165,
+      "step": 886
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978695542437389e-05,
+      "loss": 0.3092,
+      "step": 887
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978625075316419e-05,
+      "loss": 0.2844,
+      "step": 888
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9785544931073773e-05,
+      "loss": 0.2652,
+      "step": 889
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9784837958185637e-05,
+      "loss": 0.274,
+      "step": 890
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9784129834582934e-05,
+      "loss": 0.2027,
+      "step": 891
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9783420560348933e-05,
+      "loss": 0.2614,
+      "step": 892
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978271013556705e-05,
+      "loss": 0.2472,
+      "step": 893
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9781998560320832e-05,
+      "loss": 0.0935,
+      "step": 894
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978128583469396e-05,
+      "loss": 0.1297,
+      "step": 895
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9780571958770253e-05,
+      "loss": 0.1567,
+      "step": 896
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.977985693263366e-05,
+      "loss": 0.2363,
+      "step": 897
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9779140756368276e-05,
+      "loss": 0.2857,
+      "step": 898
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977842343005832e-05,
+      "loss": 0.2544,
+      "step": 899
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977770495378815e-05,
+      "loss": 0.1812,
+      "step": 900
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9776985327642264e-05,
+      "loss": 0.3876,
+      "step": 901
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9776264551705292e-05,
+      "loss": 0.2631,
+      "step": 902
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9775542626061993e-05,
+      "loss": 0.3296,
+      "step": 903
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977481955079727e-05,
+      "loss": 0.2612,
+      "step": 904
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9774095325996158e-05,
+      "loss": 0.1275,
+      "step": 905
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977336995174383e-05,
+      "loss": 0.1828,
+      "step": 906
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9772643428125585e-05,
+      "loss": 0.2965,
+      "step": 907
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9771915755226865e-05,
+      "loss": 0.3017,
+      "step": 908
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977118693313325e-05,
+      "loss": 0.1835,
+      "step": 909
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977045696193045e-05,
+      "loss": 0.2463,
+      "step": 910
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9769725841704307e-05,
+      "loss": 0.3461,
+      "step": 911
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.97689935725408e-05,
+      "loss": 0.2464,
+      "step": 912
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9768260154526056e-05,
+      "loss": 0.1981,
+      "step": 913
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9767525587746318e-05,
+      "loss": 0.1704,
+      "step": 914
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.976678987228797e-05,
+      "loss": 0.3053,
+      "step": 915
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9766053008237543e-05,
+      "loss": 0.2918,
+      "step": 916
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9765314995681687e-05,
+      "loss": 0.2886,
+      "step": 917
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9764575834707194e-05,
+      "loss": 0.2715,
+      "step": 918
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976383552540099e-05,
+      "loss": 0.2385,
+      "step": 919
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976309406785014e-05,
+      "loss": 0.275,
+      "step": 920
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9762351462141835e-05,
+      "loss": 0.3184,
+      "step": 921
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976160770836341e-05,
+      "loss": 0.1751,
+      "step": 922
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976086280660234e-05,
+      "loss": 0.1862,
+      "step": 923
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976011675694621e-05,
+      "loss": 0.2146,
+      "step": 924
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975936955948277e-05,
+      "loss": 0.2916,
+      "step": 925
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975862121429989e-05,
+      "loss": 0.1843,
+      "step": 926
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9757871721485575e-05,
+      "loss": 0.1343,
+      "step": 927
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975712108112796e-05,
+      "loss": 0.1698,
+      "step": 928
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975636929331533e-05,
+      "loss": 0.3007,
+      "step": 929
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9755616358136096e-05,
+      "loss": 0.3978,
+      "step": 930
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.97548622756788e-05,
+      "loss": 0.2306,
+      "step": 931
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9754107046032128e-05,
+      "loss": 0.1514,
+      "step": 932
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.97533506692849e-05,
+      "loss": 0.2095,
+      "step": 933
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9752593145526053e-05,
+      "loss": 0.1638,
+      "step": 934
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9751834474844687e-05,
+      "loss": 0.234,
+      "step": 935
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9751074657330015e-05,
+      "loss": 0.1651,
+      "step": 936
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9750313693071402e-05,
+      "loss": 0.1683,
+      "step": 937
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9749551582158325e-05,
+      "loss": 0.2788,
+      "step": 938
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9748788324680424e-05,
+      "loss": 0.265,
+      "step": 939
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974802392072745e-05,
+      "loss": 0.2385,
+      "step": 940
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.97472583703893e-05,
+      "loss": 0.1556,
+      "step": 941
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974649167375601e-05,
+      "loss": 0.1458,
+      "step": 942
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9745723830917738e-05,
+      "loss": 0.2496,
+      "step": 943
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9744954841964782e-05,
+      "loss": 0.3032,
+      "step": 944
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9744184706987582e-05,
+      "loss": 0.2186,
+      "step": 945
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974341342607671e-05,
+      "loss": 0.3392,
+      "step": 946
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9742640999322863e-05,
+      "loss": 0.2763,
+      "step": 947
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9741867426816885e-05,
+      "loss": 0.3098,
+      "step": 948
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974109270864974e-05,
+      "loss": 0.1698,
+      "step": 949
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9740316844912548e-05,
+      "loss": 0.1586,
+      "step": 950
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9739539835696546e-05,
+      "loss": 0.1538,
+      "step": 951
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9738761681093113e-05,
+      "loss": 0.1364,
+      "step": 952
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9737982381193765e-05,
+      "loss": 0.2318,
+      "step": 953
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.973720193609014e-05,
+      "loss": 0.3293,
+      "step": 954
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9736420345874026e-05,
+      "loss": 0.1767,
+      "step": 955
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.973563761063734e-05,
+      "loss": 0.2253,
+      "step": 956
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.973485373047213e-05,
+      "loss": 0.3146,
+      "step": 957
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9734068705470584e-05,
+      "loss": 0.2941,
+      "step": 958
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9733282535725022e-05,
+      "loss": 0.2359,
+      "step": 959
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9732495221327897e-05,
+      "loss": 0.2223,
+      "step": 960
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9731706762371803e-05,
+      "loss": 0.1922,
+      "step": 961
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.973091715894946e-05,
+      "loss": 0.2616,
+      "step": 962
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.973012641115373e-05,
+      "loss": 0.1962,
+      "step": 963
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.97293345190776e-05,
+      "loss": 0.2626,
+      "step": 964
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9728541482814206e-05,
+      "loss": 0.1771,
+      "step": 965
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9727747302456806e-05,
+      "loss": 0.2018,
+      "step": 966
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.97269519780988e-05,
+      "loss": 0.1227,
+      "step": 967
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9726155509833716e-05,
+      "loss": 0.2522,
+      "step": 968
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9725357897755223e-05,
+      "loss": 0.3823,
+      "step": 969
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9724559141957117e-05,
+      "loss": 0.214,
+      "step": 970
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972375924253334e-05,
+      "loss": 0.1412,
+      "step": 971
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9722958199577956e-05,
+      "loss": 0.2626,
+      "step": 972
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.972215601318517e-05,
+      "loss": 0.1708,
+      "step": 973
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9721352683449326e-05,
+      "loss": 0.1271,
+      "step": 974
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.972054821046489e-05,
+      "loss": 0.309,
+      "step": 975
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971974259432647e-05,
+      "loss": 0.2551,
+      "step": 976
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9718935835128815e-05,
+      "loss": 0.2904,
+      "step": 977
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9718127932966792e-05,
+      "loss": 0.2507,
+      "step": 978
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9717318887935417e-05,
+      "loss": 0.3428,
+      "step": 979
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9716508700129836e-05,
+      "loss": 0.2526,
+      "step": 980
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9715697369645327e-05,
+      "loss": 0.1575,
+      "step": 981
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9714884896577303e-05,
+      "loss": 0.2305,
+      "step": 982
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9714071281021316e-05,
+      "loss": 0.1874,
+      "step": 983
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971325652307304e-05,
+      "loss": 0.2085,
+      "step": 984
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.97124406228283e-05,
+      "loss": 0.2325,
+      "step": 985
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9711623580383044e-05,
+      "loss": 0.2458,
+      "step": 986
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9710805395833365e-05,
+      "loss": 0.2143,
+      "step": 987
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.970998606927547e-05,
+      "loss": 0.176,
+      "step": 988
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9709165600805723e-05,
+      "loss": 0.3346,
+      "step": 989
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.970834399052061e-05,
+      "loss": 0.1862,
+      "step": 990
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.970752123851675e-05,
+      "loss": 0.2819,
+      "step": 991
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.970669734489091e-05,
+      "loss": 0.2975,
+      "step": 992
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.970587230973997e-05,
+      "loss": 0.2153,
+      "step": 993
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.970504613316096e-05,
+      "loss": 0.2243,
+      "step": 994
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.970421881525104e-05,
+      "loss": 0.0933,
+      "step": 995
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.970339035610751e-05,
+      "loss": 0.3152,
+      "step": 996
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9702560755827788e-05,
+      "loss": 0.2311,
+      "step": 997
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9701730014509444e-05,
+      "loss": 0.216,
+      "step": 998
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.970089813225017e-05,
+      "loss": 0.1034,
+      "step": 999
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.97000651091478e-05,
+      "loss": 0.2811,
+      "step": 1000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9699230945300297e-05,
+      "loss": 0.132,
+      "step": 1001
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9698395640805762e-05,
+      "loss": 0.2958,
+      "step": 1002
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9697559195762427e-05,
+      "loss": 0.2023,
+      "step": 1003
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969672161026866e-05,
+      "loss": 0.298,
+      "step": 1004
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969588288442296e-05,
+      "loss": 0.1617,
+      "step": 1005
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9695043018323967e-05,
+      "loss": 0.1336,
+      "step": 1006
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969420201207045e-05,
+      "loss": 0.2256,
+      "step": 1007
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9693359865761308e-05,
+      "loss": 0.2441,
+      "step": 1008
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9692516579495584e-05,
+      "loss": 0.3728,
+      "step": 1009
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969167215337245e-05,
+      "loss": 0.2378,
+      "step": 1010
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.969082658749121e-05,
+      "loss": 0.2162,
+      "step": 1011
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9689979881951303e-05,
+      "loss": 0.3822,
+      "step": 1012
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9689132036852305e-05,
+      "loss": 0.2104,
+      "step": 1013
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9688283052293923e-05,
+      "loss": 0.2683,
+      "step": 1014
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9687432928376006e-05,
+      "loss": 0.217,
+      "step": 1015
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9686581665198518e-05,
+      "loss": 0.1587,
+      "step": 1016
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9685729262861577e-05,
+      "loss": 0.2902,
+      "step": 1017
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9684875721465426e-05,
+      "loss": 0.3171,
+      "step": 1018
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968402104111044e-05,
+      "loss": 0.2493,
+      "step": 1019
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9683165221897132e-05,
+      "loss": 0.204,
+      "step": 1020
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9682308263926154e-05,
+      "loss": 0.1488,
+      "step": 1021
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9681450167298278e-05,
+      "loss": 0.2084,
+      "step": 1022
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968059093211442e-05,
+      "loss": 0.2233,
+      "step": 1023
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.967973055847563e-05,
+      "loss": 0.1277,
+      "step": 1024
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9678869046483085e-05,
+      "loss": 0.3581,
+      "step": 1025
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9678006396238106e-05,
+      "loss": 0.1982,
+      "step": 1026
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9677142607842136e-05,
+      "loss": 0.3801,
+      "step": 1027
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9676277681396763e-05,
+      "loss": 0.2929,
+      "step": 1028
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.96754116170037e-05,
+      "loss": 0.2391,
+      "step": 1029
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9674544414764804e-05,
+      "loss": 0.1938,
+      "step": 1030
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9673676074782054e-05,
+      "loss": 0.1528,
+      "step": 1031
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.967280659715757e-05,
+      "loss": 0.2734,
+      "step": 1032
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.96719359819936e-05,
+      "loss": 0.2014,
+      "step": 1033
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9671064229392538e-05,
+      "loss": 0.3444,
+      "step": 1034
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.96701913394569e-05,
+      "loss": 0.301,
+      "step": 1035
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9669317312289336e-05,
+      "loss": 0.37,
+      "step": 1036
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966844214799264e-05,
+      "loss": 0.1441,
+      "step": 1037
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9667565846669725e-05,
+      "loss": 0.2402,
+      "step": 1038
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966668840842365e-05,
+      "loss": 0.273,
+      "step": 1039
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9665809833357606e-05,
+      "loss": 0.1766,
+      "step": 1040
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966493012157491e-05,
+      "loss": 0.2498,
+      "step": 1041
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966404927317902e-05,
+      "loss": 0.0752,
+      "step": 1042
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9663167288273523e-05,
+      "loss": 0.1603,
+      "step": 1043
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9662284166962147e-05,
+      "loss": 0.2523,
+      "step": 1044
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9661399909348743e-05,
+      "loss": 0.2296,
+      "step": 1045
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9660514515537307e-05,
+      "loss": 0.2516,
+      "step": 1046
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.965962798563196e-05,
+      "loss": 0.1408,
+      "step": 1047
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9658740319736958e-05,
+      "loss": 0.2341,
+      "step": 1048
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9657851517956693e-05,
+      "loss": 0.1969,
+      "step": 1049
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.965696158039569e-05,
+      "loss": 0.2395,
+      "step": 1050
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9656070507158606e-05,
+      "loss": 0.1543,
+      "step": 1051
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9655178298350236e-05,
+      "loss": 0.2465,
+      "step": 1052
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9654284954075505e-05,
+      "loss": 0.2708,
+      "step": 1053
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9653390474439467e-05,
+      "loss": 0.1824,
+      "step": 1054
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9652494859547323e-05,
+      "loss": 0.232,
+      "step": 1055
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9651598109504394e-05,
+      "loss": 0.1362,
+      "step": 1056
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9650700224416136e-05,
+      "loss": 0.1864,
+      "step": 1057
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9649801204388144e-05,
+      "loss": 0.3072,
+      "step": 1058
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964890104952615e-05,
+      "loss": 0.1758,
+      "step": 1059
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9647999759936005e-05,
+      "loss": 0.1737,
+      "step": 1060
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964709733572371e-05,
+      "loss": 0.2002,
+      "step": 1061
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9646193776995383e-05,
+      "loss": 0.368,
+      "step": 1062
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9645289083857295e-05,
+      "loss": 0.144,
+      "step": 1063
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9644383256415834e-05,
+      "loss": 0.4139,
+      "step": 1064
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9643476294777525e-05,
+      "loss": 0.1764,
+      "step": 1065
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964256819904903e-05,
+      "loss": 0.2327,
+      "step": 1066
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9641658969337146e-05,
+      "loss": 0.3948,
+      "step": 1067
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9640748605748793e-05,
+      "loss": 0.3572,
+      "step": 1068
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9639837108391038e-05,
+      "loss": 0.2703,
+      "step": 1069
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963892447737107e-05,
+      "loss": 0.1806,
+      "step": 1070
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963801071279622e-05,
+      "loss": 0.1526,
+      "step": 1071
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9637095814773947e-05,
+      "loss": 0.1524,
+      "step": 1072
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9636179783411845e-05,
+      "loss": 0.2414,
+      "step": 1073
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9635262618817637e-05,
+      "loss": 0.2099,
+      "step": 1074
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963434432109919e-05,
+      "loss": 0.1792,
+      "step": 1075
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9633424890364493e-05,
+      "loss": 0.1342,
+      "step": 1076
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963250432672167e-05,
+      "loss": 0.1416,
+      "step": 1077
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9631582630278992e-05,
+      "loss": 0.2,
+      "step": 1078
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963065980114484e-05,
+      "loss": 0.2333,
+      "step": 1079
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9629735839427745e-05,
+      "loss": 0.2384,
+      "step": 1080
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9628810745236367e-05,
+      "loss": 0.1867,
+      "step": 1081
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9627884518679495e-05,
+      "loss": 0.2061,
+      "step": 1082
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9626957159866064e-05,
+      "loss": 0.3485,
+      "step": 1083
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9626028668905125e-05,
+      "loss": 0.2785,
+      "step": 1084
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9625099045905872e-05,
+      "loss": 0.1563,
+      "step": 1085
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9624168290977634e-05,
+      "loss": 0.2906,
+      "step": 1086
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9623236404229863e-05,
+      "loss": 0.2105,
+      "step": 1087
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9622303385772155e-05,
+      "loss": 0.2326,
+      "step": 1088
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9621369235714233e-05,
+      "loss": 0.1755,
+      "step": 1089
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9620433954165955e-05,
+      "loss": 0.2255,
+      "step": 1090
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9619497541237312e-05,
+      "loss": 0.1886,
+      "step": 1091
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961855999703843e-05,
+      "loss": 0.2121,
+      "step": 1092
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9617621321679567e-05,
+      "loss": 0.1819,
+      "step": 1093
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.96166815152711e-05,
+      "loss": 0.148,
+      "step": 1094
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961574057792357e-05,
+      "loss": 0.2027,
+      "step": 1095
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961479850974762e-05,
+      "loss": 0.1469,
+      "step": 1096
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9613855310854047e-05,
+      "loss": 0.2118,
+      "step": 1097
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961291098135377e-05,
+      "loss": 0.2613,
+      "step": 1098
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9611965521357838e-05,
+      "loss": 0.3377,
+      "step": 1099
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9611018930977444e-05,
+      "loss": 0.1808,
+      "step": 1100
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9610071210323916e-05,
+      "loss": 0.3001,
+      "step": 1101
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9609122359508696e-05,
+      "loss": 0.3536,
+      "step": 1102
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9608172378643375e-05,
+      "loss": 0.1815,
+      "step": 1103
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.960722126783967e-05,
+      "loss": 0.4018,
+      "step": 1104
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.960626902720944e-05,
+      "loss": 0.1831,
+      "step": 1105
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9605315656864667e-05,
+      "loss": 0.2771,
+      "step": 1106
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9604361156917466e-05,
+      "loss": 0.2214,
+      "step": 1107
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9603405527480088e-05,
+      "loss": 0.1722,
+      "step": 1108
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9602448768664922e-05,
+      "loss": 0.2414,
+      "step": 1109
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.960149088058448e-05,
+      "loss": 0.1707,
+      "step": 1110
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.960053186335141e-05,
+      "loss": 0.2213,
+      "step": 1111
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.95995717170785e-05,
+      "loss": 0.0974,
+      "step": 1112
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9598610441878664e-05,
+      "loss": 0.2996,
+      "step": 1113
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9597648037864942e-05,
+      "loss": 0.2827,
+      "step": 1114
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9596684505150522e-05,
+      "loss": 0.1868,
+      "step": 1115
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9595719843848718e-05,
+      "loss": 0.2708,
+      "step": 1116
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9594754054072972e-05,
+      "loss": 0.2159,
+      "step": 1117
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959378713593686e-05,
+      "loss": 0.2345,
+      "step": 1118
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9592819089554102e-05,
+      "loss": 0.2306,
+      "step": 1119
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959184991503853e-05,
+      "loss": 0.2445,
+      "step": 1120
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9590879612504135e-05,
+      "loss": 0.2116,
+      "step": 1121
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9589908182065015e-05,
+      "loss": 0.086,
+      "step": 1122
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9588935623835416e-05,
+      "loss": 0.1988,
+      "step": 1123
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9587961937929712e-05,
+      "loss": 0.2034,
+      "step": 1124
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9586987124462413e-05,
+      "loss": 0.3266,
+      "step": 1125
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9586011183548155e-05,
+      "loss": 0.2543,
+      "step": 1126
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9585034115301712e-05,
+      "loss": 0.1635,
+      "step": 1127
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.958405591983799e-05,
+      "loss": 0.1636,
+      "step": 1128
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9583076597272025e-05,
+      "loss": 0.1945,
+      "step": 1129
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.958209614771899e-05,
+      "loss": 0.3579,
+      "step": 1130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.958111457129418e-05,
+      "loss": 0.2334,
+      "step": 1131
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.958013186811304e-05,
+      "loss": 0.2555,
+      "step": 1132
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957914803829113e-05,
+      "loss": 0.2919,
+      "step": 1133
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9578163081944153e-05,
+      "loss": 0.1487,
+      "step": 1134
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9577176999187943e-05,
+      "loss": 0.2203,
+      "step": 1135
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9576189790138466e-05,
+      "loss": 0.181,
+      "step": 1136
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9575201454911815e-05,
+      "loss": 0.2186,
+      "step": 1137
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9574211993624223e-05,
+      "loss": 0.2858,
+      "step": 1138
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9573221406392054e-05,
+      "loss": 0.2332,
+      "step": 1139
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9572229693331797e-05,
+      "loss": 0.1675,
+      "step": 1140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9571236854560084e-05,
+      "loss": 0.3713,
+      "step": 1141
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9570242890193673e-05,
+      "loss": 0.1899,
+      "step": 1142
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956924780034946e-05,
+      "loss": 0.2585,
+      "step": 1143
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9568251585144464e-05,
+      "loss": 0.184,
+      "step": 1144
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9567254244695847e-05,
+      "loss": 0.1987,
+      "step": 1145
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9566255779120888e-05,
+      "loss": 0.2882,
+      "step": 1146
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956525618853702e-05,
+      "loss": 0.2854,
+      "step": 1147
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956425547306179e-05,
+      "loss": 0.2279,
+      "step": 1148
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9563253632812885e-05,
+      "loss": 0.2558,
+      "step": 1149
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9562250667908126e-05,
+      "loss": 0.2544,
+      "step": 1150
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9561246578465462e-05,
+      "loss": 0.1571,
+      "step": 1151
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9560241364602972e-05,
+      "loss": 0.2617,
+      "step": 1152
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9559235026438876e-05,
+      "loss": 0.2714,
+      "step": 1153
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9558227564091517e-05,
+      "loss": 0.2637,
+      "step": 1154
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955721897767938e-05,
+      "loss": 0.2538,
+      "step": 1155
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9556209267321073e-05,
+      "loss": 0.2045,
+      "step": 1156
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955519843313534e-05,
+      "loss": 0.2941,
+      "step": 1157
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9554186475241054e-05,
+      "loss": 0.2176,
+      "step": 1158
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955317339375723e-05,
+      "loss": 0.1546,
+      "step": 1159
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9552159188803e-05,
+      "loss": 0.2807,
+      "step": 1160
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9551143860497645e-05,
+      "loss": 0.2889,
+      "step": 1161
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.955012740896056e-05,
+      "loss": 0.2632,
+      "step": 1162
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.954910983431129e-05,
+      "loss": 0.282,
+      "step": 1163
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9548091136669497e-05,
+      "loss": 0.227,
+      "step": 1164
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9547071316154985e-05,
+      "loss": 0.1438,
+      "step": 1165
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.954605037288769e-05,
+      "loss": 0.3144,
+      "step": 1166
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.954502830698767e-05,
+      "loss": 0.2189,
+      "step": 1167
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9544005118575127e-05,
+      "loss": 0.2347,
+      "step": 1168
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.954298080777039e-05,
+      "loss": 0.2196,
+      "step": 1169
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9541955374693912e-05,
+      "loss": 0.1557,
+      "step": 1170
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.95409288194663e-05,
+      "loss": 0.2376,
+      "step": 1171
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9539901142208262e-05,
+      "loss": 0.2249,
+      "step": 1172
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9538872343040663e-05,
+      "loss": 0.2425,
+      "step": 1173
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9537842422084496e-05,
+      "loss": 0.1308,
+      "step": 1174
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9536811379460876e-05,
+      "loss": 0.2313,
+      "step": 1175
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9535779215291055e-05,
+      "loss": 0.2905,
+      "step": 1176
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9534745929696418e-05,
+      "loss": 0.248,
+      "step": 1177
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9533711522798485e-05,
+      "loss": 0.2744,
+      "step": 1178
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.95326759947189e-05,
+      "loss": 0.1781,
+      "step": 1179
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9531639345579445e-05,
+      "loss": 0.2285,
+      "step": 1180
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.953060157550203e-05,
+      "loss": 0.1916,
+      "step": 1181
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.95295626846087e-05,
+      "loss": 0.39,
+      "step": 1182
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9528522673021634e-05,
+      "loss": 0.2644,
+      "step": 1183
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9527481540863128e-05,
+      "loss": 0.3036,
+      "step": 1184
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9526439288255636e-05,
+      "loss": 0.1886,
+      "step": 1185
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9525395915321713e-05,
+      "loss": 0.2658,
+      "step": 1186
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9524351422184072e-05,
+      "loss": 0.3219,
+      "step": 1187
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952330580896555e-05,
+      "loss": 0.286,
+      "step": 1188
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9522259075789102e-05,
+      "loss": 0.1122,
+      "step": 1189
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9521211222777836e-05,
+      "loss": 0.2777,
+      "step": 1190
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9520162250054974e-05,
+      "loss": 0.138,
+      "step": 1191
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951911215774388e-05,
+      "loss": 0.3283,
+      "step": 1192
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951806094596805e-05,
+      "loss": 0.2576,
+      "step": 1193
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9517008614851108e-05,
+      "loss": 0.2784,
+      "step": 1194
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.95159551645168e-05,
+      "loss": 0.2882,
+      "step": 1195
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9514900595089026e-05,
+      "loss": 0.231,
+      "step": 1196
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.95138449066918e-05,
+      "loss": 0.2591,
+      "step": 1197
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9512788099449272e-05,
+      "loss": 0.2427,
+      "step": 1198
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9511730173485724e-05,
+      "loss": 0.2306,
+      "step": 1199
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9510671128925574e-05,
+      "loss": 0.2023,
+      "step": 1200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950961096589336e-05,
+      "loss": 0.2781,
+      "step": 1201
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950854968451377e-05,
+      "loss": 0.1792,
+      "step": 1202
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9507487284911604e-05,
+      "loss": 0.1428,
+      "step": 1203
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9506423767211808e-05,
+      "loss": 0.2511,
+      "step": 1204
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9505359131539445e-05,
+      "loss": 0.2125,
+      "step": 1205
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950429337801973e-05,
+      "loss": 0.23,
+      "step": 1206
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9503226506777987e-05,
+      "loss": 0.3689,
+      "step": 1207
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9502158517939688e-05,
+      "loss": 0.115,
+      "step": 1208
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9501089411630425e-05,
+      "loss": 0.187,
+      "step": 1209
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9500019187975935e-05,
+      "loss": 0.2389,
+      "step": 1210
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.949894784710207e-05,
+      "loss": 0.1993,
+      "step": 1211
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9497875389134827e-05,
+      "loss": 0.2884,
+      "step": 1212
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9496801814200326e-05,
+      "loss": 0.253,
+      "step": 1213
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9495727122424824e-05,
+      "loss": 0.1692,
+      "step": 1214
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9494651313934704e-05,
+      "loss": 0.2256,
+      "step": 1215
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9493574388856483e-05,
+      "loss": 0.2942,
+      "step": 1216
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9492496347316813e-05,
+      "loss": 0.1315,
+      "step": 1217
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.949141718944247e-05,
+      "loss": 0.1859,
+      "step": 1218
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9490336915360365e-05,
+      "loss": 0.2195,
+      "step": 1219
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9489255525197544e-05,
+      "loss": 0.246,
+      "step": 1220
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948817301908118e-05,
+      "loss": 0.1878,
+      "step": 1221
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948708939713857e-05,
+      "loss": 0.3291,
+      "step": 1222
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9486004659497156e-05,
+      "loss": 0.334,
+      "step": 1223
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9484918806284506e-05,
+      "loss": 0.2814,
+      "step": 1224
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9483831837628316e-05,
+      "loss": 0.1316,
+      "step": 1225
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9482743753656415e-05,
+      "loss": 0.1793,
+      "step": 1226
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9481654554496764e-05,
+      "loss": 0.3554,
+      "step": 1227
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9480564240277456e-05,
+      "loss": 0.2298,
+      "step": 1228
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9479472811126715e-05,
+      "loss": 0.2168,
+      "step": 1229
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9478380267172892e-05,
+      "loss": 0.2202,
+      "step": 1230
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9477286608544477e-05,
+      "loss": 0.2331,
+      "step": 1231
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947619183537008e-05,
+      "loss": 0.2161,
+      "step": 1232
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9475095947778452e-05,
+      "loss": 0.1463,
+      "step": 1233
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947399894589847e-05,
+      "loss": 0.1232,
+      "step": 1234
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9472900829859145e-05,
+      "loss": 0.2919,
+      "step": 1235
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947180159978962e-05,
+      "loss": 0.1745,
+      "step": 1236
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9470701255819152e-05,
+      "loss": 0.2978,
+      "step": 1237
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946959979807716e-05,
+      "loss": 0.2326,
+      "step": 1238
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9468497226693176e-05,
+      "loss": 0.2565,
+      "step": 1239
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9467393541796856e-05,
+      "loss": 0.3535,
+      "step": 1240
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9466288743517997e-05,
+      "loss": 0.245,
+      "step": 1241
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946518283198653e-05,
+      "loss": 0.2595,
+      "step": 1242
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9464075807332514e-05,
+      "loss": 0.2275,
+      "step": 1243
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9462967669686125e-05,
+      "loss": 0.2255,
+      "step": 1244
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9461858419177692e-05,
+      "loss": 0.1869,
+      "step": 1245
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9460748055937666e-05,
+      "loss": 0.1723,
+      "step": 1246
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9459636580096624e-05,
+      "loss": 0.2814,
+      "step": 1247
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9458523991785277e-05,
+      "loss": 0.1526,
+      "step": 1248
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9457410291134465e-05,
+      "loss": 0.242,
+      "step": 1249
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945629547827517e-05,
+      "loss": 0.1691,
+      "step": 1250
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9455179553338488e-05,
+      "loss": 0.2316,
+      "step": 1251
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9454062516455656e-05,
+      "loss": 0.1382,
+      "step": 1252
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945294436775804e-05,
+      "loss": 0.402,
+      "step": 1253
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9451825107377134e-05,
+      "loss": 0.2767,
+      "step": 1254
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9450704735444574e-05,
+      "loss": 0.3264,
+      "step": 1255
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9449583252092107e-05,
+      "loss": 0.2093,
+      "step": 1256
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9448460657451625e-05,
+      "loss": 0.4007,
+      "step": 1257
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9447336951655148e-05,
+      "loss": 0.2115,
+      "step": 1258
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9446212134834824e-05,
+      "loss": 0.0645,
+      "step": 1259
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9445086207122937e-05,
+      "loss": 0.1686,
+      "step": 1260
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9443959168651893e-05,
+      "loss": 0.3036,
+      "step": 1261
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.944283101955424e-05,
+      "loss": 0.136,
+      "step": 1262
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.944170175996265e-05,
+      "loss": 0.1231,
+      "step": 1263
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9440571390009916e-05,
+      "loss": 0.3368,
+      "step": 1264
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9439439909828984e-05,
+      "loss": 0.2634,
+      "step": 1265
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.943830731955291e-05,
+      "loss": 0.1939,
+      "step": 1266
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9437173619314892e-05,
+      "loss": 0.1925,
+      "step": 1267
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9436038809248256e-05,
+      "loss": 0.3684,
+      "step": 1268
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9434902889486457e-05,
+      "loss": 0.4294,
+      "step": 1269
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9433765860163077e-05,
+      "loss": 0.241,
+      "step": 1270
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.943262772141184e-05,
+      "loss": 0.247,
+      "step": 1271
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9431488473366588e-05,
+      "loss": 0.1654,
+      "step": 1272
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.94303481161613e-05,
+      "loss": 0.1558,
+      "step": 1273
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9429206649930085e-05,
+      "loss": 0.3138,
+      "step": 1274
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9428064074807178e-05,
+      "loss": 0.2506,
+      "step": 1275
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9426920390926952e-05,
+      "loss": 0.2667,
+      "step": 1276
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9425775598423903e-05,
+      "loss": 0.1473,
+      "step": 1277
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9424629697432668e-05,
+      "loss": 0.1367,
+      "step": 1278
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9423482688087996e-05,
+      "loss": 0.2106,
+      "step": 1279
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9422334570524783e-05,
+      "loss": 0.2063,
+      "step": 1280
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9421185344878052e-05,
+      "loss": 0.1977,
+      "step": 1281
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9420035011282953e-05,
+      "loss": 0.2218,
+      "step": 1282
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9418883569874763e-05,
+      "loss": 0.1182,
+      "step": 1283
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9417731020788896e-05,
+      "loss": 0.2592,
+      "step": 1284
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9416577364160894e-05,
+      "loss": 0.2345,
+      "step": 1285
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941542260012643e-05,
+      "loss": 0.2426,
+      "step": 1286
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9414266728821305e-05,
+      "loss": 0.1643,
+      "step": 1287
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9413109750381457e-05,
+      "loss": 0.2082,
+      "step": 1288
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9411951664942936e-05,
+      "loss": 0.2421,
+      "step": 1289
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9410792472641947e-05,
+      "loss": 0.24,
+      "step": 1290
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.940963217361481e-05,
+      "loss": 0.2385,
+      "step": 1291
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9408470767997977e-05,
+      "loss": 0.2667,
+      "step": 1292
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9407308255928032e-05,
+      "loss": 0.2387,
+      "step": 1293
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9406144637541686e-05,
+      "loss": 0.259,
+      "step": 1294
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9404979912975788e-05,
+      "loss": 0.3143,
+      "step": 1295
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9403814082367307e-05,
+      "loss": 0.2028,
+      "step": 1296
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9402647145853346e-05,
+      "loss": 0.2931,
+      "step": 1297
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9401479103571143e-05,
+      "loss": 0.13,
+      "step": 1298
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.940030995565806e-05,
+      "loss": 0.1537,
+      "step": 1299
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9399139702251593e-05,
+      "loss": 0.2832,
+      "step": 1300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9397968343489362e-05,
+      "loss": 0.2694,
+      "step": 1301
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9396795879509122e-05,
+      "loss": 0.1629,
+      "step": 1302
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9395622310448757e-05,
+      "loss": 0.1833,
+      "step": 1303
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9394447636446286e-05,
+      "loss": 0.1858,
+      "step": 1304
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9393271857639843e-05,
+      "loss": 0.2814,
+      "step": 1305
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.939209497416771e-05,
+      "loss": 0.3063,
+      "step": 1306
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9390916986168285e-05,
+      "loss": 0.1975,
+      "step": 1307
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9389737893780106e-05,
+      "loss": 0.2183,
+      "step": 1308
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9388557697141836e-05,
+      "loss": 0.1671,
+      "step": 1309
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.938737639639226e-05,
+      "loss": 0.2644,
+      "step": 1310
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9386193991670313e-05,
+      "loss": 0.2739,
+      "step": 1311
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9385010483115042e-05,
+      "loss": 0.2603,
+      "step": 1312
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.938382587086563e-05,
+      "loss": 0.1023,
+      "step": 1313
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.938264015506139e-05,
+      "loss": 0.246,
+      "step": 1314
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9381453335841766e-05,
+      "loss": 0.2591,
+      "step": 1315
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9380265413346328e-05,
+      "loss": 0.2055,
+      "step": 1316
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9379076387714776e-05,
+      "loss": 0.2952,
+      "step": 1317
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9377886259086943e-05,
+      "loss": 0.2137,
+      "step": 1318
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9376695027602795e-05,
+      "loss": 0.2033,
+      "step": 1319
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9375502693402417e-05,
+      "loss": 0.284,
+      "step": 1320
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9374309256626033e-05,
+      "loss": 0.1042,
+      "step": 1321
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9373114717413986e-05,
+      "loss": 0.1126,
+      "step": 1322
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9371919075906768e-05,
+      "loss": 0.4085,
+      "step": 1323
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937072233224498e-05,
+      "loss": 0.1127,
+      "step": 1324
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9369524486569363e-05,
+      "loss": 0.2494,
+      "step": 1325
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.936832553902079e-05,
+      "loss": 0.3111,
+      "step": 1326
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9367125489740255e-05,
+      "loss": 0.1759,
+      "step": 1327
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9365924338868886e-05,
+      "loss": 0.2598,
+      "step": 1328
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9364722086547944e-05,
+      "loss": 0.3083,
+      "step": 1329
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.936351873291881e-05,
+      "loss": 0.2815,
+      "step": 1330
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9362314278123e-05,
+      "loss": 0.273,
+      "step": 1331
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.936110872230217e-05,
+      "loss": 0.2435,
+      "step": 1332
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935990206559809e-05,
+      "loss": 0.1684,
+      "step": 1333
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935869430815266e-05,
+      "loss": 0.1997,
+      "step": 1334
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9357485450107926e-05,
+      "loss": 0.2982,
+      "step": 1335
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9356275491606042e-05,
+      "loss": 0.1853,
+      "step": 1336
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9355064432789298e-05,
+      "loss": 0.2232,
+      "step": 1337
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935385227380013e-05,
+      "loss": 0.1443,
+      "step": 1338
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9352639014781082e-05,
+      "loss": 0.187,
+      "step": 1339
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9351424655874834e-05,
+      "loss": 0.169,
+      "step": 1340
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9350209197224203e-05,
+      "loss": 0.2849,
+      "step": 1341
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9348992638972123e-05,
+      "loss": 0.2824,
+      "step": 1342
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9347774981261666e-05,
+      "loss": 0.1937,
+      "step": 1343
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934655622423603e-05,
+      "loss": 0.2285,
+      "step": 1344
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9345336368038544e-05,
+      "loss": 0.2055,
+      "step": 1345
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934411541281266e-05,
+      "loss": 0.2019,
+      "step": 1346
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9342893358701978e-05,
+      "loss": 0.2124,
+      "step": 1347
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9341670205850202e-05,
+      "loss": 0.0712,
+      "step": 1348
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934044595440118e-05,
+      "loss": 0.1095,
+      "step": 1349
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9339220604498884e-05,
+      "loss": 0.2891,
+      "step": 1350
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9337994156287422e-05,
+      "loss": 0.2315,
+      "step": 1351
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933676660991102e-05,
+      "loss": 0.1481,
+      "step": 1352
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933553796551405e-05,
+      "loss": 0.1738,
+      "step": 1353
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9334308223240992e-05,
+      "loss": 0.2271,
+      "step": 1354
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9333077383236472e-05,
+      "loss": 0.2408,
+      "step": 1355
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933184544564524e-05,
+      "loss": 0.0858,
+      "step": 1356
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9330612410612168e-05,
+      "loss": 0.2023,
+      "step": 1357
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9329378278282268e-05,
+      "loss": 0.237,
+      "step": 1358
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9328143048800675e-05,
+      "loss": 0.2599,
+      "step": 1359
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9326906722312658e-05,
+      "loss": 0.2816,
+      "step": 1360
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.93256692989636e-05,
+      "loss": 0.129,
+      "step": 1361
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9324430778899036e-05,
+      "loss": 0.1766,
+      "step": 1362
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9323191162264615e-05,
+      "loss": 0.1471,
+      "step": 1363
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9321950449206114e-05,
+      "loss": 0.1175,
+      "step": 1364
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932070863986945e-05,
+      "loss": 0.3068,
+      "step": 1365
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9319465734400653e-05,
+      "loss": 0.1385,
+      "step": 1366
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9318221732945898e-05,
+      "loss": 0.2332,
+      "step": 1367
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.931697663565148e-05,
+      "loss": 0.1461,
+      "step": 1368
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9315730442663827e-05,
+      "loss": 0.3662,
+      "step": 1369
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.931448315412949e-05,
+      "loss": 0.1823,
+      "step": 1370
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.931323477019515e-05,
+      "loss": 0.2806,
+      "step": 1371
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9311985291007625e-05,
+      "loss": 0.1355,
+      "step": 1372
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9310734716713853e-05,
+      "loss": 0.2709,
+      "step": 1373
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9309483047460905e-05,
+      "loss": 0.2423,
+      "step": 1374
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9308230283395975e-05,
+      "loss": 0.3683,
+      "step": 1375
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.93069764246664e-05,
+      "loss": 0.1169,
+      "step": 1376
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.930572147141963e-05,
+      "loss": 0.1897,
+      "step": 1377
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9304465423803247e-05,
+      "loss": 0.2888,
+      "step": 1378
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.930320828196497e-05,
+      "loss": 0.2355,
+      "step": 1379
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9301950046052635e-05,
+      "loss": 0.2062,
+      "step": 1380
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.930069071621422e-05,
+      "loss": 0.2082,
+      "step": 1381
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929943029259782e-05,
+      "loss": 0.1881,
+      "step": 1382
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9298168775351663e-05,
+      "loss": 0.273,
+      "step": 1383
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9296906164624107e-05,
+      "loss": 0.2492,
+      "step": 1384
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929564246056364e-05,
+      "loss": 0.1773,
+      "step": 1385
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929437766331887e-05,
+      "loss": 0.319,
+      "step": 1386
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929311177303854e-05,
+      "loss": 0.1396,
+      "step": 1387
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9291844789871526e-05,
+      "loss": 0.3044,
+      "step": 1388
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9290576713966828e-05,
+      "loss": 0.1323,
+      "step": 1389
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9289307545473566e-05,
+      "loss": 0.2148,
+      "step": 1390
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9288037284541e-05,
+      "loss": 0.2116,
+      "step": 1391
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.928676593131852e-05,
+      "loss": 0.1855,
+      "step": 1392
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9285493485955634e-05,
+      "loss": 0.1948,
+      "step": 1393
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.928421994860199e-05,
+      "loss": 0.1949,
+      "step": 1394
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.928294531940735e-05,
+      "loss": 0.2034,
+      "step": 1395
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9281669598521616e-05,
+      "loss": 0.1506,
+      "step": 1396
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9280392786094816e-05,
+      "loss": 0.3205,
+      "step": 1397
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9279114882277105e-05,
+      "loss": 0.2807,
+      "step": 1398
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.927783588721877e-05,
+      "loss": 0.1097,
+      "step": 1399
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9276555801070217e-05,
+      "loss": 0.2422,
+      "step": 1400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.927527462398199e-05,
+      "loss": 0.3132,
+      "step": 1401
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9273992356104757e-05,
+      "loss": 0.2219,
+      "step": 1402
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9272708997589314e-05,
+      "loss": 0.1431,
+      "step": 1403
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.927142454858659e-05,
+      "loss": 0.1785,
+      "step": 1404
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9270139009247635e-05,
+      "loss": 0.3382,
+      "step": 1405
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9268852379723633e-05,
+      "loss": 0.1861,
+      "step": 1406
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9267564660165888e-05,
+      "loss": 0.1984,
+      "step": 1407
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9266275850725845e-05,
+      "loss": 0.2059,
+      "step": 1408
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.926498595155507e-05,
+      "loss": 0.1309,
+      "step": 1409
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9263694962805255e-05,
+      "loss": 0.1391,
+      "step": 1410
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9262402884628223e-05,
+      "loss": 0.2607,
+      "step": 1411
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9261109717175924e-05,
+      "loss": 0.2104,
+      "step": 1412
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.925981546060044e-05,
+      "loss": 0.3161,
+      "step": 1413
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9258520115053973e-05,
+      "loss": 0.209,
+      "step": 1414
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9257223680688862e-05,
+      "loss": 0.3029,
+      "step": 1415
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.925592615765757e-05,
+      "loss": 0.1875,
+      "step": 1416
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9254627546112683e-05,
+      "loss": 0.3341,
+      "step": 1417
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9253327846206925e-05,
+      "loss": 0.2471,
+      "step": 1418
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9252027058093143e-05,
+      "loss": 0.25,
+      "step": 1419
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.925072518192431e-05,
+      "loss": 0.153,
+      "step": 1420
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.924942221785353e-05,
+      "loss": 0.2234,
+      "step": 1421
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9248118166034035e-05,
+      "loss": 0.362,
+      "step": 1422
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.924681302661918e-05,
+      "loss": 0.2801,
+      "step": 1423
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9245506799762458e-05,
+      "loss": 0.2309,
+      "step": 1424
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.924419948561748e-05,
+      "loss": 0.1593,
+      "step": 1425
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9242891084337983e-05,
+      "loss": 0.082,
+      "step": 1426
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9241581596077848e-05,
+      "loss": 0.2125,
+      "step": 1427
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9240271020991062e-05,
+      "loss": 0.2812,
+      "step": 1428
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9238959359231763e-05,
+      "loss": 0.1827,
+      "step": 1429
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.92376466109542e-05,
+      "loss": 0.3418,
+      "step": 1430
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9236332776312745e-05,
+      "loss": 0.2756,
+      "step": 1431
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9235017855461922e-05,
+      "loss": 0.1568,
+      "step": 1432
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9233701848556358e-05,
+      "loss": 0.1266,
+      "step": 1433
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9232384755750824e-05,
+      "loss": 0.3294,
+      "step": 1434
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923106657720021e-05,
+      "loss": 0.2792,
+      "step": 1435
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9229747313059534e-05,
+      "loss": 0.3072,
+      "step": 1436
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.922842696348395e-05,
+      "loss": 0.2719,
+      "step": 1437
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9227105528628724e-05,
+      "loss": 0.1845,
+      "step": 1438
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.922578300864927e-05,
+      "loss": 0.2762,
+      "step": 1439
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.922445940370111e-05,
+      "loss": 0.2368,
+      "step": 1440
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9223134713939906e-05,
+      "loss": 0.1316,
+      "step": 1441
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9221808939521445e-05,
+      "loss": 0.4091,
+      "step": 1442
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9220482080601637e-05,
+      "loss": 0.1858,
+      "step": 1443
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9219154137336524e-05,
+      "loss": 0.2939,
+      "step": 1444
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9217825109882277e-05,
+      "loss": 0.1666,
+      "step": 1445
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9216494998395192e-05,
+      "loss": 0.314,
+      "step": 1446
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.921516380303169e-05,
+      "loss": 0.2499,
+      "step": 1447
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9213831523948324e-05,
+      "loss": 0.2007,
+      "step": 1448
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.921249816130177e-05,
+      "loss": 0.2009,
+      "step": 1449
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9211163715248836e-05,
+      "loss": 0.2835,
+      "step": 1450
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9209828185946453e-05,
+      "loss": 0.1989,
+      "step": 1451
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9208491573551683e-05,
+      "loss": 0.214,
+      "step": 1452
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.920715387822172e-05,
+      "loss": 0.0847,
+      "step": 1453
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9205815100113866e-05,
+      "loss": 0.2672,
+      "step": 1454
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9204475239385572e-05,
+      "loss": 0.1516,
+      "step": 1455
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9203134296194413e-05,
+      "loss": 0.2207,
+      "step": 1456
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9201792270698077e-05,
+      "loss": 0.1817,
+      "step": 1457
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9200449163054392e-05,
+      "loss": 0.1222,
+      "step": 1458
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.919910497342131e-05,
+      "loss": 0.253,
+      "step": 1459
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9197759701956914e-05,
+      "loss": 0.2189,
+      "step": 1460
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9196413348819403e-05,
+      "loss": 0.191,
+      "step": 1461
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9195065914167115e-05,
+      "loss": 0.1421,
+      "step": 1462
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9193717398158508e-05,
+      "loss": 0.2542,
+      "step": 1463
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9192367800952175e-05,
+      "loss": 0.2626,
+      "step": 1464
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9191017122706827e-05,
+      "loss": 0.1547,
+      "step": 1465
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9189665363581308e-05,
+      "loss": 0.149,
+      "step": 1466
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9188312523734586e-05,
+      "loss": 0.1091,
+      "step": 1467
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9186958603325755e-05,
+      "loss": 0.2462,
+      "step": 1468
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9185603602514044e-05,
+      "loss": 0.3413,
+      "step": 1469
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9184247521458802e-05,
+      "loss": 0.2129,
+      "step": 1470
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9182890360319503e-05,
+      "loss": 0.3162,
+      "step": 1471
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9181532119255754e-05,
+      "loss": 0.319,
+      "step": 1472
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9180172798427286e-05,
+      "loss": 0.1363,
+      "step": 1473
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.917881239799396e-05,
+      "loss": 0.2586,
+      "step": 1474
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.917745091811576e-05,
+      "loss": 0.2862,
+      "step": 1475
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9176088358952798e-05,
+      "loss": 0.3382,
+      "step": 1476
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9174724720665315e-05,
+      "loss": 0.2177,
+      "step": 1477
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9173360003413674e-05,
+      "loss": 0.3068,
+      "step": 1478
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9171994207358367e-05,
+      "loss": 0.1364,
+      "step": 1479
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9170627332660023e-05,
+      "loss": 0.1571,
+      "step": 1480
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.916925937947938e-05,
+      "loss": 0.2141,
+      "step": 1481
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9167890347977316e-05,
+      "loss": 0.3066,
+      "step": 1482
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.916652023831483e-05,
+      "loss": 0.1431,
+      "step": 1483
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.916514905065305e-05,
+      "loss": 0.1269,
+      "step": 1484
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.916377678515323e-05,
+      "loss": 0.3288,
+      "step": 1485
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.916240344197675e-05,
+      "loss": 0.2117,
+      "step": 1486
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.916102902128512e-05,
+      "loss": 0.3893,
+      "step": 1487
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.915965352323997e-05,
+      "loss": 0.1636,
+      "step": 1488
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9158276948003066e-05,
+      "loss": 0.1647,
+      "step": 1489
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9156899295736292e-05,
+      "loss": 0.3167,
+      "step": 1490
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9155520566601668e-05,
+      "loss": 0.1799,
+      "step": 1491
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.915414076076133e-05,
+      "loss": 0.1713,
+      "step": 1492
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9152759878377546e-05,
+      "loss": 0.2054,
+      "step": 1493
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.915137791961271e-05,
+      "loss": 0.2042,
+      "step": 1494
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9149994884629343e-05,
+      "loss": 0.2828,
+      "step": 1495
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9148610773590094e-05,
+      "loss": 0.2048,
+      "step": 1496
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9147225586657737e-05,
+      "loss": 0.1067,
+      "step": 1497
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9145839323995173e-05,
+      "loss": 0.2051,
+      "step": 1498
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9144451985765426e-05,
+      "loss": 0.1862,
+      "step": 1499
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9143063572131652e-05,
+      "loss": 0.1832,
+      "step": 1500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9141674083257127e-05,
+      "loss": 0.2058,
+      "step": 1501
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9140283519305262e-05,
+      "loss": 0.2469,
+      "step": 1502
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913889188043959e-05,
+      "loss": 0.3074,
+      "step": 1503
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9137499166823764e-05,
+      "loss": 0.3983,
+      "step": 1504
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9136105378621578e-05,
+      "loss": 0.2346,
+      "step": 1505
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9134710515996943e-05,
+      "loss": 0.1511,
+      "step": 1506
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9133314579113892e-05,
+      "loss": 0.1233,
+      "step": 1507
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9131917568136587e-05,
+      "loss": 0.2915,
+      "step": 1508
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913051948322933e-05,
+      "loss": 0.1368,
+      "step": 1509
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912912032455653e-05,
+      "loss": 0.2165,
+      "step": 1510
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9127720092282737e-05,
+      "loss": 0.381,
+      "step": 1511
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9126318786572612e-05,
+      "loss": 0.1338,
+      "step": 1512
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9124916407590957e-05,
+      "loss": 0.2474,
+      "step": 1513
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9123512955502696e-05,
+      "loss": 0.2137,
+      "step": 1514
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9122108430472874e-05,
+      "loss": 0.2029,
+      "step": 1515
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9120702832666666e-05,
+      "loss": 0.2653,
+      "step": 1516
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9119296162249372e-05,
+      "loss": 0.2876,
+      "step": 1517
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.911788841938642e-05,
+      "loss": 0.1649,
+      "step": 1518
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9116479604243365e-05,
+      "loss": 0.157,
+      "step": 1519
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9115069716985886e-05,
+      "loss": 0.1465,
+      "step": 1520
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9113658757779786e-05,
+      "loss": 0.1798,
+      "step": 1521
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9112246726790995e-05,
+      "loss": 0.2831,
+      "step": 1522
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9110833624185576e-05,
+      "loss": 0.1643,
+      "step": 1523
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9109419450129706e-05,
+      "loss": 0.2406,
+      "step": 1524
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9108004204789695e-05,
+      "loss": 0.2538,
+      "step": 1525
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9106587888331986e-05,
+      "loss": 0.2377,
+      "step": 1526
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910517050092313e-05,
+      "loss": 0.0962,
+      "step": 1527
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9103752042729824e-05,
+      "loss": 0.3055,
+      "step": 1528
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9102332513918876e-05,
+      "loss": 0.3131,
+      "step": 1529
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9100911914657224e-05,
+      "loss": 0.2834,
+      "step": 1530
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909949024511194e-05,
+      "loss": 0.2819,
+      "step": 1531
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9098067505450203e-05,
+      "loss": 0.1905,
+      "step": 1532
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9096643695839342e-05,
+      "loss": 0.1108,
+      "step": 1533
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909521881644679e-05,
+      "loss": 0.1783,
+      "step": 1534
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909379286744012e-05,
+      "loss": 0.3113,
+      "step": 1535
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909236584898703e-05,
+      "loss": 0.1862,
+      "step": 1536
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9090937761255332e-05,
+      "loss": 0.2951,
+      "step": 1537
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9089508604412973e-05,
+      "loss": 0.1758,
+      "step": 1538
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.908807837862803e-05,
+      "loss": 0.1755,
+      "step": 1539
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9086647084068697e-05,
+      "loss": 0.1984,
+      "step": 1540
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9085214720903297e-05,
+      "loss": 0.2696,
+      "step": 1541
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9083781289300278e-05,
+      "loss": 0.2242,
+      "step": 1542
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9082346789428213e-05,
+      "loss": 0.1566,
+      "step": 1543
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9080911221455804e-05,
+      "loss": 0.2886,
+      "step": 1544
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9079474585551872e-05,
+      "loss": 0.1305,
+      "step": 1545
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.907803688188538e-05,
+      "loss": 0.2469,
+      "step": 1546
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9076598110625387e-05,
+      "loss": 0.216,
+      "step": 1547
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.907515827194111e-05,
+      "loss": 0.2334,
+      "step": 1548
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9073717366001866e-05,
+      "loss": 0.2136,
+      "step": 1549
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9072275392977116e-05,
+      "loss": 0.2292,
+      "step": 1550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9070832353036435e-05,
+      "loss": 0.2705,
+      "step": 1551
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.906938824634953e-05,
+      "loss": 0.1986,
+      "step": 1552
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9067943073086223e-05,
+      "loss": 0.2823,
+      "step": 1553
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9066496833416476e-05,
+      "loss": 0.1043,
+      "step": 1554
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.906504952751037e-05,
+      "loss": 0.2261,
+      "step": 1555
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9063601155538106e-05,
+      "loss": 0.0852,
+      "step": 1556
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9062151717670018e-05,
+      "loss": 0.2044,
+      "step": 1557
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9060701214076557e-05,
+      "loss": 0.2471,
+      "step": 1558
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9059249644928317e-05,
+      "loss": 0.3217,
+      "step": 1559
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9057797010395995e-05,
+      "loss": 0.2904,
+      "step": 1560
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9056343310650428e-05,
+      "loss": 0.1021,
+      "step": 1561
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.905488854586257e-05,
+      "loss": 0.2098,
+      "step": 1562
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9053432716203505e-05,
+      "loss": 0.2554,
+      "step": 1563
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9051975821844445e-05,
+      "loss": 0.2934,
+      "step": 1564
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9050517862956716e-05,
+      "loss": 0.1774,
+      "step": 1565
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9049058839711786e-05,
+      "loss": 0.2235,
+      "step": 1566
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9047598752281235e-05,
+      "loss": 0.2022,
+      "step": 1567
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9046137600836765e-05,
+      "loss": 0.2838,
+      "step": 1568
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.904467538555022e-05,
+      "loss": 0.2526,
+      "step": 1569
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9043212106593554e-05,
+      "loss": 0.1994,
+      "step": 1570
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.904174776413885e-05,
+      "loss": 0.2697,
+      "step": 1571
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9040282358358324e-05,
+      "loss": 0.4012,
+      "step": 1572
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9038815889424303e-05,
+      "loss": 0.19,
+      "step": 1573
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9037348357509252e-05,
+      "loss": 0.4491,
+      "step": 1574
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.903587976278575e-05,
+      "loss": 0.1787,
+      "step": 1575
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.903441010542651e-05,
+      "loss": 0.2187,
+      "step": 1576
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.903293938560437e-05,
+      "loss": 0.2325,
+      "step": 1577
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9031467603492278e-05,
+      "loss": 0.1681,
+      "step": 1578
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9029994759263325e-05,
+      "loss": 0.1643,
+      "step": 1579
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9028520853090724e-05,
+      "loss": 0.2562,
+      "step": 1580
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9027045885147803e-05,
+      "loss": 0.3877,
+      "step": 1581
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9025569855608026e-05,
+      "loss": 0.1911,
+      "step": 1582
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.902409276464497e-05,
+      "loss": 0.1454,
+      "step": 1583
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.902261461243235e-05,
+      "loss": 0.1949,
+      "step": 1584
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9021135399143994e-05,
+      "loss": 0.2717,
+      "step": 1585
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9019655124953864e-05,
+      "loss": 0.1931,
+      "step": 1586
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9018173790036046e-05,
+      "loss": 0.2156,
+      "step": 1587
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9016691394564737e-05,
+      "loss": 0.1682,
+      "step": 1588
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.901520793871428e-05,
+      "loss": 0.2864,
+      "step": 1589
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9013723422659124e-05,
+      "loss": 0.1526,
+      "step": 1590
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9012237846573857e-05,
+      "loss": 0.2217,
+      "step": 1591
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9010751210633184e-05,
+      "loss": 0.1967,
+      "step": 1592
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9009263515011936e-05,
+      "loss": 0.3284,
+      "step": 1593
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9007774759885068e-05,
+      "loss": 0.2344,
+      "step": 1594
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9006284945427656e-05,
+      "loss": 0.1824,
+      "step": 1595
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.900479407181491e-05,
+      "loss": 0.2684,
+      "step": 1596
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.900330213922216e-05,
+      "loss": 0.2889,
+      "step": 1597
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.900180914782486e-05,
+      "loss": 0.2359,
+      "step": 1598
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9000315097798583e-05,
+      "loss": 0.2007,
+      "step": 1599
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8998819989319036e-05,
+      "loss": 0.2379,
+      "step": 1600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8997323822562048e-05,
+      "loss": 0.2615,
+      "step": 1601
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.899582659770357e-05,
+      "loss": 0.1554,
+      "step": 1602
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.899432831491968e-05,
+      "loss": 0.238,
+      "step": 1603
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.899282897438657e-05,
+      "loss": 0.2378,
+      "step": 1604
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8991328576280576e-05,
+      "loss": 0.2303,
+      "step": 1605
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.898982712077814e-05,
+      "loss": 0.3307,
+      "step": 1606
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.898832460805584e-05,
+      "loss": 0.2949,
+      "step": 1607
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8986821038290372e-05,
+      "loss": 0.1521,
+      "step": 1608
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.898531641165856e-05,
+      "loss": 0.2921,
+      "step": 1609
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8983810728337353e-05,
+      "loss": 0.1899,
+      "step": 1610
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8982303988503815e-05,
+      "loss": 0.2143,
+      "step": 1611
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.898079619233515e-05,
+      "loss": 0.208,
+      "step": 1612
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.897928734000867e-05,
+      "loss": 0.2674,
+      "step": 1613
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8977777431701824e-05,
+      "loss": 0.1706,
+      "step": 1614
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8976266467592175e-05,
+      "loss": 0.1362,
+      "step": 1615
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8974754447857423e-05,
+      "loss": 0.2654,
+      "step": 1616
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8973241372675373e-05,
+      "loss": 0.1917,
+      "step": 1617
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8971727242223976e-05,
+      "loss": 0.2,
+      "step": 1618
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.897021205668129e-05,
+      "loss": 0.2222,
+      "step": 1619
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.896869581622551e-05,
+      "loss": 0.2897,
+      "step": 1620
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.896717852103494e-05,
+      "loss": 0.1937,
+      "step": 1621
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.896566017128802e-05,
+      "loss": 0.2557,
+      "step": 1622
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8964140767163313e-05,
+      "loss": 0.2436,
+      "step": 1623
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8962620308839507e-05,
+      "loss": 0.2878,
+      "step": 1624
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8961098796495402e-05,
+      "loss": 0.242,
+      "step": 1625
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8959576230309934e-05,
+      "loss": 0.2904,
+      "step": 1626
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.895805261046216e-05,
+      "loss": 0.2657,
+      "step": 1627
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.895652793713126e-05,
+      "loss": 0.187,
+      "step": 1628
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8955002210496537e-05,
+      "loss": 0.2951,
+      "step": 1629
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.895347543073743e-05,
+      "loss": 0.1905,
+      "step": 1630
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.895194759803347e-05,
+      "loss": 0.2403,
+      "step": 1631
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8950418712564353e-05,
+      "loss": 0.2843,
+      "step": 1632
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8948888774509866e-05,
+      "loss": 0.2434,
+      "step": 1633
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.894735778404994e-05,
+      "loss": 0.2887,
+      "step": 1634
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8945825741364617e-05,
+      "loss": 0.1564,
+      "step": 1635
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8944292646634072e-05,
+      "loss": 0.1768,
+      "step": 1636
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.89427585000386e-05,
+      "loss": 0.2596,
+      "step": 1637
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8941223301758612e-05,
+      "loss": 0.2368,
+      "step": 1638
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.893968705197466e-05,
+      "loss": 0.1647,
+      "step": 1639
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8938149750867404e-05,
+      "loss": 0.1835,
+      "step": 1640
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8936611398617636e-05,
+      "loss": 0.2634,
+      "step": 1641
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8935071995406264e-05,
+      "loss": 0.2078,
+      "step": 1642
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8933531541414328e-05,
+      "loss": 0.2841,
+      "step": 1643
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8931990036822992e-05,
+      "loss": 0.3286,
+      "step": 1644
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8930447481813536e-05,
+      "loss": 0.1522,
+      "step": 1645
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8928903876567366e-05,
+      "loss": 0.1467,
+      "step": 1646
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8927359221266015e-05,
+      "loss": 0.1777,
+      "step": 1647
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8925813516091134e-05,
+      "loss": 0.2074,
+      "step": 1648
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8924266761224506e-05,
+      "loss": 0.2809,
+      "step": 1649
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8922718956848028e-05,
+      "loss": 0.2763,
+      "step": 1650
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8921170103143726e-05,
+      "loss": 0.0915,
+      "step": 1651
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8919620200293746e-05,
+      "loss": 0.3159,
+      "step": 1652
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.891806924848036e-05,
+      "loss": 0.1789,
+      "step": 1653
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8916517247885968e-05,
+      "loss": 0.3104,
+      "step": 1654
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.891496419869308e-05,
+      "loss": 0.1355,
+      "step": 1655
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8913410101084344e-05,
+      "loss": 0.2456,
+      "step": 1656
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8911854955242523e-05,
+      "loss": 0.1344,
+      "step": 1657
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.89102987613505e-05,
+      "loss": 0.3448,
+      "step": 1658
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8908741519591297e-05,
+      "loss": 0.2098,
+      "step": 1659
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8907183230148035e-05,
+      "loss": 0.1523,
+      "step": 1660
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8905623893203978e-05,
+      "loss": 0.1222,
+      "step": 1661
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.890406350894251e-05,
+      "loss": 0.1947,
+      "step": 1662
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8902502077547133e-05,
+      "loss": 0.1568,
+      "step": 1663
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8900939599201472e-05,
+      "loss": 0.2382,
+      "step": 1664
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8899376074089276e-05,
+      "loss": 0.1767,
+      "step": 1665
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8897811502394425e-05,
+      "loss": 0.1949,
+      "step": 1666
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.889624588430091e-05,
+      "loss": 0.2776,
+      "step": 1667
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8894679219992847e-05,
+      "loss": 0.2298,
+      "step": 1668
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.889311150965449e-05,
+      "loss": 0.2832,
+      "step": 1669
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8891542753470195e-05,
+      "loss": 0.3042,
+      "step": 1670
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8889972951624454e-05,
+      "loss": 0.2299,
+      "step": 1671
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8888402104301878e-05,
+      "loss": 0.3395,
+      "step": 1672
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.88868302116872e-05,
+      "loss": 0.2277,
+      "step": 1673
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.888525727396528e-05,
+      "loss": 0.2594,
+      "step": 1674
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.88836832913211e-05,
+      "loss": 0.2661,
+      "step": 1675
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8882108263939754e-05,
+      "loss": 0.1913,
+      "step": 1676
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.888053219200648e-05,
+      "loss": 0.3667,
+      "step": 1677
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.887895507570662e-05,
+      "loss": 0.1759,
+      "step": 1678
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8877376915225646e-05,
+      "loss": 0.1525,
+      "step": 1679
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8875797710749152e-05,
+      "loss": 0.1729,
+      "step": 1680
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.887421746246286e-05,
+      "loss": 0.3038,
+      "step": 1681
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8872636170552606e-05,
+      "loss": 0.2211,
+      "step": 1682
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.887105383520435e-05,
+      "loss": 0.2747,
+      "step": 1683
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.886947045660419e-05,
+      "loss": 0.3284,
+      "step": 1684
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8867886034938318e-05,
+      "loss": 0.2778,
+      "step": 1685
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8866300570393074e-05,
+      "loss": 0.2867,
+      "step": 1686
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.886471406315491e-05,
+      "loss": 0.1729,
+      "step": 1687
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.88631265134104e-05,
+      "loss": 0.4128,
+      "step": 1688
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8861537921346248e-05,
+      "loss": 0.2411,
+      "step": 1689
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.885994828714927e-05,
+      "loss": 0.1094,
+      "step": 1690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.885835761100641e-05,
+      "loss": 0.2582,
+      "step": 1691
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8856765893104736e-05,
+      "loss": 0.3346,
+      "step": 1692
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.885517313363144e-05,
+      "loss": 0.2822,
+      "step": 1693
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8853579332773825e-05,
+      "loss": 0.2713,
+      "step": 1694
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8851984490719334e-05,
+      "loss": 0.2561,
+      "step": 1695
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.885038860765552e-05,
+      "loss": 0.2509,
+      "step": 1696
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.884879168377006e-05,
+      "loss": 0.2507,
+      "step": 1697
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.884719371925076e-05,
+      "loss": 0.215,
+      "step": 1698
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8845594714285535e-05,
+      "loss": 0.3245,
+      "step": 1699
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8843994669062436e-05,
+      "loss": 0.226,
+      "step": 1700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.884239358376963e-05,
+      "loss": 0.1931,
+      "step": 1701
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8840791458595414e-05,
+      "loss": 0.0939,
+      "step": 1702
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.883918829372819e-05,
+      "loss": 0.2322,
+      "step": 1703
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8837584089356498e-05,
+      "loss": 0.1199,
+      "step": 1704
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8835978845668998e-05,
+      "loss": 0.3347,
+      "step": 1705
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8834372562854465e-05,
+      "loss": 0.1551,
+      "step": 1706
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8832765241101807e-05,
+      "loss": 0.1799,
+      "step": 1707
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.883115688060004e-05,
+      "loss": 0.2437,
+      "step": 1708
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8829547481538314e-05,
+      "loss": 0.1061,
+      "step": 1709
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.88279370441059e-05,
+      "loss": 0.1334,
+      "step": 1710
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.882632556849218e-05,
+      "loss": 0.2521,
+      "step": 1711
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8824713054886676e-05,
+      "loss": 0.3507,
+      "step": 1712
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.882309950347902e-05,
+      "loss": 0.1849,
+      "step": 1713
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8821484914458964e-05,
+      "loss": 0.1491,
+      "step": 1714
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8819869288016387e-05,
+      "loss": 0.2472,
+      "step": 1715
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.88182526243413e-05,
+      "loss": 0.2569,
+      "step": 1716
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.881663492362381e-05,
+      "loss": 0.1564,
+      "step": 1717
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8815016186054174e-05,
+      "loss": 0.2951,
+      "step": 1718
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.881339641182275e-05,
+      "loss": 0.1771,
+      "step": 1719
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.881177560112003e-05,
+      "loss": 0.1853,
+      "step": 1720
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.881015375413663e-05,
+      "loss": 0.301,
+      "step": 1721
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8808530871063272e-05,
+      "loss": 0.2594,
+      "step": 1722
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8806906952090814e-05,
+      "loss": 0.1709,
+      "step": 1723
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8805281997410232e-05,
+      "loss": 0.1418,
+      "step": 1724
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.880365600721263e-05,
+      "loss": 0.2315,
+      "step": 1725
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8802028981689214e-05,
+      "loss": 0.1543,
+      "step": 1726
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8800400921031335e-05,
+      "loss": 0.1459,
+      "step": 1727
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.879877182543045e-05,
+      "loss": 0.19,
+      "step": 1728
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.879714169507815e-05,
+      "loss": 0.2072,
+      "step": 1729
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8795510530166135e-05,
+      "loss": 0.3728,
+      "step": 1730
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.879387833088624e-05,
+      "loss": 0.1944,
+      "step": 1731
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.879224509743041e-05,
+      "loss": 0.3615,
+      "step": 1732
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8790610829990715e-05,
+      "loss": 0.1924,
+      "step": 1733
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.878897552875935e-05,
+      "loss": 0.2361,
+      "step": 1734
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.878733919392863e-05,
+      "loss": 0.2021,
+      "step": 1735
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8785701825690988e-05,
+      "loss": 0.2072,
+      "step": 1736
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8784063424238983e-05,
+      "loss": 0.322,
+      "step": 1737
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8782423989765295e-05,
+      "loss": 0.1845,
+      "step": 1738
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8780783522462724e-05,
+      "loss": 0.3244,
+      "step": 1739
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.877914202252419e-05,
+      "loss": 0.2256,
+      "step": 1740
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8777499490142742e-05,
+      "loss": 0.162,
+      "step": 1741
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8775855925511537e-05,
+      "loss": 0.26,
+      "step": 1742
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8774211328823866e-05,
+      "loss": 0.2309,
+      "step": 1743
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8772565700273137e-05,
+      "loss": 0.211,
+      "step": 1744
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8770919040052873e-05,
+      "loss": 0.1671,
+      "step": 1745
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8769271348356734e-05,
+      "loss": 0.2785,
+      "step": 1746
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.876762262537848e-05,
+      "loss": 0.2299,
+      "step": 1747
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.876597287131201e-05,
+      "loss": 0.2601,
+      "step": 1748
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8764322086351342e-05,
+      "loss": 0.2133,
+      "step": 1749
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8762670270690605e-05,
+      "loss": 0.1704,
+      "step": 1750
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.876101742452406e-05,
+      "loss": 0.2721,
+      "step": 1751
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.875936354804608e-05,
+      "loss": 0.1719,
+      "step": 1752
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8757708641451166e-05,
+      "loss": 0.1677,
+      "step": 1753
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8756052704933938e-05,
+      "loss": 0.2157,
+      "step": 1754
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.875439573868914e-05,
+      "loss": 0.1783,
+      "step": 1755
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8752737742911628e-05,
+      "loss": 0.244,
+      "step": 1756
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.875107871779639e-05,
+      "loss": 0.2698,
+      "step": 1757
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.874941866353853e-05,
+      "loss": 0.3474,
+      "step": 1758
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8747757580333273e-05,
+      "loss": 0.2528,
+      "step": 1759
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.874609546837596e-05,
+      "loss": 0.1518,
+      "step": 1760
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.874443232786207e-05,
+      "loss": 0.1469,
+      "step": 1761
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.874276815898718e-05,
+      "loss": 0.2212,
+      "step": 1762
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8741102961947005e-05,
+      "loss": 0.276,
+      "step": 1763
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8739436736937374e-05,
+      "loss": 0.3411,
+      "step": 1764
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.873776948415424e-05,
+      "loss": 0.3201,
+      "step": 1765
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8736101203793672e-05,
+      "loss": 0.3237,
+      "step": 1766
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8734431896051865e-05,
+      "loss": 0.2722,
+      "step": 1767
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8732761561125126e-05,
+      "loss": 0.2104,
+      "step": 1768
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8731090199209898e-05,
+      "loss": 0.1891,
+      "step": 1769
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8729417810502734e-05,
+      "loss": 0.2903,
+      "step": 1770
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.872774439520031e-05,
+      "loss": 0.2212,
+      "step": 1771
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8726069953499417e-05,
+      "loss": 0.313,
+      "step": 1772
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.872439448559698e-05,
+      "loss": 0.1781,
+      "step": 1773
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8722717991690033e-05,
+      "loss": 0.1735,
+      "step": 1774
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8721040471975732e-05,
+      "loss": 0.1983,
+      "step": 1775
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8719361926651367e-05,
+      "loss": 0.2261,
+      "step": 1776
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8717682355914326e-05,
+      "loss": 0.1883,
+      "step": 1777
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8716001759962136e-05,
+      "loss": 0.1723,
+      "step": 1778
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8714320138992438e-05,
+      "loss": 0.3138,
+      "step": 1779
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.871263749320299e-05,
+      "loss": 0.1958,
+      "step": 1780
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8710953822791676e-05,
+      "loss": 0.1991,
+      "step": 1781
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8709269127956503e-05,
+      "loss": 0.2423,
+      "step": 1782
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8707583408895584e-05,
+      "loss": 0.2367,
+      "step": 1783
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8705896665807174e-05,
+      "loss": 0.2328,
+      "step": 1784
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8704208898889627e-05,
+      "loss": 0.152,
+      "step": 1785
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8702520108341437e-05,
+      "loss": 0.1998,
+      "step": 1786
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.87008302943612e-05,
+      "loss": 0.2625,
+      "step": 1787
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8699139457147648e-05,
+      "loss": 0.298,
+      "step": 1788
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8697447596899625e-05,
+      "loss": 0.1762,
+      "step": 1789
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8695754713816094e-05,
+      "loss": 0.1775,
+      "step": 1790
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8694060808096142e-05,
+      "loss": 0.1747,
+      "step": 1791
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8692365879938976e-05,
+      "loss": 0.1772,
+      "step": 1792
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8690669929543923e-05,
+      "loss": 0.2504,
+      "step": 1793
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.868897295711043e-05,
+      "loss": 0.2975,
+      "step": 1794
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8687274962838064e-05,
+      "loss": 0.2965,
+      "step": 1795
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.868557594692651e-05,
+      "loss": 0.1508,
+      "step": 1796
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8683875909575574e-05,
+      "loss": 0.1433,
+      "step": 1797
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8682174850985194e-05,
+      "loss": 0.1954,
+      "step": 1798
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8680472771355404e-05,
+      "loss": 0.2653,
+      "step": 1799
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.867876967088638e-05,
+      "loss": 0.2274,
+      "step": 1800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.867706554977841e-05,
+      "loss": 0.3312,
+      "step": 1801
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8675360408231894e-05,
+      "loss": 0.2156,
+      "step": 1802
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8673654246447365e-05,
+      "loss": 0.2173,
+      "step": 1803
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.867194706462547e-05,
+      "loss": 0.2768,
+      "step": 1804
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8670238862966982e-05,
+      "loss": 0.2373,
+      "step": 1805
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.866852964167278e-05,
+      "loss": 0.1864,
+      "step": 1806
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8666819400943877e-05,
+      "loss": 0.1394,
+      "step": 1807
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8665108140981397e-05,
+      "loss": 0.2081,
+      "step": 1808
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.866339586198659e-05,
+      "loss": 0.2135,
+      "step": 1809
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8661682564160818e-05,
+      "loss": 0.2899,
+      "step": 1810
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8659968247705577e-05,
+      "loss": 0.2852,
+      "step": 1811
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8658252912822466e-05,
+      "loss": 0.3317,
+      "step": 1812
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8656536559713214e-05,
+      "loss": 0.2043,
+      "step": 1813
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.865481918857967e-05,
+      "loss": 0.1514,
+      "step": 1814
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8653100799623794e-05,
+      "loss": 0.1011,
+      "step": 1815
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.865138139304768e-05,
+      "loss": 0.1644,
+      "step": 1816
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.864966096905352e-05,
+      "loss": 0.194,
+      "step": 1817
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.864793952784365e-05,
+      "loss": 0.3367,
+      "step": 1818
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8646217069620512e-05,
+      "loss": 0.0921,
+      "step": 1819
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.864449359458667e-05,
+      "loss": 0.1428,
+      "step": 1820
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8642769102944812e-05,
+      "loss": 0.2963,
+      "step": 1821
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.864104359489773e-05,
+      "loss": 0.163,
+      "step": 1822
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8639317070648358e-05,
+      "loss": 0.2945,
+      "step": 1823
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8637589530399735e-05,
+      "loss": 0.1998,
+      "step": 1824
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.863586097435502e-05,
+      "loss": 0.2589,
+      "step": 1825
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8634131402717495e-05,
+      "loss": 0.2396,
+      "step": 1826
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8632400815690565e-05,
+      "loss": 0.2426,
+      "step": 1827
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8630669213477747e-05,
+      "loss": 0.2813,
+      "step": 1828
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.862893659628268e-05,
+      "loss": 0.285,
+      "step": 1829
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.862720296430912e-05,
+      "loss": 0.1173,
+      "step": 1830
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8625468317760953e-05,
+      "loss": 0.2383,
+      "step": 1831
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8623732656842172e-05,
+      "loss": 0.1785,
+      "step": 1832
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.862199598175689e-05,
+      "loss": 0.2496,
+      "step": 1833
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8620258292709352e-05,
+      "loss": 0.2855,
+      "step": 1834
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.861851958990391e-05,
+      "loss": 0.1903,
+      "step": 1835
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8616779873545035e-05,
+      "loss": 0.17,
+      "step": 1836
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8615039143837325e-05,
+      "loss": 0.2519,
+      "step": 1837
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8613297400985487e-05,
+      "loss": 0.3833,
+      "step": 1838
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.861155464519436e-05,
+      "loss": 0.2679,
+      "step": 1839
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8609810876668892e-05,
+      "loss": 0.3379,
+      "step": 1840
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8608066095614156e-05,
+      "loss": 0.1552,
+      "step": 1841
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8606320302235335e-05,
+      "loss": 0.2583,
+      "step": 1842
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8604573496737743e-05,
+      "loss": 0.1433,
+      "step": 1843
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.860282567932681e-05,
+      "loss": 0.3246,
+      "step": 1844
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8601076850208072e-05,
+      "loss": 0.2452,
+      "step": 1845
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8599327009587204e-05,
+      "loss": 0.2605,
+      "step": 1846
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8597576157669986e-05,
+      "loss": 0.3274,
+      "step": 1847
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8595824294662327e-05,
+      "loss": 0.2001,
+      "step": 1848
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.859407142077024e-05,
+      "loss": 0.3322,
+      "step": 1849
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8592317536199877e-05,
+      "loss": 0.1803,
+      "step": 1850
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8590562641157487e-05,
+      "loss": 0.1471,
+      "step": 1851
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8588806735849455e-05,
+      "loss": 0.1199,
+      "step": 1852
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8587049820482277e-05,
+      "loss": 0.1629,
+      "step": 1853
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.858529189526257e-05,
+      "loss": 0.2188,
+      "step": 1854
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8583532960397072e-05,
+      "loss": 0.1808,
+      "step": 1855
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.858177301609263e-05,
+      "loss": 0.2806,
+      "step": 1856
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8580012062556224e-05,
+      "loss": 0.2824,
+      "step": 1857
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8578250099994938e-05,
+      "loss": 0.2944,
+      "step": 1858
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.857648712861599e-05,
+      "loss": 0.1727,
+      "step": 1859
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8574723148626703e-05,
+      "loss": 0.3105,
+      "step": 1860
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8572958160234525e-05,
+      "loss": 0.2051,
+      "step": 1861
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8571192163647024e-05,
+      "loss": 0.2912,
+      "step": 1862
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8569425159071883e-05,
+      "loss": 0.1787,
+      "step": 1863
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8567657146716905e-05,
+      "loss": 0.1755,
+      "step": 1864
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.856588812679001e-05,
+      "loss": 0.2148,
+      "step": 1865
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.856411809949924e-05,
+      "loss": 0.2812,
+      "step": 1866
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.856234706505275e-05,
+      "loss": 0.1967,
+      "step": 1867
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.856057502365882e-05,
+      "loss": 0.1558,
+      "step": 1868
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8558801975525852e-05,
+      "loss": 0.2483,
+      "step": 1869
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8557027920862344e-05,
+      "loss": 0.2592,
+      "step": 1870
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8555252859876942e-05,
+      "loss": 0.2629,
+      "step": 1871
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8553476792778385e-05,
+      "loss": 0.2562,
+      "step": 1872
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8551699719775552e-05,
+      "loss": 0.1992,
+      "step": 1873
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8549921641077426e-05,
+      "loss": 0.2186,
+      "step": 1874
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8548142556893108e-05,
+      "loss": 0.2461,
+      "step": 1875
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.854636246743183e-05,
+      "loss": 0.136,
+      "step": 1876
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8544581372902924e-05,
+      "loss": 0.1922,
+      "step": 1877
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8542799273515858e-05,
+      "loss": 0.192,
+      "step": 1878
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8541016169480205e-05,
+      "loss": 0.1706,
+      "step": 1879
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8539232061005665e-05,
+      "loss": 0.1609,
+      "step": 1880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.853744694830205e-05,
+      "loss": 0.1716,
+      "step": 1881
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8535660831579292e-05,
+      "loss": 0.216,
+      "step": 1882
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8533873711047444e-05,
+      "loss": 0.3389,
+      "step": 1883
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8532085586916672e-05,
+      "loss": 0.204,
+      "step": 1884
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8530296459397267e-05,
+      "loss": 0.4067,
+      "step": 1885
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8528506328699626e-05,
+      "loss": 0.2446,
+      "step": 1886
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.852671519503428e-05,
+      "loss": 0.3089,
+      "step": 1887
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.852492305861186e-05,
+      "loss": 0.2159,
+      "step": 1888
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8523129919643134e-05,
+      "loss": 0.2777,
+      "step": 1889
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8521335778338976e-05,
+      "loss": 0.0902,
+      "step": 1890
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8519540634910375e-05,
+      "loss": 0.194,
+      "step": 1891
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.851774448956845e-05,
+      "loss": 0.112,
+      "step": 1892
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8515947342524425e-05,
+      "loss": 0.193,
+      "step": 1893
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8514149193989654e-05,
+      "loss": 0.2418,
+      "step": 1894
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8512350044175596e-05,
+      "loss": 0.1016,
+      "step": 1895
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8510549893293837e-05,
+      "loss": 0.1744,
+      "step": 1896
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.850874874155608e-05,
+      "loss": 0.1653,
+      "step": 1897
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8506946589174142e-05,
+      "loss": 0.1297,
+      "step": 1898
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8505143436359957e-05,
+      "loss": 0.1555,
+      "step": 1899
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8503339283325585e-05,
+      "loss": 0.1329,
+      "step": 1900
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.850153413028319e-05,
+      "loss": 0.1371,
+      "step": 1901
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8499727977445066e-05,
+      "loss": 0.1862,
+      "step": 1902
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8497920825023622e-05,
+      "loss": 0.2633,
+      "step": 1903
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8496112673231375e-05,
+      "loss": 0.3586,
+      "step": 1904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8494303522280974e-05,
+      "loss": 0.2513,
+      "step": 1905
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8492493372385178e-05,
+      "loss": 0.3168,
+      "step": 1906
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8490682223756855e-05,
+      "loss": 0.0725,
+      "step": 1907
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.848887007660901e-05,
+      "loss": 0.139,
+      "step": 1908
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.848705693115475e-05,
+      "loss": 0.2051,
+      "step": 1909
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8485242787607306e-05,
+      "loss": 0.2172,
+      "step": 1910
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8483427646180023e-05,
+      "loss": 0.1716,
+      "step": 1911
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8481611507086367e-05,
+      "loss": 0.237,
+      "step": 1912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8479794370539915e-05,
+      "loss": 0.1711,
+      "step": 1913
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8477976236754368e-05,
+      "loss": 0.1887,
+      "step": 1914
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8476157105943544e-05,
+      "loss": 0.1351,
+      "step": 1915
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.847433697832137e-05,
+      "loss": 0.0897,
+      "step": 1916
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8472515854101902e-05,
+      "loss": 0.203,
+      "step": 1917
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8470693733499307e-05,
+      "loss": 0.2131,
+      "step": 1918
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8468870616727867e-05,
+      "loss": 0.1638,
+      "step": 1919
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8467046504001985e-05,
+      "loss": 0.1451,
+      "step": 1920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8465221395536185e-05,
+      "loss": 0.1829,
+      "step": 1921
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.846339529154509e-05,
+      "loss": 0.2458,
+      "step": 1922
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8461568192243467e-05,
+      "loss": 0.1917,
+      "step": 1923
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.845974009784618e-05,
+      "loss": 0.2585,
+      "step": 1924
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8457911008568218e-05,
+      "loss": 0.1831,
+      "step": 1925
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.845608092462468e-05,
+      "loss": 0.1131,
+      "step": 1926
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8454249846230794e-05,
+      "loss": 0.1655,
+      "step": 1927
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8452417773601895e-05,
+      "loss": 0.1841,
+      "step": 1928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8450584706953436e-05,
+      "loss": 0.119,
+      "step": 1929
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8448750646500992e-05,
+      "loss": 0.1868,
+      "step": 1930
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8446915592460253e-05,
+      "loss": 0.2091,
+      "step": 1931
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.844507954504702e-05,
+      "loss": 0.1373,
+      "step": 1932
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.844324250447722e-05,
+      "loss": 0.1537,
+      "step": 1933
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8441404470966887e-05,
+      "loss": 0.1863,
+      "step": 1934
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8439565444732185e-05,
+      "loss": 0.2371,
+      "step": 1935
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.843772542598938e-05,
+      "loss": 0.1078,
+      "step": 1936
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8435884414954865e-05,
+      "loss": 0.2168,
+      "step": 1937
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8434042411845144e-05,
+      "loss": 0.223,
+      "step": 1938
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.843219941687684e-05,
+      "loss": 0.1702,
+      "step": 1939
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8430355430266692e-05,
+      "loss": 0.1849,
+      "step": 1940
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.842851045223156e-05,
+      "loss": 0.214,
+      "step": 1941
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8426664482988414e-05,
+      "loss": 0.1065,
+      "step": 1942
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8424817522754342e-05,
+      "loss": 0.1827,
+      "step": 1943
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8422969571746553e-05,
+      "loss": 0.1757,
+      "step": 1944
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.842112063018237e-05,
+      "loss": 0.1571,
+      "step": 1945
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8419270698279226e-05,
+      "loss": 0.2437,
+      "step": 1946
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.841741977625468e-05,
+      "loss": 0.0897,
+      "step": 1947
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8415567864326406e-05,
+      "loss": 0.1741,
+      "step": 1948
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.841371496271219e-05,
+      "loss": 0.136,
+      "step": 1949
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8411861071629936e-05,
+      "loss": 0.16,
+      "step": 1950
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8410006191297666e-05,
+      "loss": 0.2035,
+      "step": 1951
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8408150321933514e-05,
+      "loss": 0.2677,
+      "step": 1952
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.840629346375574e-05,
+      "loss": 0.2014,
+      "step": 1953
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8404435616982707e-05,
+      "loss": 0.173,
+      "step": 1954
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8402576781832905e-05,
+      "loss": 0.1804,
+      "step": 1955
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8400716958524935e-05,
+      "loss": 0.1551,
+      "step": 1956
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.839885614727752e-05,
+      "loss": 0.1795,
+      "step": 1957
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8396994348309486e-05,
+      "loss": 0.2249,
+      "step": 1958
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.839513156183979e-05,
+      "loss": 0.0991,
+      "step": 1959
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.83932677880875e-05,
+      "loss": 0.2029,
+      "step": 1960
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8391403027271794e-05,
+      "loss": 0.1166,
+      "step": 1961
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8389537279611978e-05,
+      "loss": 0.1378,
+      "step": 1962
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8387670545327463e-05,
+      "loss": 0.1667,
+      "step": 1963
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8385802824637777e-05,
+      "loss": 0.1506,
+      "step": 1964
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8383934117762576e-05,
+      "loss": 0.1844,
+      "step": 1965
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8382064424921616e-05,
+      "loss": 0.2588,
+      "step": 1966
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.838019374633478e-05,
+      "loss": 0.1866,
+      "step": 1967
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.837832208222206e-05,
+      "loss": 0.1875,
+      "step": 1968
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8376449432803573e-05,
+      "loss": 0.2236,
+      "step": 1969
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8374575798299537e-05,
+      "loss": 0.1946,
+      "step": 1970
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8372701178930303e-05,
+      "loss": 0.1753,
+      "step": 1971
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8370825574916324e-05,
+      "loss": 0.2689,
+      "step": 1972
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8368948986478176e-05,
+      "loss": 0.219,
+      "step": 1973
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8367071413836554e-05,
+      "loss": 0.285,
+      "step": 1974
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.836519285721226e-05,
+      "loss": 0.2227,
+      "step": 1975
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.836331331682621e-05,
+      "loss": 0.1803,
+      "step": 1976
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8361432792899452e-05,
+      "loss": 0.1308,
+      "step": 1977
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.835955128565313e-05,
+      "loss": 0.1952,
+      "step": 1978
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.835766879530852e-05,
+      "loss": 0.203,
+      "step": 1979
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8355785322087e-05,
+      "loss": 0.25,
+      "step": 1980
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8353900866210075e-05,
+      "loss": 0.2088,
+      "step": 1981
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.835201542789936e-05,
+      "loss": 0.1578,
+      "step": 1982
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8350129007376578e-05,
+      "loss": 0.0832,
+      "step": 1983
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8348241604863585e-05,
+      "loss": 0.147,
+      "step": 1984
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8346353220582337e-05,
+      "loss": 0.2001,
+      "step": 1985
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8344463854754915e-05,
+      "loss": 0.2965,
+      "step": 1986
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.834257350760351e-05,
+      "loss": 0.2228,
+      "step": 1987
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8340682179350434e-05,
+      "loss": 0.1268,
+      "step": 1988
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8338789870218107e-05,
+      "loss": 0.1941,
+      "step": 1989
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8336896580429062e-05,
+      "loss": 0.2626,
+      "step": 1990
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8335002310205964e-05,
+      "loss": 0.1502,
+      "step": 1991
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8333107059771576e-05,
+      "loss": 0.1318,
+      "step": 1992
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8331210829348792e-05,
+      "loss": 0.1502,
+      "step": 1993
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.83293136191606e-05,
+      "loss": 0.2946,
+      "step": 1994
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.832741542943012e-05,
+      "loss": 0.1438,
+      "step": 1995
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8325516260380584e-05,
+      "loss": 0.0573,
+      "step": 1996
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8323616112235337e-05,
+      "loss": 0.181,
+      "step": 1997
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.832171498521784e-05,
+      "loss": 0.2141,
+      "step": 1998
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8319812879551667e-05,
+      "loss": 0.2217,
+      "step": 1999
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8317909795460514e-05,
+      "loss": 0.2678,
+      "step": 2000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.831600573316818e-05,
+      "loss": 0.2364,
+      "step": 2001
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8314100692898594e-05,
+      "loss": 0.2072,
+      "step": 2002
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8312194674875785e-05,
+      "loss": 0.1116,
+      "step": 2003
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8310287679323904e-05,
+      "loss": 0.1449,
+      "step": 2004
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8308379706467225e-05,
+      "loss": 0.1402,
+      "step": 2005
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8306470756530126e-05,
+      "loss": 0.1407,
+      "step": 2006
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.83045608297371e-05,
+      "loss": 0.1359,
+      "step": 2007
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8302649926312755e-05,
+      "loss": 0.1723,
+      "step": 2008
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8300738046481824e-05,
+      "loss": 0.1796,
+      "step": 2009
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8298825190469145e-05,
+      "loss": 0.2626,
+      "step": 2010
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8296911358499672e-05,
+      "loss": 0.187,
+      "step": 2011
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8294996550798477e-05,
+      "loss": 0.1653,
+      "step": 2012
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.829308076759074e-05,
+      "loss": 0.0929,
+      "step": 2013
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8291164009101766e-05,
+      "loss": 0.1831,
+      "step": 2014
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8289246275556972e-05,
+      "loss": 0.1509,
+      "step": 2015
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8287327567181878e-05,
+      "loss": 0.0974,
+      "step": 2016
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.828540788420213e-05,
+      "loss": 0.2032,
+      "step": 2017
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.828348722684349e-05,
+      "loss": 0.1738,
+      "step": 2018
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.828156559533183e-05,
+      "loss": 0.1415,
+      "step": 2019
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8279642989893136e-05,
+      "loss": 0.1919,
+      "step": 2020
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8277719410753508e-05,
+      "loss": 0.0944,
+      "step": 2021
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8275794858139168e-05,
+      "loss": 0.2332,
+      "step": 2022
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.827386933227644e-05,
+      "loss": 0.2643,
+      "step": 2023
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8271942833391775e-05,
+      "loss": 0.1972,
+      "step": 2024
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.827001536171173e-05,
+      "loss": 0.1653,
+      "step": 2025
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8268086917462975e-05,
+      "loss": 0.1807,
+      "step": 2026
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8266157500872307e-05,
+      "loss": 0.1321,
+      "step": 2027
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8264227112166625e-05,
+      "loss": 0.2042,
+      "step": 2028
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8262295751572943e-05,
+      "loss": 0.0952,
+      "step": 2029
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.82603634193184e-05,
+      "loss": 0.2264,
+      "step": 2030
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8258430115630236e-05,
+      "loss": 0.1602,
+      "step": 2031
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.825649584073581e-05,
+      "loss": 0.1703,
+      "step": 2032
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.82545605948626e-05,
+      "loss": 0.2702,
+      "step": 2033
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8252624378238193e-05,
+      "loss": 0.2033,
+      "step": 2034
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.825068719109029e-05,
+      "loss": 0.1481,
+      "step": 2035
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.824874903364671e-05,
+      "loss": 0.1654,
+      "step": 2036
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8246809906135387e-05,
+      "loss": 0.1863,
+      "step": 2037
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8244869808784355e-05,
+      "loss": 0.0803,
+      "step": 2038
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8242928741821783e-05,
+      "loss": 0.1023,
+      "step": 2039
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8240986705475943e-05,
+      "loss": 0.1517,
+      "step": 2040
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8239043699975214e-05,
+      "loss": 0.1767,
+      "step": 2041
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8237099725548107e-05,
+      "loss": 0.1223,
+      "step": 2042
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.823515478242323e-05,
+      "loss": 0.1413,
+      "step": 2043
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8233208870829317e-05,
+      "loss": 0.143,
+      "step": 2044
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.823126199099521e-05,
+      "loss": 0.1716,
+      "step": 2045
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.822931414314986e-05,
+      "loss": 0.2502,
+      "step": 2046
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.822736532752234e-05,
+      "loss": 0.0807,
+      "step": 2047
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.822541554434184e-05,
+      "loss": 0.269,
+      "step": 2048
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.822346479383765e-05,
+      "loss": 0.2369,
+      "step": 2049
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8221513076239183e-05,
+      "loss": 0.2597,
+      "step": 2050
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.821956039177597e-05,
+      "loss": 0.2655,
+      "step": 2051
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8217606740677647e-05,
+      "loss": 0.176,
+      "step": 2052
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8215652123173964e-05,
+      "loss": 0.1581,
+      "step": 2053
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.821369653949479e-05,
+      "loss": 0.2742,
+      "step": 2054
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.821173998987011e-05,
+      "loss": 0.1493,
+      "step": 2055
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8209782474530013e-05,
+      "loss": 0.1479,
+      "step": 2056
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.82078239937047e-05,
+      "loss": 0.1768,
+      "step": 2057
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8205864547624503e-05,
+      "loss": 0.1859,
+      "step": 2058
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.820390413651985e-05,
+      "loss": 0.221,
+      "step": 2059
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8201942760621293e-05,
+      "loss": 0.1693,
+      "step": 2060
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.819998042015949e-05,
+      "loss": 0.176,
+      "step": 2061
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8198017115365217e-05,
+      "loss": 0.1607,
+      "step": 2062
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.819605284646936e-05,
+      "loss": 0.1197,
+      "step": 2063
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8194087613702923e-05,
+      "loss": 0.1551,
+      "step": 2064
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.819212141729702e-05,
+      "loss": 0.2326,
+      "step": 2065
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8190154257482886e-05,
+      "loss": 0.2267,
+      "step": 2066
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.818818613449185e-05,
+      "loss": 0.1671,
+      "step": 2067
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8186217048555376e-05,
+      "loss": 0.1268,
+      "step": 2068
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8184246999905024e-05,
+      "loss": 0.2321,
+      "step": 2069
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8182275988772485e-05,
+      "loss": 0.2777,
+      "step": 2070
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8180304015389545e-05,
+      "loss": 0.1855,
+      "step": 2071
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.817833107998812e-05,
+      "loss": 0.1632,
+      "step": 2072
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8176357182800223e-05,
+      "loss": 0.1946,
+      "step": 2073
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8174382324057994e-05,
+      "loss": 0.1594,
+      "step": 2074
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8172406503993672e-05,
+      "loss": 0.175,
+      "step": 2075
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8170429722839625e-05,
+      "loss": 0.1399,
+      "step": 2076
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8168451980828323e-05,
+      "loss": 0.1075,
+      "step": 2077
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8166473278192355e-05,
+      "loss": 0.1964,
+      "step": 2078
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8164493615164413e-05,
+      "loss": 0.2056,
+      "step": 2079
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8162512991977314e-05,
+      "loss": 0.172,
+      "step": 2080
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.816053140886398e-05,
+      "loss": 0.2924,
+      "step": 2081
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8158548866057453e-05,
+      "loss": 0.2017,
+      "step": 2082
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8156565363790882e-05,
+      "loss": 0.2173,
+      "step": 2083
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8154580902297526e-05,
+      "loss": 0.149,
+      "step": 2084
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8152595481810768e-05,
+      "loss": 0.127,
+      "step": 2085
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8150609102564096e-05,
+      "loss": 0.1,
+      "step": 2086
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8148621764791105e-05,
+      "loss": 0.1971,
+      "step": 2087
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8146633468725512e-05,
+      "loss": 0.149,
+      "step": 2088
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.814464421460115e-05,
+      "loss": 0.1086,
+      "step": 2089
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8142654002651958e-05,
+      "loss": 0.0939,
+      "step": 2090
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.814066283311198e-05,
+      "loss": 0.1701,
+      "step": 2091
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.813867070621539e-05,
+      "loss": 0.1988,
+      "step": 2092
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8136677622196463e-05,
+      "loss": 0.206,
+      "step": 2093
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.813468358128959e-05,
+      "loss": 0.2649,
+      "step": 2094
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.813268858372927e-05,
+      "loss": 0.16,
+      "step": 2095
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.813069262975012e-05,
+      "loss": 0.1302,
+      "step": 2096
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.812869571958687e-05,
+      "loss": 0.225,
+      "step": 2097
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.812669785347436e-05,
+      "loss": 0.1604,
+      "step": 2098
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8124699031647543e-05,
+      "loss": 0.1136,
+      "step": 2099
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.812269925434148e-05,
+      "loss": 0.3018,
+      "step": 2100
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8120698521791352e-05,
+      "loss": 0.257,
+      "step": 2101
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8118696834232445e-05,
+      "loss": 0.2902,
+      "step": 2102
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.811669419190017e-05,
+      "loss": 0.1653,
+      "step": 2103
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8114690595030028e-05,
+      "loss": 0.2025,
+      "step": 2104
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.811268604385766e-05,
+      "loss": 0.1768,
+      "step": 2105
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8110680538618793e-05,
+      "loss": 0.2385,
+      "step": 2106
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8108674079549283e-05,
+      "loss": 0.1503,
+      "step": 2107
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8106666666885093e-05,
+      "loss": 0.2407,
+      "step": 2108
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8104658300862303e-05,
+      "loss": 0.1855,
+      "step": 2109
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.810264898171709e-05,
+      "loss": 0.1306,
+      "step": 2110
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.810063870968576e-05,
+      "loss": 0.1764,
+      "step": 2111
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8098627485004727e-05,
+      "loss": 0.2033,
+      "step": 2112
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.809661530791051e-05,
+      "loss": 0.1845,
+      "step": 2113
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8094602178639743e-05,
+      "loss": 0.2223,
+      "step": 2114
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.809258809742918e-05,
+      "loss": 0.2363,
+      "step": 2115
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8090573064515676e-05,
+      "loss": 0.1788,
+      "step": 2116
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8088557080136206e-05,
+      "loss": 0.2267,
+      "step": 2117
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8086540144527853e-05,
+      "loss": 0.2372,
+      "step": 2118
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8084522257927807e-05,
+      "loss": 0.2119,
+      "step": 2119
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.808250342057338e-05,
+      "loss": 0.1831,
+      "step": 2120
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8080483632701988e-05,
+      "loss": 0.2038,
+      "step": 2121
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.807846289455116e-05,
+      "loss": 0.1497,
+      "step": 2122
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8076441206358548e-05,
+      "loss": 0.1681,
+      "step": 2123
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8074418568361896e-05,
+      "loss": 0.1599,
+      "step": 2124
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8072394980799077e-05,
+      "loss": 0.1468,
+      "step": 2125
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8070370443908064e-05,
+      "loss": 0.1773,
+      "step": 2126
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8068344957926946e-05,
+      "loss": 0.1754,
+      "step": 2127
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8066318523093927e-05,
+      "loss": 0.0886,
+      "step": 2128
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8064291139647318e-05,
+      "loss": 0.2006,
+      "step": 2129
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8062262807825538e-05,
+      "loss": 0.2101,
+      "step": 2130
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8060233527867135e-05,
+      "loss": 0.2149,
+      "step": 2131
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8058203300010743e-05,
+      "loss": 0.1726,
+      "step": 2132
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8056172124495126e-05,
+      "loss": 0.1902,
+      "step": 2133
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8054140001559154e-05,
+      "loss": 0.1184,
+      "step": 2134
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8052106931441806e-05,
+      "loss": 0.1582,
+      "step": 2135
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.805007291438218e-05,
+      "loss": 0.2056,
+      "step": 2136
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.804803795061947e-05,
+      "loss": 0.2455,
+      "step": 2137
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8046002040393004e-05,
+      "loss": 0.1312,
+      "step": 2138
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8043965183942203e-05,
+      "loss": 0.2,
+      "step": 2139
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8041927381506602e-05,
+      "loss": 0.1821,
+      "step": 2140
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.803988863332585e-05,
+      "loss": 0.1089,
+      "step": 2141
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8037848939639717e-05,
+      "loss": 0.1252,
+      "step": 2142
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8035808300688063e-05,
+      "loss": 0.1521,
+      "step": 2143
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8033766716710877e-05,
+      "loss": 0.204,
+      "step": 2144
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.803172418794825e-05,
+      "loss": 0.1804,
+      "step": 2145
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8029680714640393e-05,
+      "loss": 0.2375,
+      "step": 2146
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.802763629702762e-05,
+      "loss": 0.2283,
+      "step": 2147
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8025590935350347e-05,
+      "loss": 0.1857,
+      "step": 2148
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8023544629849127e-05,
+      "loss": 0.2756,
+      "step": 2149
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8021497380764602e-05,
+      "loss": 0.1535,
+      "step": 2150
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8019449188337535e-05,
+      "loss": 0.166,
+      "step": 2151
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8017400052808796e-05,
+      "loss": 0.1827,
+      "step": 2152
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8015349974419365e-05,
+      "loss": 0.1239,
+      "step": 2153
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8013298953410338e-05,
+      "loss": 0.1773,
+      "step": 2154
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.801124699002292e-05,
+      "loss": 0.1985,
+      "step": 2155
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8009194084498422e-05,
+      "loss": 0.3806,
+      "step": 2156
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8007140237078266e-05,
+      "loss": 0.208,
+      "step": 2157
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8005085448003997e-05,
+      "loss": 0.098,
+      "step": 2158
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.800302971751726e-05,
+      "loss": 0.3039,
+      "step": 2159
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8000973045859802e-05,
+      "loss": 0.1355,
+      "step": 2160
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7998915433273506e-05,
+      "loss": 0.155,
+      "step": 2161
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.799685688000034e-05,
+      "loss": 0.1943,
+      "step": 2162
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7994797386282404e-05,
+      "loss": 0.1288,
+      "step": 2163
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7992736952361887e-05,
+      "loss": 0.1431,
+      "step": 2164
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.799067557848111e-05,
+      "loss": 0.1904,
+      "step": 2165
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7988613264882485e-05,
+      "loss": 0.2711,
+      "step": 2166
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.798655001180855e-05,
+      "loss": 0.1352,
+      "step": 2167
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7984485819501943e-05,
+      "loss": 0.1858,
+      "step": 2168
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.798242068820542e-05,
+      "loss": 0.1703,
+      "step": 2169
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7980354618161845e-05,
+      "loss": 0.2539,
+      "step": 2170
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7978287609614188e-05,
+      "loss": 0.1968,
+      "step": 2171
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7976219662805535e-05,
+      "loss": 0.1622,
+      "step": 2172
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.797415077797908e-05,
+      "loss": 0.1633,
+      "step": 2173
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7972080955378126e-05,
+      "loss": 0.1359,
+      "step": 2174
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.797001019524609e-05,
+      "loss": 0.1567,
+      "step": 2175
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.79679384978265e-05,
+      "loss": 0.0846,
+      "step": 2176
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7965865863362985e-05,
+      "loss": 0.1711,
+      "step": 2177
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7963792292099292e-05,
+      "loss": 0.1405,
+      "step": 2178
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.796171778427928e-05,
+      "loss": 0.1315,
+      "step": 2179
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7959642340146915e-05,
+      "loss": 0.1603,
+      "step": 2180
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7957565959946268e-05,
+      "loss": 0.205,
+      "step": 2181
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.795548864392153e-05,
+      "loss": 0.2349,
+      "step": 2182
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.795341039231699e-05,
+      "loss": 0.114,
+      "step": 2183
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7951331205377063e-05,
+      "loss": 0.1763,
+      "step": 2184
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7949251083346264e-05,
+      "loss": 0.1943,
+      "step": 2185
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7947170026469215e-05,
+      "loss": 0.2358,
+      "step": 2186
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7945088034990652e-05,
+      "loss": 0.3292,
+      "step": 2187
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7943005109155423e-05,
+      "loss": 0.1316,
+      "step": 2188
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.794092124920848e-05,
+      "loss": 0.1434,
+      "step": 2189
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7938836455394898e-05,
+      "loss": 0.1778,
+      "step": 2190
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7936750727959845e-05,
+      "loss": 0.1668,
+      "step": 2191
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7934664067148602e-05,
+      "loss": 0.1964,
+      "step": 2192
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7932576473206577e-05,
+      "loss": 0.14,
+      "step": 2193
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7930487946379265e-05,
+      "loss": 0.154,
+      "step": 2194
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7928398486912282e-05,
+      "loss": 0.1785,
+      "step": 2195
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7926308095051354e-05,
+      "loss": 0.1585,
+      "step": 2196
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7924216771042315e-05,
+      "loss": 0.2215,
+      "step": 2197
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7922124515131108e-05,
+      "loss": 0.211,
+      "step": 2198
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7920031327563784e-05,
+      "loss": 0.181,
+      "step": 2199
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.791793720858651e-05,
+      "loss": 0.1592,
+      "step": 2200
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7915842158445557e-05,
+      "loss": 0.2219,
+      "step": 2201
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.79137461773873e-05,
+      "loss": 0.2127,
+      "step": 2202
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7911649265658237e-05,
+      "loss": 0.0847,
+      "step": 2203
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7909551423504966e-05,
+      "loss": 0.2177,
+      "step": 2204
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.79074526511742e-05,
+      "loss": 0.1806,
+      "step": 2205
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7905352948912757e-05,
+      "loss": 0.18,
+      "step": 2206
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7903252316967563e-05,
+      "loss": 0.1299,
+      "step": 2207
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.790115075558566e-05,
+      "loss": 0.1434,
+      "step": 2208
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7899048265014194e-05,
+      "loss": 0.2535,
+      "step": 2209
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.789694484550042e-05,
+      "loss": 0.143,
+      "step": 2210
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7894840497291705e-05,
+      "loss": 0.1196,
+      "step": 2211
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7892735220635526e-05,
+      "loss": 0.1514,
+      "step": 2212
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7890629015779465e-05,
+      "loss": 0.1496,
+      "step": 2213
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7888521882971218e-05,
+      "loss": 0.1522,
+      "step": 2214
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7886413822458584e-05,
+      "loss": 0.122,
+      "step": 2215
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7884304834489475e-05,
+      "loss": 0.2081,
+      "step": 2216
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.788219491931192e-05,
+      "loss": 0.2124,
+      "step": 2217
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.788008407717403e-05,
+      "loss": 0.0788,
+      "step": 2218
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7877972308324066e-05,
+      "loss": 0.1693,
+      "step": 2219
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7875859613010362e-05,
+      "loss": 0.2664,
+      "step": 2220
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.787374599148138e-05,
+      "loss": 0.1277,
+      "step": 2221
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7871631443985684e-05,
+      "loss": 0.1429,
+      "step": 2222
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7869515970771952e-05,
+      "loss": 0.2901,
+      "step": 2223
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7867399572088958e-05,
+      "loss": 0.193,
+      "step": 2224
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7865282248185604e-05,
+      "loss": 0.1406,
+      "step": 2225
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7863163999310885e-05,
+      "loss": 0.2272,
+      "step": 2226
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7861044825713922e-05,
+      "loss": 0.2116,
+      "step": 2227
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.785892472764392e-05,
+      "loss": 0.1302,
+      "step": 2228
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7856803705350207e-05,
+      "loss": 0.3932,
+      "step": 2229
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7854681759082228e-05,
+      "loss": 0.1902,
+      "step": 2230
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7852558889089523e-05,
+      "loss": 0.1317,
+      "step": 2231
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7850435095621746e-05,
+      "loss": 0.1564,
+      "step": 2232
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.784831037892866e-05,
+      "loss": 0.1889,
+      "step": 2233
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7846184739260132e-05,
+      "loss": 0.2245,
+      "step": 2234
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7844058176866147e-05,
+      "loss": 0.1232,
+      "step": 2235
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.784193069199678e-05,
+      "loss": 0.2085,
+      "step": 2236
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7839802284902243e-05,
+      "loss": 0.1843,
+      "step": 2237
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7837672955832836e-05,
+      "loss": 0.1654,
+      "step": 2238
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7835542705038963e-05,
+      "loss": 0.2769,
+      "step": 2239
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.783341153277115e-05,
+      "loss": 0.1892,
+      "step": 2240
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7831279439280034e-05,
+      "loss": 0.1221,
+      "step": 2241
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7829146424816345e-05,
+      "loss": 0.2193,
+      "step": 2242
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7827012489630932e-05,
+      "loss": 0.1415,
+      "step": 2243
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7824877633974752e-05,
+      "loss": 0.1943,
+      "step": 2244
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7822741858098864e-05,
+      "loss": 0.182,
+      "step": 2245
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7820605162254437e-05,
+      "loss": 0.1435,
+      "step": 2246
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7818467546692756e-05,
+      "loss": 0.1544,
+      "step": 2247
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7816329011665202e-05,
+      "loss": 0.1733,
+      "step": 2248
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.781418955742328e-05,
+      "loss": 0.064,
+      "step": 2249
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7812049184218586e-05,
+      "loss": 0.2267,
+      "step": 2250
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7809907892302836e-05,
+      "loss": 0.1728,
+      "step": 2251
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7807765681927845e-05,
+      "loss": 0.1564,
+      "step": 2252
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7805622553345545e-05,
+      "loss": 0.2128,
+      "step": 2253
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.780347850680797e-05,
+      "loss": 0.1611,
+      "step": 2254
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7801333542567265e-05,
+      "loss": 0.2356,
+      "step": 2255
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.779918766087568e-05,
+      "loss": 0.2321,
+      "step": 2256
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7797040861985574e-05,
+      "loss": 0.1958,
+      "step": 2257
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7794893146149418e-05,
+      "loss": 0.2322,
+      "step": 2258
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7792744513619783e-05,
+      "loss": 0.1332,
+      "step": 2259
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7790594964649353e-05,
+      "loss": 0.2794,
+      "step": 2260
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7788444499490922e-05,
+      "loss": 0.1293,
+      "step": 2261
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7786293118397385e-05,
+      "loss": 0.1758,
+      "step": 2262
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7784140821621753e-05,
+      "loss": 0.1719,
+      "step": 2263
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7781987609417134e-05,
+      "loss": 0.164,
+      "step": 2264
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7779833482036752e-05,
+      "loss": 0.1472,
+      "step": 2265
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.777767843973394e-05,
+      "loss": 0.2324,
+      "step": 2266
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7775522482762125e-05,
+      "loss": 0.1857,
+      "step": 2267
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7773365611374865e-05,
+      "loss": 0.2232,
+      "step": 2268
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7771207825825803e-05,
+      "loss": 0.1573,
+      "step": 2269
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.77690491263687e-05,
+      "loss": 0.229,
+      "step": 2270
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.776688951325742e-05,
+      "loss": 0.1635,
+      "step": 2271
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7764728986745942e-05,
+      "loss": 0.2717,
+      "step": 2272
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7762567547088346e-05,
+      "loss": 0.2342,
+      "step": 2273
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7760405194538823e-05,
+      "loss": 0.1172,
+      "step": 2274
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7758241929351672e-05,
+      "loss": 0.1857,
+      "step": 2275
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.775607775178129e-05,
+      "loss": 0.2706,
+      "step": 2276
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7753912662082188e-05,
+      "loss": 0.1536,
+      "step": 2277
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.775174666050899e-05,
+      "loss": 0.1695,
+      "step": 2278
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.774957974731642e-05,
+      "loss": 0.1133,
+      "step": 2279
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.774741192275931e-05,
+      "loss": 0.1285,
+      "step": 2280
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.77452431870926e-05,
+      "loss": 0.2335,
+      "step": 2281
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7743073540571337e-05,
+      "loss": 0.1768,
+      "step": 2282
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.774090298345068e-05,
+      "loss": 0.1828,
+      "step": 2283
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.773873151598588e-05,
+      "loss": 0.1786,
+      "step": 2284
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7736559138432316e-05,
+      "loss": 0.1781,
+      "step": 2285
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.773438585104546e-05,
+      "loss": 0.1475,
+      "step": 2286
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7732211654080895e-05,
+      "loss": 0.197,
+      "step": 2287
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7730036547794306e-05,
+      "loss": 0.2203,
+      "step": 2288
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7727860532441493e-05,
+      "loss": 0.1418,
+      "step": 2289
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.772568360827836e-05,
+      "loss": 0.1504,
+      "step": 2290
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7723505775560915e-05,
+      "loss": 0.1516,
+      "step": 2291
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7721327034545276e-05,
+      "loss": 0.2065,
+      "step": 2292
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.771914738548767e-05,
+      "loss": 0.157,
+      "step": 2293
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7716966828644425e-05,
+      "loss": 0.2035,
+      "step": 2294
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7714785364271975e-05,
+      "loss": 0.164,
+      "step": 2295
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.771260299262687e-05,
+      "loss": 0.2343,
+      "step": 2296
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7710419713965756e-05,
+      "loss": 0.1504,
+      "step": 2297
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7708235528545398e-05,
+      "loss": 0.122,
+      "step": 2298
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7706050436622648e-05,
+      "loss": 0.3432,
+      "step": 2299
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.770386443845449e-05,
+      "loss": 0.1754,
+      "step": 2300
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.770167753429799e-05,
+      "loss": 0.218,
+      "step": 2301
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7699489724410338e-05,
+      "loss": 0.1653,
+      "step": 2302
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7697301009048823e-05,
+      "loss": 0.1571,
+      "step": 2303
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7695111388470844e-05,
+      "loss": 0.1977,
+      "step": 2304
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7692920862933903e-05,
+      "loss": 0.1892,
+      "step": 2305
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.769072943269561e-05,
+      "loss": 0.1333,
+      "step": 2306
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7688537098013675e-05,
+      "loss": 0.2545,
+      "step": 2307
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.768634385914593e-05,
+      "loss": 0.1945,
+      "step": 2308
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7684149716350298e-05,
+      "loss": 0.1178,
+      "step": 2309
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7681954669884817e-05,
+      "loss": 0.167,
+      "step": 2310
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7679758720007626e-05,
+      "loss": 0.1425,
+      "step": 2311
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7677561866976973e-05,
+      "loss": 0.1757,
+      "step": 2312
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7675364111051215e-05,
+      "loss": 0.2889,
+      "step": 2313
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.767316545248881e-05,
+      "loss": 0.2299,
+      "step": 2314
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.767096589154832e-05,
+      "loss": 0.1562,
+      "step": 2315
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.766876542848843e-05,
+      "loss": 0.184,
+      "step": 2316
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.76665640635679e-05,
+      "loss": 0.2646,
+      "step": 2317
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.766436179704563e-05,
+      "loss": 0.0729,
+      "step": 2318
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7662158629180607e-05,
+      "loss": 0.1911,
+      "step": 2319
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.765995456023192e-05,
+      "loss": 0.1687,
+      "step": 2320
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.765774959045878e-05,
+      "loss": 0.1653,
+      "step": 2321
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7655543720120488e-05,
+      "loss": 0.1408,
+      "step": 2322
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7653336949476467e-05,
+      "loss": 0.1459,
+      "step": 2323
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.765112927878623e-05,
+      "loss": 0.1687,
+      "step": 2324
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7648920708309406e-05,
+      "loss": 0.1337,
+      "step": 2325
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.764671123830573e-05,
+      "loss": 0.1123,
+      "step": 2326
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.764450086903503e-05,
+      "loss": 0.1359,
+      "step": 2327
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7642289600757263e-05,
+      "loss": 0.1477,
+      "step": 2328
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7640077433732465e-05,
+      "loss": 0.2341,
+      "step": 2329
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.76378643682208e-05,
+      "loss": 0.2019,
+      "step": 2330
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7635650404482523e-05,
+      "loss": 0.1408,
+      "step": 2331
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7633435542778003e-05,
+      "loss": 0.1656,
+      "step": 2332
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.763121978336771e-05,
+      "loss": 0.1905,
+      "step": 2333
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7629003126512226e-05,
+      "loss": 0.2631,
+      "step": 2334
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.762678557247223e-05,
+      "loss": 0.2133,
+      "step": 2335
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7624567121508506e-05,
+      "loss": 0.199,
+      "step": 2336
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7622347773881956e-05,
+      "loss": 0.1339,
+      "step": 2337
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.762012752985357e-05,
+      "loss": 0.1713,
+      "step": 2338
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7617906389684466e-05,
+      "loss": 0.111,
+      "step": 2339
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7615684353635843e-05,
+      "loss": 0.1462,
+      "step": 2340
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.761346142196902e-05,
+      "loss": 0.1479,
+      "step": 2341
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7611237594945415e-05,
+      "loss": 0.1477,
+      "step": 2342
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.760901287282656e-05,
+      "loss": 0.1626,
+      "step": 2343
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7606787255874084e-05,
+      "loss": 0.1408,
+      "step": 2344
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7604560744349723e-05,
+      "loss": 0.1584,
+      "step": 2345
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7602333338515314e-05,
+      "loss": 0.2435,
+      "step": 2346
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7600105038632808e-05,
+      "loss": 0.143,
+      "step": 2347
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.759787584496426e-05,
+      "loss": 0.2704,
+      "step": 2348
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7595645757771824e-05,
+      "loss": 0.0878,
+      "step": 2349
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7593414777317763e-05,
+      "loss": 0.1889,
+      "step": 2350
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7591182903864443e-05,
+      "loss": 0.2507,
+      "step": 2351
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.758895013767434e-05,
+      "loss": 0.1712,
+      "step": 2352
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7586716479010026e-05,
+      "loss": 0.1928,
+      "step": 2353
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7584481928134186e-05,
+      "loss": 0.1599,
+      "step": 2354
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7582246485309613e-05,
+      "loss": 0.2555,
+      "step": 2355
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7580010150799185e-05,
+      "loss": 0.1547,
+      "step": 2356
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7577772924865913e-05,
+      "loss": 0.213,
+      "step": 2357
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.757553480777289e-05,
+      "loss": 0.141,
+      "step": 2358
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7573295799783328e-05,
+      "loss": 0.2518,
+      "step": 2359
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7571055901160537e-05,
+      "loss": 0.1283,
+      "step": 2360
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.756881511216793e-05,
+      "loss": 0.2337,
+      "step": 2361
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.756657343306903e-05,
+      "loss": 0.2162,
+      "step": 2362
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7564330864127464e-05,
+      "loss": 0.1629,
+      "step": 2363
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.756208740560696e-05,
+      "loss": 0.1343,
+      "step": 2364
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7559843057771352e-05,
+      "loss": 0.1155,
+      "step": 2365
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.755759782088458e-05,
+      "loss": 0.1077,
+      "step": 2366
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7555351695210695e-05,
+      "loss": 0.1079,
+      "step": 2367
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7553104681013833e-05,
+      "loss": 0.208,
+      "step": 2368
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7550856778558256e-05,
+      "loss": 0.2097,
+      "step": 2369
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7548607988108317e-05,
+      "loss": 0.2261,
+      "step": 2370
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7546358309928478e-05,
+      "loss": 0.0999,
+      "step": 2371
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.754410774428331e-05,
+      "loss": 0.1204,
+      "step": 2372
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.754185629143748e-05,
+      "loss": 0.1969,
+      "step": 2373
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7539603951655762e-05,
+      "loss": 0.1592,
+      "step": 2374
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7537350725203035e-05,
+      "loss": 0.2633,
+      "step": 2375
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7535096612344285e-05,
+      "loss": 0.1466,
+      "step": 2376
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7532841613344595e-05,
+      "loss": 0.2279,
+      "step": 2377
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7530585728469165e-05,
+      "loss": 0.1242,
+      "step": 2378
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7528328957983282e-05,
+      "loss": 0.2717,
+      "step": 2379
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7526071302152355e-05,
+      "loss": 0.1739,
+      "step": 2380
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7523812761241882e-05,
+      "loss": 0.1452,
+      "step": 2381
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7521553335517475e-05,
+      "loss": 0.2127,
+      "step": 2382
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7519293025244842e-05,
+      "loss": 0.1795,
+      "step": 2383
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7517031830689806e-05,
+      "loss": 0.2172,
+      "step": 2384
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.751476975211828e-05,
+      "loss": 0.1749,
+      "step": 2385
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7512506789796297e-05,
+      "loss": 0.1743,
+      "step": 2386
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.751024294398998e-05,
+      "loss": 0.0987,
+      "step": 2387
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7507978214965563e-05,
+      "loss": 0.1205,
+      "step": 2388
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.750571260298938e-05,
+      "loss": 0.1375,
+      "step": 2389
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7503446108327875e-05,
+      "loss": 0.1866,
+      "step": 2390
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7501178731247586e-05,
+      "loss": 0.2714,
+      "step": 2391
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7498910472015168e-05,
+      "loss": 0.1563,
+      "step": 2392
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7496641330897368e-05,
+      "loss": 0.1197,
+      "step": 2393
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7494371308161042e-05,
+      "loss": 0.1989,
+      "step": 2394
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7492100404073146e-05,
+      "loss": 0.2288,
+      "step": 2395
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.748982861890075e-05,
+      "loss": 0.1597,
+      "step": 2396
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.748755595291101e-05,
+      "loss": 0.1432,
+      "step": 2397
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7485282406371203e-05,
+      "loss": 0.2337,
+      "step": 2398
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.74830079795487e-05,
+      "loss": 0.1142,
+      "step": 2399
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7480732672710976e-05,
+      "loss": 0.143,
+      "step": 2400
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7478456486125614e-05,
+      "loss": 0.2129,
+      "step": 2401
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7476179420060298e-05,
+      "loss": 0.1027,
+      "step": 2402
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.747390147478281e-05,
+      "loss": 0.1712,
+      "step": 2403
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7471622650561044e-05,
+      "loss": 0.259,
+      "step": 2404
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7469342947662996e-05,
+      "loss": 0.1755,
+      "step": 2405
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7467062366356756e-05,
+      "loss": 0.1795,
+      "step": 2406
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7464780906910534e-05,
+      "loss": 0.2355,
+      "step": 2407
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7462498569592627e-05,
+      "loss": 0.1278,
+      "step": 2408
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7460215354671445e-05,
+      "loss": 0.213,
+      "step": 2409
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.74579312624155e-05,
+      "loss": 0.2698,
+      "step": 2410
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.74556462930934e-05,
+      "loss": 0.2845,
+      "step": 2411
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.745336044697386e-05,
+      "loss": 0.1451,
+      "step": 2412
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7451073724325706e-05,
+      "loss": 0.23,
+      "step": 2413
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.744878612541786e-05,
+      "loss": 0.2159,
+      "step": 2414
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7446497650519347e-05,
+      "loss": 0.1432,
+      "step": 2415
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7444208299899293e-05,
+      "loss": 0.2855,
+      "step": 2416
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.744191807382693e-05,
+      "loss": 0.0743,
+      "step": 2417
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7439626972571598e-05,
+      "loss": 0.1922,
+      "step": 2418
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.743733499640273e-05,
+      "loss": 0.1723,
+      "step": 2419
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7435042145589866e-05,
+      "loss": 0.2313,
+      "step": 2420
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7432748420402653e-05,
+      "loss": 0.1953,
+      "step": 2421
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7430453821110832e-05,
+      "loss": 0.2202,
+      "step": 2422
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7428158347984257e-05,
+      "loss": 0.2771,
+      "step": 2423
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7425862001292882e-05,
+      "loss": 0.2567,
+      "step": 2424
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7423564781306753e-05,
+      "loss": 0.2053,
+      "step": 2425
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.742126668829603e-05,
+      "loss": 0.162,
+      "step": 2426
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.741896772253098e-05,
+      "loss": 0.1994,
+      "step": 2427
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7416667884281958e-05,
+      "loss": 0.1345,
+      "step": 2428
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7414367173819428e-05,
+      "loss": 0.1746,
+      "step": 2429
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7412065591413965e-05,
+      "loss": 0.206,
+      "step": 2430
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.740976313733623e-05,
+      "loss": 0.2251,
+      "step": 2431
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7407459811857007e-05,
+      "loss": 0.1464,
+      "step": 2432
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.740515561524716e-05,
+      "loss": 0.1216,
+      "step": 2433
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7402850547777673e-05,
+      "loss": 0.2175,
+      "step": 2434
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7400544609719626e-05,
+      "loss": 0.1727,
+      "step": 2435
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7398237801344204e-05,
+      "loss": 0.241,
+      "step": 2436
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.739593012292268e-05,
+      "loss": 0.2485,
+      "step": 2437
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7393621574726457e-05,
+      "loss": 0.1055,
+      "step": 2438
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.739131215702701e-05,
+      "loss": 0.1723,
+      "step": 2439
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7389001870095942e-05,
+      "loss": 0.2625,
+      "step": 2440
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.738669071420494e-05,
+      "loss": 0.1739,
+      "step": 2441
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.73843786896258e-05,
+      "loss": 0.164,
+      "step": 2442
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.738206579663043e-05,
+      "loss": 0.2933,
+      "step": 2443
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7379752035490817e-05,
+      "loss": 0.168,
+      "step": 2444
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7377437406479072e-05,
+      "loss": 0.2146,
+      "step": 2445
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7375121909867396e-05,
+      "loss": 0.222,
+      "step": 2446
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7372805545928097e-05,
+      "loss": 0.283,
+      "step": 2447
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.737048831493358e-05,
+      "loss": 0.3651,
+      "step": 2448
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7368170217156363e-05,
+      "loss": 0.2472,
+      "step": 2449
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7365851252869054e-05,
+      "loss": 0.1841,
+      "step": 2450
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7363531422344365e-05,
+      "loss": 0.1103,
+      "step": 2451
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.736121072585512e-05,
+      "loss": 0.1857,
+      "step": 2452
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7358889163674227e-05,
+      "loss": 0.1447,
+      "step": 2453
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.735656673607471e-05,
+      "loss": 0.2231,
+      "step": 2454
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7354243443329694e-05,
+      "loss": 0.1425,
+      "step": 2455
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.73519192857124e-05,
+      "loss": 0.1579,
+      "step": 2456
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7349594263496152e-05,
+      "loss": 0.2199,
+      "step": 2457
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.734726837695438e-05,
+      "loss": 0.2503,
+      "step": 2458
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7344941626360606e-05,
+      "loss": 0.2561,
+      "step": 2459
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7342614011988468e-05,
+      "loss": 0.3827,
+      "step": 2460
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7340285534111695e-05,
+      "loss": 0.2257,
+      "step": 2461
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7337956193004113e-05,
+      "loss": 0.1584,
+      "step": 2462
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.733562598893967e-05,
+      "loss": 0.1664,
+      "step": 2463
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.733329492219239e-05,
+      "loss": 0.1308,
+      "step": 2464
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7330962993036416e-05,
+      "loss": 0.2276,
+      "step": 2465
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.732863020174599e-05,
+      "loss": 0.1884,
+      "step": 2466
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7326296548595447e-05,
+      "loss": 0.1293,
+      "step": 2467
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7323962033859232e-05,
+      "loss": 0.1915,
+      "step": 2468
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7321626657811887e-05,
+      "loss": 0.1311,
+      "step": 2469
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7319290420728056e-05,
+      "loss": 0.1551,
+      "step": 2470
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7316953322882487e-05,
+      "loss": 0.1887,
+      "step": 2471
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7314615364550024e-05,
+      "loss": 0.2203,
+      "step": 2472
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7312276546005615e-05,
+      "loss": 0.1471,
+      "step": 2473
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7309936867524314e-05,
+      "loss": 0.2768,
+      "step": 2474
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7307596329381263e-05,
+      "loss": 0.2719,
+      "step": 2475
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.730525493185172e-05,
+      "loss": 0.0854,
+      "step": 2476
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.730291267521104e-05,
+      "loss": 0.1523,
+      "step": 2477
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.730056955973467e-05,
+      "loss": 0.2208,
+      "step": 2478
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.729822558569817e-05,
+      "loss": 0.2001,
+      "step": 2479
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.729588075337719e-05,
+      "loss": 0.2361,
+      "step": 2480
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7293535063047487e-05,
+      "loss": 0.1166,
+      "step": 2481
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7291188514984924e-05,
+      "loss": 0.1454,
+      "step": 2482
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7288841109465454e-05,
+      "loss": 0.0983,
+      "step": 2483
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7286492846765142e-05,
+      "loss": 0.2073,
+      "step": 2484
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.728414372716014e-05,
+      "loss": 0.1268,
+      "step": 2485
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7281793750926717e-05,
+      "loss": 0.143,
+      "step": 2486
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7279442918341227e-05,
+      "loss": 0.2117,
+      "step": 2487
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7277091229680136e-05,
+      "loss": 0.0796,
+      "step": 2488
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7274738685220005e-05,
+      "loss": 0.1059,
+      "step": 2489
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.72723852852375e-05,
+      "loss": 0.253,
+      "step": 2490
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7270031030009385e-05,
+      "loss": 0.2042,
+      "step": 2491
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7267675919812524e-05,
+      "loss": 0.1804,
+      "step": 2492
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7265319954923878e-05,
+      "loss": 0.1745,
+      "step": 2493
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7262963135620518e-05,
+      "loss": 0.1734,
+      "step": 2494
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.726060546217961e-05,
+      "loss": 0.2806,
+      "step": 2495
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.725824693487842e-05,
+      "loss": 0.0769,
+      "step": 2496
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.725588755399431e-05,
+      "loss": 0.1786,
+      "step": 2497
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7253527319804754e-05,
+      "loss": 0.0852,
+      "step": 2498
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7251166232587318e-05,
+      "loss": 0.1722,
+      "step": 2499
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.724880429261967e-05,
+      "loss": 0.2642,
+      "step": 2500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7246441500179573e-05,
+      "loss": 0.1824,
+      "step": 2501
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7244077855544904e-05,
+      "loss": 0.1661,
+      "step": 2502
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7241713358993634e-05,
+      "loss": 0.2386,
+      "step": 2503
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7239348010803818e-05,
+      "loss": 0.1133,
+      "step": 2504
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7236981811253638e-05,
+      "loss": 0.1591,
+      "step": 2505
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7234614760621357e-05,
+      "loss": 0.1516,
+      "step": 2506
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.723224685918535e-05,
+      "loss": 0.1034,
+      "step": 2507
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7229878107224077e-05,
+      "loss": 0.0995,
+      "step": 2508
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.722750850501612e-05,
+      "loss": 0.1021,
+      "step": 2509
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7225138052840137e-05,
+      "loss": 0.2762,
+      "step": 2510
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7222766750974906e-05,
+      "loss": 0.1723,
+      "step": 2511
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.722039459969929e-05,
+      "loss": 0.1678,
+      "step": 2512
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7218021599292264e-05,
+      "loss": 0.1895,
+      "step": 2513
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7215647750032893e-05,
+      "loss": 0.1747,
+      "step": 2514
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7213273052200346e-05,
+      "loss": 0.1607,
+      "step": 2515
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7210897506073895e-05,
+      "loss": 0.197,
+      "step": 2516
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.72085211119329e-05,
+      "loss": 0.2523,
+      "step": 2517
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7206143870056843e-05,
+      "loss": 0.2375,
+      "step": 2518
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.720376578072528e-05,
+      "loss": 0.2328,
+      "step": 2519
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7201386844217887e-05,
+      "loss": 0.1404,
+      "step": 2520
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7199007060814426e-05,
+      "loss": 0.225,
+      "step": 2521
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7196626430794765e-05,
+      "loss": 0.1215,
+      "step": 2522
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.719424495443887e-05,
+      "loss": 0.1455,
+      "step": 2523
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7191862632026807e-05,
+      "loss": 0.1104,
+      "step": 2524
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.718947946383874e-05,
+      "loss": 0.328,
+      "step": 2525
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7187095450154938e-05,
+      "loss": 0.2066,
+      "step": 2526
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7184710591255766e-05,
+      "loss": 0.2062,
+      "step": 2527
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.718232488742168e-05,
+      "loss": 0.1906,
+      "step": 2528
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.717993833893325e-05,
+      "loss": 0.2014,
+      "step": 2529
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7177550946071132e-05,
+      "loss": 0.1138,
+      "step": 2530
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.71751627091161e-05,
+      "loss": 0.1464,
+      "step": 2531
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7172773628349002e-05,
+      "loss": 0.2253,
+      "step": 2532
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7170383704050802e-05,
+      "loss": 0.2292,
+      "step": 2533
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.716799293650256e-05,
+      "loss": 0.1689,
+      "step": 2534
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7165601325985437e-05,
+      "loss": 0.1921,
+      "step": 2535
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.716320887278069e-05,
+      "loss": 0.1228,
+      "step": 2536
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7160815577169674e-05,
+      "loss": 0.1069,
+      "step": 2537
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7158421439433845e-05,
+      "loss": 0.2052,
+      "step": 2538
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7156026459854757e-05,
+      "loss": 0.0874,
+      "step": 2539
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7153630638714064e-05,
+      "loss": 0.1381,
+      "step": 2540
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7151233976293526e-05,
+      "loss": 0.2154,
+      "step": 2541
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7148836472874984e-05,
+      "loss": 0.1172,
+      "step": 2542
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7146438128740395e-05,
+      "loss": 0.2245,
+      "step": 2543
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7144038944171808e-05,
+      "loss": 0.1928,
+      "step": 2544
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7141638919451367e-05,
+      "loss": 0.1948,
+      "step": 2545
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7139238054861325e-05,
+      "loss": 0.1474,
+      "step": 2546
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.713683635068403e-05,
+      "loss": 0.1139,
+      "step": 2547
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7134433807201918e-05,
+      "loss": 0.212,
+      "step": 2548
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7132030424697536e-05,
+      "loss": 0.1962,
+      "step": 2549
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.712962620345353e-05,
+      "loss": 0.1718,
+      "step": 2550
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.712722114375264e-05,
+      "loss": 0.2016,
+      "step": 2551
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7124815245877703e-05,
+      "loss": 0.1264,
+      "step": 2552
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7122408510111653e-05,
+      "loss": 0.2133,
+      "step": 2553
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.712000093673753e-05,
+      "loss": 0.2911,
+      "step": 2554
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7117592526038477e-05,
+      "loss": 0.187,
+      "step": 2555
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7115183278297715e-05,
+      "loss": 0.1843,
+      "step": 2556
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.711277319379858e-05,
+      "loss": 0.1795,
+      "step": 2557
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7110362272824505e-05,
+      "loss": 0.1484,
+      "step": 2558
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7107950515659014e-05,
+      "loss": 0.2446,
+      "step": 2559
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7105537922585744e-05,
+      "loss": 0.1871,
+      "step": 2560
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7103124493888406e-05,
+      "loss": 0.1532,
+      "step": 2561
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7100710229850835e-05,
+      "loss": 0.3215,
+      "step": 2562
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7098295130756945e-05,
+      "loss": 0.1474,
+      "step": 2563
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7095879196890765e-05,
+      "loss": 0.215,
+      "step": 2564
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7093462428536402e-05,
+      "loss": 0.145,
+      "step": 2565
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.709104482597808e-05,
+      "loss": 0.1067,
+      "step": 2566
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.708862638950011e-05,
+      "loss": 0.0992,
+      "step": 2567
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.708620711938691e-05,
+      "loss": 0.2177,
+      "step": 2568
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7083787015922985e-05,
+      "loss": 0.148,
+      "step": 2569
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7081366079392946e-05,
+      "loss": 0.2095,
+      "step": 2570
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7078944310081497e-05,
+      "loss": 0.1632,
+      "step": 2571
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7076521708273442e-05,
+      "loss": 0.1675,
+      "step": 2572
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7074098274253684e-05,
+      "loss": 0.1864,
+      "step": 2573
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.707167400830723e-05,
+      "loss": 0.1501,
+      "step": 2574
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.706924891071917e-05,
+      "loss": 0.1548,
+      "step": 2575
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.70668229817747e-05,
+      "loss": 0.2251,
+      "step": 2576
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7064396221759122e-05,
+      "loss": 0.1583,
+      "step": 2577
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7061968630957817e-05,
+      "loss": 0.1096,
+      "step": 2578
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7059540209656278e-05,
+      "loss": 0.1304,
+      "step": 2579
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7057110958140092e-05,
+      "loss": 0.2354,
+      "step": 2580
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7054680876694947e-05,
+      "loss": 0.2302,
+      "step": 2581
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7052249965606616e-05,
+      "loss": 0.1162,
+      "step": 2582
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7049818225160988e-05,
+      "loss": 0.1527,
+      "step": 2583
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7047385655644035e-05,
+      "loss": 0.1681,
+      "step": 2584
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7044952257341832e-05,
+      "loss": 0.11,
+      "step": 2585
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7042518030540553e-05,
+      "loss": 0.1036,
+      "step": 2586
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7040082975526467e-05,
+      "loss": 0.267,
+      "step": 2587
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7037647092585935e-05,
+      "loss": 0.2747,
+      "step": 2588
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7035210382005432e-05,
+      "loss": 0.2948,
+      "step": 2589
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7032772844071507e-05,
+      "loss": 0.1635,
+      "step": 2590
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7030334479070832e-05,
+      "loss": 0.2145,
+      "step": 2591
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.702789528729015e-05,
+      "loss": 0.1468,
+      "step": 2592
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7025455269016328e-05,
+      "loss": 0.1123,
+      "step": 2593
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7023014424536306e-05,
+      "loss": 0.1828,
+      "step": 2594
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7020572754137134e-05,
+      "loss": 0.2391,
+      "step": 2595
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.701813025810596e-05,
+      "loss": 0.1395,
+      "step": 2596
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7015686936730022e-05,
+      "loss": 0.1079,
+      "step": 2597
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.701324279029666e-05,
+      "loss": 0.2438,
+      "step": 2598
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7010797819093316e-05,
+      "loss": 0.1389,
+      "step": 2599
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7008352023407515e-05,
+      "loss": 0.1682,
+      "step": 2600
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7005905403526892e-05,
+      "loss": 0.1697,
+      "step": 2601
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.700345795973917e-05,
+      "loss": 0.1949,
+      "step": 2602
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7001009692332178e-05,
+      "loss": 0.1538,
+      "step": 2603
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.699856060159383e-05,
+      "loss": 0.2733,
+      "step": 2604
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6996110687812145e-05,
+      "loss": 0.1367,
+      "step": 2605
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6993659951275243e-05,
+      "loss": 0.1212,
+      "step": 2606
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.699120839227133e-05,
+      "loss": 0.1431,
+      "step": 2607
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6988756011088714e-05,
+      "loss": 0.1977,
+      "step": 2608
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6986302808015803e-05,
+      "loss": 0.1584,
+      "step": 2609
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.698384878334109e-05,
+      "loss": 0.1219,
+      "step": 2610
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6981393937353182e-05,
+      "loss": 0.1769,
+      "step": 2611
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6978938270340772e-05,
+      "loss": 0.1521,
+      "step": 2612
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6976481782592643e-05,
+      "loss": 0.1001,
+      "step": 2613
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.697402447439769e-05,
+      "loss": 0.129,
+      "step": 2614
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6971566346044898e-05,
+      "loss": 0.2185,
+      "step": 2615
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6969107397823338e-05,
+      "loss": 0.1648,
+      "step": 2616
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6966647630022198e-05,
+      "loss": 0.0684,
+      "step": 2617
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.696418704293074e-05,
+      "loss": 0.11,
+      "step": 2618
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6961725636838346e-05,
+      "loss": 0.1235,
+      "step": 2619
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.695926341203447e-05,
+      "loss": 0.2034,
+      "step": 2620
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6956800368808682e-05,
+      "loss": 0.1189,
+      "step": 2621
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.695433650745064e-05,
+      "loss": 0.2264,
+      "step": 2622
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6951871828250087e-05,
+      "loss": 0.2022,
+      "step": 2623
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6949406331496888e-05,
+      "loss": 0.1263,
+      "step": 2624
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6946940017480986e-05,
+      "loss": 0.1383,
+      "step": 2625
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.694447288649242e-05,
+      "loss": 0.1645,
+      "step": 2626
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6942004938821336e-05,
+      "loss": 0.1783,
+      "step": 2627
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.693953617475796e-05,
+      "loss": 0.1733,
+      "step": 2628
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.693706659459263e-05,
+      "loss": 0.2201,
+      "step": 2629
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.693459619861577e-05,
+      "loss": 0.1103,
+      "step": 2630
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.69321249871179e-05,
+      "loss": 0.1375,
+      "step": 2631
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.692965296038965e-05,
+      "loss": 0.1201,
+      "step": 2632
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6927180118721725e-05,
+      "loss": 0.1256,
+      "step": 2633
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6924706462404936e-05,
+      "loss": 0.1257,
+      "step": 2634
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6922231991730193e-05,
+      "loss": 0.19,
+      "step": 2635
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6919756706988497e-05,
+      "loss": 0.191,
+      "step": 2636
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.691728060847095e-05,
+      "loss": 0.277,
+      "step": 2637
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6914803696468735e-05,
+      "loss": 0.1551,
+      "step": 2638
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.691232597127315e-05,
+      "loss": 0.2132,
+      "step": 2639
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6909847433175577e-05,
+      "loss": 0.209,
+      "step": 2640
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.69073680824675e-05,
+      "loss": 0.1445,
+      "step": 2641
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.690488791944049e-05,
+      "loss": 0.2641,
+      "step": 2642
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6902406944386224e-05,
+      "loss": 0.178,
+      "step": 2643
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6899925157596466e-05,
+      "loss": 0.1747,
+      "step": 2644
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6897442559363076e-05,
+      "loss": 0.2519,
+      "step": 2645
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6894959149978013e-05,
+      "loss": 0.137,
+      "step": 2646
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6892474929733334e-05,
+      "loss": 0.2212,
+      "step": 2647
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.688998989892119e-05,
+      "loss": 0.1539,
+      "step": 2648
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6887504057833815e-05,
+      "loss": 0.1403,
+      "step": 2649
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.688501740676356e-05,
+      "loss": 0.2052,
+      "step": 2650
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.688252994600285e-05,
+      "loss": 0.08,
+      "step": 2651
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6880041675844217e-05,
+      "loss": 0.2357,
+      "step": 2652
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6877552596580295e-05,
+      "loss": 0.2255,
+      "step": 2653
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6875062708503795e-05,
+      "loss": 0.192,
+      "step": 2654
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6872572011907533e-05,
+      "loss": 0.1335,
+      "step": 2655
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.687008050708442e-05,
+      "loss": 0.1699,
+      "step": 2656
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6867588194327466e-05,
+      "loss": 0.1445,
+      "step": 2657
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6865095073929763e-05,
+      "loss": 0.202,
+      "step": 2658
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6862601146184517e-05,
+      "loss": 0.1557,
+      "step": 2659
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6860106411385008e-05,
+      "loss": 0.1673,
+      "step": 2660
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.685761086982463e-05,
+      "loss": 0.1377,
+      "step": 2661
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6855114521796855e-05,
+      "loss": 0.1563,
+      "step": 2662
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6852617367595265e-05,
+      "loss": 0.1139,
+      "step": 2663
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6850119407513526e-05,
+      "loss": 0.2157,
+      "step": 2664
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6847620641845405e-05,
+      "loss": 0.1071,
+      "step": 2665
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.684512107088476e-05,
+      "loss": 0.0998,
+      "step": 2666
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6842620694925544e-05,
+      "loss": 0.166,
+      "step": 2667
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6840119514261807e-05,
+      "loss": 0.1722,
+      "step": 2668
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6837617529187692e-05,
+      "loss": 0.2063,
+      "step": 2669
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6835114739997438e-05,
+      "loss": 0.1644,
+      "step": 2670
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6832611146985372e-05,
+      "loss": 0.1844,
+      "step": 2671
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6830106750445933e-05,
+      "loss": 0.0943,
+      "step": 2672
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.682760155067363e-05,
+      "loss": 0.1164,
+      "step": 2673
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.682509554796309e-05,
+      "loss": 0.1865,
+      "step": 2674
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6822588742609012e-05,
+      "loss": 0.2387,
+      "step": 2675
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.682008113490621e-05,
+      "loss": 0.1695,
+      "step": 2676
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6817572725149575e-05,
+      "loss": 0.144,
+      "step": 2677
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.681506351363411e-05,
+      "loss": 0.1609,
+      "step": 2678
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6812553500654897e-05,
+      "loss": 0.1916,
+      "step": 2679
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6810042686507117e-05,
+      "loss": 0.1749,
+      "step": 2680
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.680753107148605e-05,
+      "loss": 0.1892,
+      "step": 2681
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6805018655887066e-05,
+      "loss": 0.1797,
+      "step": 2682
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6802505440005624e-05,
+      "loss": 0.2239,
+      "step": 2683
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.679999142413729e-05,
+      "loss": 0.1503,
+      "step": 2684
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6797476608577717e-05,
+      "loss": 0.2506,
+      "step": 2685
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6794960993622646e-05,
+      "loss": 0.1425,
+      "step": 2686
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.679244457956792e-05,
+      "loss": 0.2479,
+      "step": 2687
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6789927366709474e-05,
+      "loss": 0.1896,
+      "step": 2688
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6787409355343334e-05,
+      "loss": 0.2012,
+      "step": 2689
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6784890545765634e-05,
+      "loss": 0.2274,
+      "step": 2690
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6782370938272577e-05,
+      "loss": 0.1015,
+      "step": 2691
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.677985053316048e-05,
+      "loss": 0.2399,
+      "step": 2692
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.677732933072574e-05,
+      "loss": 0.2384,
+      "step": 2693
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.677480733126487e-05,
+      "loss": 0.1642,
+      "step": 2694
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.677228453507445e-05,
+      "loss": 0.1663,
+      "step": 2695
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6769760942451165e-05,
+      "loss": 0.2222,
+      "step": 2696
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.67672365536918e-05,
+      "loss": 0.2464,
+      "step": 2697
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6764711369093223e-05,
+      "loss": 0.23,
+      "step": 2698
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6762185388952402e-05,
+      "loss": 0.1166,
+      "step": 2699
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6759658613566398e-05,
+      "loss": 0.2839,
+      "step": 2700
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.675713104323236e-05,
+      "loss": 0.2236,
+      "step": 2701
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6754602678247544e-05,
+      "loss": 0.1822,
+      "step": 2702
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6752073518909278e-05,
+      "loss": 0.1901,
+      "step": 2703
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6749543565515002e-05,
+      "loss": 0.1942,
+      "step": 2704
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6747012818362244e-05,
+      "loss": 0.0907,
+      "step": 2705
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.674448127774863e-05,
+      "loss": 0.0508,
+      "step": 2706
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6741948943971858e-05,
+      "loss": 0.1954,
+      "step": 2707
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6739415817329743e-05,
+      "loss": 0.143,
+      "step": 2708
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6736881898120195e-05,
+      "loss": 0.148,
+      "step": 2709
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6734347186641194e-05,
+      "loss": 0.17,
+      "step": 2710
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6731811683190834e-05,
+      "loss": 0.2122,
+      "step": 2711
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.672927538806729e-05,
+      "loss": 0.166,
+      "step": 2712
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6726738301568833e-05,
+      "loss": 0.1359,
+      "step": 2713
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6724200423993843e-05,
+      "loss": 0.2292,
+      "step": 2714
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.672166175564076e-05,
+      "loss": 0.0877,
+      "step": 2715
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.671912229680815e-05,
+      "loss": 0.1715,
+      "step": 2716
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.671658204779465e-05,
+      "loss": 0.1531,
+      "step": 2717
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6714041008899004e-05,
+      "loss": 0.2421,
+      "step": 2718
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6711499180420035e-05,
+      "loss": 0.1982,
+      "step": 2719
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6708956562656674e-05,
+      "loss": 0.2268,
+      "step": 2720
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6706413155907936e-05,
+      "loss": 0.1813,
+      "step": 2721
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6703868960472926e-05,
+      "loss": 0.1851,
+      "step": 2722
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.670132397665085e-05,
+      "loss": 0.1479,
+      "step": 2723
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6698778204741002e-05,
+      "loss": 0.1585,
+      "step": 2724
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6696231645042768e-05,
+      "loss": 0.1382,
+      "step": 2725
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.669368429785563e-05,
+      "loss": 0.207,
+      "step": 2726
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6691136163479157e-05,
+      "loss": 0.2532,
+      "step": 2727
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6688587242213018e-05,
+      "loss": 0.1598,
+      "step": 2728
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6686037534356973e-05,
+      "loss": 0.2216,
+      "step": 2729
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6683487040210865e-05,
+      "loss": 0.3255,
+      "step": 2730
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6680935760074638e-05,
+      "loss": 0.1133,
+      "step": 2731
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6678383694248332e-05,
+      "loss": 0.2096,
+      "step": 2732
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.667583084303207e-05,
+      "loss": 0.2746,
+      "step": 2733
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6673277206726077e-05,
+      "loss": 0.0941,
+      "step": 2734
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.667072278563066e-05,
+      "loss": 0.1336,
+      "step": 2735
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6668167580046228e-05,
+      "loss": 0.1594,
+      "step": 2736
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6665611590273275e-05,
+      "loss": 0.1538,
+      "step": 2737
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.666305481661239e-05,
+      "loss": 0.2263,
+      "step": 2738
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6660497259364252e-05,
+      "loss": 0.143,
+      "step": 2739
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6657938918829638e-05,
+      "loss": 0.1937,
+      "step": 2740
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6655379795309415e-05,
+      "loss": 0.2255,
+      "step": 2741
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6652819889104538e-05,
+      "loss": 0.1851,
+      "step": 2742
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6650259200516053e-05,
+      "loss": 0.292,
+      "step": 2743
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6647697729845104e-05,
+      "loss": 0.2371,
+      "step": 2744
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.664513547739293e-05,
+      "loss": 0.1599,
+      "step": 2745
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6642572443460853e-05,
+      "loss": 0.2002,
+      "step": 2746
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6640008628350286e-05,
+      "loss": 0.1554,
+      "step": 2747
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6637444032362745e-05,
+      "loss": 0.1878,
+      "step": 2748
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6634878655799825e-05,
+      "loss": 0.0857,
+      "step": 2749
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6632312498963222e-05,
+      "loss": 0.2091,
+      "step": 2750
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.662974556215472e-05,
+      "loss": 0.1577,
+      "step": 2751
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.66271778456762e-05,
+      "loss": 0.3094,
+      "step": 2752
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6624609349829623e-05,
+      "loss": 0.1421,
+      "step": 2753
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6622040074917053e-05,
+      "loss": 0.1365,
+      "step": 2754
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.661947002124064e-05,
+      "loss": 0.1768,
+      "step": 2755
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6616899189102627e-05,
+      "loss": 0.1725,
+      "step": 2756
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6614327578805346e-05,
+      "loss": 0.0974,
+      "step": 2757
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6611755190651232e-05,
+      "loss": 0.1733,
+      "step": 2758
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.660918202494279e-05,
+      "loss": 0.1162,
+      "step": 2759
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6606608081982635e-05,
+      "loss": 0.1897,
+      "step": 2760
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.660403336207347e-05,
+      "loss": 0.146,
+      "step": 2761
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6601457865518082e-05,
+      "loss": 0.2005,
+      "step": 2762
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6598881592619357e-05,
+      "loss": 0.1787,
+      "step": 2763
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6596304543680267e-05,
+      "loss": 0.2806,
+      "step": 2764
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.659372671900388e-05,
+      "loss": 0.2405,
+      "step": 2765
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6591148118893355e-05,
+      "loss": 0.0931,
+      "step": 2766
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6588568743651934e-05,
+      "loss": 0.1406,
+      "step": 2767
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6585988593582957e-05,
+      "loss": 0.1893,
+      "step": 2768
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6583407668989858e-05,
+      "loss": 0.1664,
+      "step": 2769
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6580825970176156e-05,
+      "loss": 0.1061,
+      "step": 2770
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6578243497445466e-05,
+      "loss": 0.1716,
+      "step": 2771
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.657566025110149e-05,
+      "loss": 0.1546,
+      "step": 2772
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.657307623144802e-05,
+      "loss": 0.1535,
+      "step": 2773
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6570491438788944e-05,
+      "loss": 0.1754,
+      "step": 2774
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6567905873428237e-05,
+      "loss": 0.2061,
+      "step": 2775
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.656531953566997e-05,
+      "loss": 0.2575,
+      "step": 2776
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.656273242581829e-05,
+      "loss": 0.1741,
+      "step": 2777
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.656014454417746e-05,
+      "loss": 0.196,
+      "step": 2778
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6557555891051812e-05,
+      "loss": 0.1672,
+      "step": 2779
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6554966466745773e-05,
+      "loss": 0.2001,
+      "step": 2780
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6552376271563874e-05,
+      "loss": 0.2305,
+      "step": 2781
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6549785305810723e-05,
+      "loss": 0.0702,
+      "step": 2782
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6547193569791014e-05,
+      "loss": 0.0491,
+      "step": 2783
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6544601063809547e-05,
+      "loss": 0.2174,
+      "step": 2784
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6542007788171206e-05,
+      "loss": 0.166,
+      "step": 2785
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6539413743180963e-05,
+      "loss": 0.2125,
+      "step": 2786
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.653681892914388e-05,
+      "loss": 0.1978,
+      "step": 2787
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.653422334636512e-05,
+      "loss": 0.1796,
+      "step": 2788
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6531626995149916e-05,
+      "loss": 0.2519,
+      "step": 2789
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6529029875803614e-05,
+      "loss": 0.246,
+      "step": 2790
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6526431988631638e-05,
+      "loss": 0.1797,
+      "step": 2791
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.65238333339395e-05,
+      "loss": 0.2405,
+      "step": 2792
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6521233912032806e-05,
+      "loss": 0.1415,
+      "step": 2793
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6518633723217257e-05,
+      "loss": 0.1804,
+      "step": 2794
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6516032767798643e-05,
+      "loss": 0.1608,
+      "step": 2795
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.651343104608283e-05,
+      "loss": 0.1081,
+      "step": 2796
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6510828558375795e-05,
+      "loss": 0.1904,
+      "step": 2797
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.650822530498359e-05,
+      "loss": 0.1834,
+      "step": 2798
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.650562128621237e-05,
+      "loss": 0.1519,
+      "step": 2799
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6503016502368358e-05,
+      "loss": 0.2008,
+      "step": 2800
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6500410953757895e-05,
+      "loss": 0.1692,
+      "step": 2801
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.649780464068739e-05,
+      "loss": 0.164,
+      "step": 2802
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6495197563463358e-05,
+      "loss": 0.1713,
+      "step": 2803
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6492589722392386e-05,
+      "loss": 0.2987,
+      "step": 2804
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6489981117781166e-05,
+      "loss": 0.2343,
+      "step": 2805
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6487371749936474e-05,
+      "loss": 0.2405,
+      "step": 2806
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6484761619165182e-05,
+      "loss": 0.198,
+      "step": 2807
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6482150725774234e-05,
+      "loss": 0.1937,
+      "step": 2808
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6479539070070686e-05,
+      "loss": 0.1401,
+      "step": 2809
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6476926652361668e-05,
+      "loss": 0.1749,
+      "step": 2810
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.647431347295441e-05,
+      "loss": 0.1802,
+      "step": 2811
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6471699532156226e-05,
+      "loss": 0.1884,
+      "step": 2812
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6469084830274513e-05,
+      "loss": 0.1152,
+      "step": 2813
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6466469367616772e-05,
+      "loss": 0.1714,
+      "step": 2814
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.646385314449058e-05,
+      "loss": 0.1882,
+      "step": 2815
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6461236161203617e-05,
+      "loss": 0.2279,
+      "step": 2816
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.645861841806364e-05,
+      "loss": 0.1815,
+      "step": 2817
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.64559999153785e-05,
+      "loss": 0.1569,
+      "step": 2818
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.645338065345614e-05,
+      "loss": 0.2352,
+      "step": 2819
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6450760632604587e-05,
+      "loss": 0.238,
+      "step": 2820
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6448139853131967e-05,
+      "loss": 0.1497,
+      "step": 2821
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.644551831534648e-05,
+      "loss": 0.1751,
+      "step": 2822
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.644289601955643e-05,
+      "loss": 0.1256,
+      "step": 2823
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6440272966070193e-05,
+      "loss": 0.2051,
+      "step": 2824
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.643764915519626e-05,
+      "loss": 0.2003,
+      "step": 2825
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6435024587243185e-05,
+      "loss": 0.2263,
+      "step": 2826
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.643239926251963e-05,
+      "loss": 0.2146,
+      "step": 2827
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6429773181334327e-05,
+      "loss": 0.1235,
+      "step": 2828
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.642714634399612e-05,
+      "loss": 0.2421,
+      "step": 2829
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.642451875081392e-05,
+      "loss": 0.2093,
+      "step": 2830
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6421890402096744e-05,
+      "loss": 0.1914,
+      "step": 2831
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6419261298153684e-05,
+      "loss": 0.2239,
+      "step": 2832
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6416631439293932e-05,
+      "loss": 0.1157,
+      "step": 2833
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.641400082582676e-05,
+      "loss": 0.1944,
+      "step": 2834
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6411369458061542e-05,
+      "loss": 0.1455,
+      "step": 2835
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.640873733630772e-05,
+      "loss": 0.2451,
+      "step": 2836
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6406104460874844e-05,
+      "loss": 0.1614,
+      "step": 2837
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.640347083207254e-05,
+      "loss": 0.2613,
+      "step": 2838
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6400836450210533e-05,
+      "loss": 0.1099,
+      "step": 2839
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6398201315598626e-05,
+      "loss": 0.1964,
+      "step": 2840
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6395565428546715e-05,
+      "loss": 0.173,
+      "step": 2841
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.639292878936479e-05,
+      "loss": 0.1651,
+      "step": 2842
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6390291398362922e-05,
+      "loss": 0.1853,
+      "step": 2843
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6387653255851276e-05,
+      "loss": 0.1033,
+      "step": 2844
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6385014362140095e-05,
+      "loss": 0.1467,
+      "step": 2845
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6382374717539722e-05,
+      "loss": 0.3215,
+      "step": 2846
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.637973432236059e-05,
+      "loss": 0.2242,
+      "step": 2847
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.63770931769132e-05,
+      "loss": 0.146,
+      "step": 2848
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6374451281508164e-05,
+      "loss": 0.1419,
+      "step": 2849
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6371808636456176e-05,
+      "loss": 0.1418,
+      "step": 2850
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6369165242068017e-05,
+      "loss": 0.1615,
+      "step": 2851
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6366521098654548e-05,
+      "loss": 0.2147,
+      "step": 2852
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6363876206526726e-05,
+      "loss": 0.0648,
+      "step": 2853
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6361230565995598e-05,
+      "loss": 0.1182,
+      "step": 2854
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6358584177372296e-05,
+      "loss": 0.1278,
+      "step": 2855
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.635593704096804e-05,
+      "loss": 0.2683,
+      "step": 2856
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6353289157094137e-05,
+      "loss": 0.1994,
+      "step": 2857
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6350640526061984e-05,
+      "loss": 0.1997,
+      "step": 2858
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6347991148183063e-05,
+      "loss": 0.1953,
+      "step": 2859
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6345341023768947e-05,
+      "loss": 0.1454,
+      "step": 2860
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6342690153131293e-05,
+      "loss": 0.1516,
+      "step": 2861
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.634003853658185e-05,
+      "loss": 0.1135,
+      "step": 2862
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6337386174432458e-05,
+      "loss": 0.1702,
+      "step": 2863
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.633473306699503e-05,
+      "loss": 0.3628,
+      "step": 2864
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.633207921458158e-05,
+      "loss": 0.1521,
+      "step": 2865
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.632942461750421e-05,
+      "loss": 0.2079,
+      "step": 2866
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.63267692760751e-05,
+      "loss": 0.1519,
+      "step": 2867
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6324113190606527e-05,
+      "loss": 0.2237,
+      "step": 2868
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6321456361410848e-05,
+      "loss": 0.105,
+      "step": 2869
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.631879878880051e-05,
+      "loss": 0.2393,
+      "step": 2870
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6316140473088058e-05,
+      "loss": 0.2038,
+      "step": 2871
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6313481414586104e-05,
+      "loss": 0.2376,
+      "step": 2872
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6310821613607357e-05,
+      "loss": 0.1628,
+      "step": 2873
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6308161070464624e-05,
+      "loss": 0.16,
+      "step": 2874
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6305499785470785e-05,
+      "loss": 0.1775,
+      "step": 2875
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6302837758938807e-05,
+      "loss": 0.1879,
+      "step": 2876
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6300174991181755e-05,
+      "loss": 0.2648,
+      "step": 2877
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6297511482512774e-05,
+      "loss": 0.1505,
+      "step": 2878
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6294847233245098e-05,
+      "loss": 0.2695,
+      "step": 2879
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.629218224369205e-05,
+      "loss": 0.2859,
+      "step": 2880
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6289516514167028e-05,
+      "loss": 0.1393,
+      "step": 2881
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6286850044983536e-05,
+      "loss": 0.1557,
+      "step": 2882
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6284182836455152e-05,
+      "loss": 0.1334,
+      "step": 2883
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6281514888895546e-05,
+      "loss": 0.1852,
+      "step": 2884
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6278846202618473e-05,
+      "loss": 0.2156,
+      "step": 2885
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6276176777937772e-05,
+      "loss": 0.1962,
+      "step": 2886
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6273506615167377e-05,
+      "loss": 0.129,
+      "step": 2887
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6270835714621304e-05,
+      "loss": 0.1825,
+      "step": 2888
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6268164076613652e-05,
+      "loss": 0.0978,
+      "step": 2889
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.626549170145861e-05,
+      "loss": 0.2062,
+      "step": 2890
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6262818589470457e-05,
+      "loss": 0.2156,
+      "step": 2891
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6260144740963556e-05,
+      "loss": 0.1082,
+      "step": 2892
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6257470156252353e-05,
+      "loss": 0.182,
+      "step": 2893
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.625479483565139e-05,
+      "loss": 0.2353,
+      "step": 2894
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6252118779475282e-05,
+      "loss": 0.2062,
+      "step": 2895
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6249441988038745e-05,
+      "loss": 0.1562,
+      "step": 2896
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.624676446165657e-05,
+      "loss": 0.1255,
+      "step": 2897
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.624408620064364e-05,
+      "loss": 0.1566,
+      "step": 2898
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6241407205314924e-05,
+      "loss": 0.165,
+      "step": 2899
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6238727475985477e-05,
+      "loss": 0.2416,
+      "step": 2900
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.623604701297044e-05,
+      "loss": 0.1423,
+      "step": 2901
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6233365816585037e-05,
+      "loss": 0.3346,
+      "step": 2902
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.623068388714459e-05,
+      "loss": 0.1281,
+      "step": 2903
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6228001224964484e-05,
+      "loss": 0.0879,
+      "step": 2904
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.622531783036022e-05,
+      "loss": 0.1838,
+      "step": 2905
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6222633703647362e-05,
+      "loss": 0.1704,
+      "step": 2906
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6219948845141572e-05,
+      "loss": 0.2215,
+      "step": 2907
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.621726325515859e-05,
+      "loss": 0.2161,
+      "step": 2908
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.621457693401425e-05,
+      "loss": 0.1249,
+      "step": 2909
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6211889882024464e-05,
+      "loss": 0.1751,
+      "step": 2910
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6209202099505236e-05,
+      "loss": 0.1705,
+      "step": 2911
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.620651358677266e-05,
+      "loss": 0.1474,
+      "step": 2912
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6203824344142898e-05,
+      "loss": 0.1819,
+      "step": 2913
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.620113437193222e-05,
+      "loss": 0.3657,
+      "step": 2914
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6198443670456968e-05,
+      "loss": 0.1757,
+      "step": 2915
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6195752240033576e-05,
+      "loss": 0.1829,
+      "step": 2916
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6193060080978552e-05,
+      "loss": 0.2077,
+      "step": 2917
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.619036719360851e-05,
+      "loss": 0.1495,
+      "step": 2918
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6187673578240134e-05,
+      "loss": 0.199,
+      "step": 2919
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6184979235190198e-05,
+      "loss": 0.1922,
+      "step": 2920
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6182284164775558e-05,
+      "loss": 0.2029,
+      "step": 2921
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6179588367313165e-05,
+      "loss": 0.2268,
+      "step": 2922
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6176891843120044e-05,
+      "loss": 0.1154,
+      "step": 2923
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6174194592513318e-05,
+      "loss": 0.152,
+      "step": 2924
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6171496615810183e-05,
+      "loss": 0.2345,
+      "step": 2925
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6168797913327928e-05,
+      "loss": 0.1093,
+      "step": 2926
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.616609848538392e-05,
+      "loss": 0.1968,
+      "step": 2927
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6163398332295633e-05,
+      "loss": 0.1793,
+      "step": 2928
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.616069745438059e-05,
+      "loss": 0.1893,
+      "step": 2929
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6157995851956428e-05,
+      "loss": 0.1795,
+      "step": 2930
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.615529352534086e-05,
+      "loss": 0.1679,
+      "step": 2931
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6152590474851682e-05,
+      "loss": 0.1207,
+      "step": 2932
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6149886700806784e-05,
+      "loss": 0.2157,
+      "step": 2933
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.614718220352413e-05,
+      "loss": 0.1163,
+      "step": 2934
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6144476983321774e-05,
+      "loss": 0.168,
+      "step": 2935
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6141771040517854e-05,
+      "loss": 0.2037,
+      "step": 2936
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6139064375430594e-05,
+      "loss": 0.1842,
+      "step": 2937
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6136356988378307e-05,
+      "loss": 0.1546,
+      "step": 2938
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.613364887967938e-05,
+      "loss": 0.2323,
+      "step": 2939
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6130940049652297e-05,
+      "loss": 0.1375,
+      "step": 2940
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.612823049861562e-05,
+      "loss": 0.1705,
+      "step": 2941
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6125520226887988e-05,
+      "loss": 0.1641,
+      "step": 2942
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6122809234788147e-05,
+      "loss": 0.1689,
+      "step": 2943
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.612009752263491e-05,
+      "loss": 0.1459,
+      "step": 2944
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.611738509074718e-05,
+      "loss": 0.1261,
+      "step": 2945
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6114671939443938e-05,
+      "loss": 0.2785,
+      "step": 2946
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6111958069044264e-05,
+      "loss": 0.0779,
+      "step": 2947
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6109243479867308e-05,
+      "loss": 0.1729,
+      "step": 2948
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.610652817223231e-05,
+      "loss": 0.1292,
+      "step": 2949
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6103812146458603e-05,
+      "loss": 0.2215,
+      "step": 2950
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6101095402865587e-05,
+      "loss": 0.2002,
+      "step": 2951
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.609837794177276e-05,
+      "loss": 0.174,
+      "step": 2952
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6095659763499703e-05,
+      "loss": 0.1325,
+      "step": 2953
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6092940868366078e-05,
+      "loss": 0.1433,
+      "step": 2954
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.609022125669162e-05,
+      "loss": 0.3045,
+      "step": 2955
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6087500928796185e-05,
+      "loss": 0.1847,
+      "step": 2956
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6084779884999662e-05,
+      "loss": 0.1392,
+      "step": 2957
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.608205812562207e-05,
+      "loss": 0.1748,
+      "step": 2958
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6079335650983478e-05,
+      "loss": 0.1391,
+      "step": 2959
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6076612461404065e-05,
+      "loss": 0.2219,
+      "step": 2960
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.607388855720408e-05,
+      "loss": 0.1112,
+      "step": 2961
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.607116393870386e-05,
+      "loss": 0.1824,
+      "step": 2962
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6068438606223816e-05,
+      "loss": 0.2309,
+      "step": 2963
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6065712560084467e-05,
+      "loss": 0.1636,
+      "step": 2964
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6062985800606384e-05,
+      "loss": 0.1814,
+      "step": 2965
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6060258328110255e-05,
+      "loss": 0.1063,
+      "step": 2966
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6057530142916825e-05,
+      "loss": 0.2481,
+      "step": 2967
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6054801245346942e-05,
+      "loss": 0.1218,
+      "step": 2968
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6052071635721523e-05,
+      "loss": 0.2633,
+      "step": 2969
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.604934131436157e-05,
+      "loss": 0.1865,
+      "step": 2970
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.604661028158819e-05,
+      "loss": 0.2076,
+      "step": 2971
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6043878537722543e-05,
+      "loss": 0.1817,
+      "step": 2972
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.604114608308589e-05,
+      "loss": 0.2601,
+      "step": 2973
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.603841291799957e-05,
+      "loss": 0.2231,
+      "step": 2974
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6035679042785022e-05,
+      "loss": 0.0655,
+      "step": 2975
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.603294445776374e-05,
+      "loss": 0.2121,
+      "step": 2976
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6030209163257322e-05,
+      "loss": 0.16,
+      "step": 2977
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6027473159587443e-05,
+      "loss": 0.1254,
+      "step": 2978
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.602473644707586e-05,
+      "loss": 0.215,
+      "step": 2979
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6021999026044417e-05,
+      "loss": 0.1935,
+      "step": 2980
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.601926089681504e-05,
+      "loss": 0.1708,
+      "step": 2981
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6016522059709736e-05,
+      "loss": 0.2656,
+      "step": 2982
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.60137825150506e-05,
+      "loss": 0.1617,
+      "step": 2983
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6011042263159806e-05,
+      "loss": 0.1926,
+      "step": 2984
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.600830130435961e-05,
+      "loss": 0.2843,
+      "step": 2985
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6005559638972357e-05,
+      "loss": 0.1574,
+      "step": 2986
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6002817267320473e-05,
+      "loss": 0.1968,
+      "step": 2987
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6000074189726468e-05,
+      "loss": 0.219,
+      "step": 2988
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5997330406512923e-05,
+      "loss": 0.139,
+      "step": 2989
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5994585918002517e-05,
+      "loss": 0.2691,
+      "step": 2990
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5991840724518008e-05,
+      "loss": 0.1262,
+      "step": 2991
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5989094826382236e-05,
+      "loss": 0.1433,
+      "step": 2992
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5986348223918124e-05,
+      "loss": 0.1383,
+      "step": 2993
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5983600917448674e-05,
+      "loss": 0.0942,
+      "step": 2994
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5980852907296977e-05,
+      "loss": 0.2143,
+      "step": 2995
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5978104193786203e-05,
+      "loss": 0.1263,
+      "step": 2996
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5975354777239608e-05,
+      "loss": 0.0716,
+      "step": 2997
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5972604657980524e-05,
+      "loss": 0.1745,
+      "step": 2998
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5969853836332372e-05,
+      "loss": 0.1531,
+      "step": 2999
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5967102312618655e-05,
+      "loss": 0.1805,
+      "step": 3000
+    }
+  ],
+  "max_steps": 9445,
+  "num_train_epochs": 5,
+  "total_flos": 1.0744401458783846e+17,
+  "trial_name": null,
+  "trial_params": null
+}