diff --git "a/checkpoint-1000/trainer_state.json" "b/checkpoint-1000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-1000/trainer_state.json"
@@ -0,0 +1,6016 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.5293105729786952,
+  "global_step": 1000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 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
+    }
+  ],
+  "max_steps": 9445,
+  "num_train_epochs": 5,
+  "total_flos": 3.577437208392499e+16,
+  "trial_name": null,
+  "trial_params": null
+}