diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,42444 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 7069,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.389671361502348e-08,
+      "loss": 1.3185,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8779342723004696e-07,
+      "loss": 1.2982,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8169014084507043e-07,
+      "loss": 1.354,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.755868544600939e-07,
+      "loss": 1.3503,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.694835680751174e-07,
+      "loss": 0.8088,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.633802816901409e-07,
+      "loss": 1.3387,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.572769953051644e-07,
+      "loss": 1.3362,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.511737089201878e-07,
+      "loss": 0.7628,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.450704225352114e-07,
+      "loss": 1.3339,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.389671361502348e-07,
+      "loss": 1.3338,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0328638497652584e-06,
+      "loss": 1.3615,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1267605633802817e-06,
+      "loss": 0.7974,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2206572769953053e-06,
+      "loss": 1.3163,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3145539906103288e-06,
+      "loss": 0.7841,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4084507042253523e-06,
+      "loss": 1.2922,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5023474178403756e-06,
+      "loss": 1.2456,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5962441314553992e-06,
+      "loss": 1.2032,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6901408450704227e-06,
+      "loss": 1.242,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.784037558685446e-06,
+      "loss": 1.2097,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8779342723004696e-06,
+      "loss": 1.23,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.971830985915493e-06,
+      "loss": 1.2593,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.065727699530517e-06,
+      "loss": 1.1742,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.15962441314554e-06,
+      "loss": 1.1549,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2535211267605635e-06,
+      "loss": 1.1769,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.347417840375587e-06,
+      "loss": 1.1533,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4413145539906105e-06,
+      "loss": 1.1735,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.535211267605634e-06,
+      "loss": 1.1737,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6291079812206576e-06,
+      "loss": 1.1864,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.723004694835681e-06,
+      "loss": 1.1407,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8169014084507046e-06,
+      "loss": 0.7287,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.910798122065728e-06,
+      "loss": 1.1252,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0046948356807513e-06,
+      "loss": 1.1508,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0985915492957746e-06,
+      "loss": 1.117,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1924882629107983e-06,
+      "loss": 0.6824,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.286384976525822e-06,
+      "loss": 1.1257,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3802816901408454e-06,
+      "loss": 1.111,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.474178403755869e-06,
+      "loss": 1.1193,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.568075117370892e-06,
+      "loss": 0.6805,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6619718309859158e-06,
+      "loss": 1.0995,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.755868544600939e-06,
+      "loss": 1.0791,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.849765258215963e-06,
+      "loss": 0.7185,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.943661971830986e-06,
+      "loss": 1.0624,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0375586854460095e-06,
+      "loss": 1.1117,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.131455399061034e-06,
+      "loss": 1.0763,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.225352112676057e-06,
+      "loss": 1.0979,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.31924882629108e-06,
+      "loss": 1.0449,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.413145539906104e-06,
+      "loss": 1.0269,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.507042253521127e-06,
+      "loss": 1.0482,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.60093896713615e-06,
+      "loss": 0.7403,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.694835680751174e-06,
+      "loss": 1.024,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.788732394366197e-06,
+      "loss": 1.0384,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.882629107981221e-06,
+      "loss": 1.0466,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.976525821596244e-06,
+      "loss": 1.0171,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.070422535211268e-06,
+      "loss": 1.0001,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.164319248826292e-06,
+      "loss": 1.0508,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.258215962441315e-06,
+      "loss": 0.6562,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.352112676056338e-06,
+      "loss": 1.0217,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.446009389671362e-06,
+      "loss": 0.9937,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.539906103286385e-06,
+      "loss": 1.035,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.633802816901409e-06,
+      "loss": 0.6444,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.727699530516433e-06,
+      "loss": 0.9829,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.821596244131456e-06,
+      "loss": 0.995,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.915492957746479e-06,
+      "loss": 0.9662,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.0093896713615026e-06,
+      "loss": 0.9852,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.103286384976527e-06,
+      "loss": 1.0078,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.197183098591549e-06,
+      "loss": 0.693,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.291079812206573e-06,
+      "loss": 0.994,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.384976525821597e-06,
+      "loss": 0.9803,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.478873239436621e-06,
+      "loss": 0.9892,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.572769953051644e-06,
+      "loss": 0.6982,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.6556,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.760563380281691e-06,
+      "loss": 0.9781,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.854460093896714e-06,
+      "loss": 0.9998,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.948356807511738e-06,
+      "loss": 0.9627,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.042253521126761e-06,
+      "loss": 0.9633,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.136150234741784e-06,
+      "loss": 0.9964,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.230046948356808e-06,
+      "loss": 0.9541,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.3239436619718316e-06,
+      "loss": 0.9723,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.417840375586856e-06,
+      "loss": 0.9255,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.511737089201878e-06,
+      "loss": 0.6734,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.6056338028169015e-06,
+      "loss": 0.9801,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.699530516431926e-06,
+      "loss": 0.9803,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.79342723004695e-06,
+      "loss": 0.9415,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.887323943661972e-06,
+      "loss": 0.9443,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.981220657276996e-06,
+      "loss": 0.9443,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.075117370892019e-06,
+      "loss": 0.9399,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.169014084507043e-06,
+      "loss": 0.9524,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.262910798122067e-06,
+      "loss": 0.969,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.35680751173709e-06,
+      "loss": 0.917,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.450704225352114e-06,
+      "loss": 0.9347,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.544600938967136e-06,
+      "loss": 0.942,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.63849765258216e-06,
+      "loss": 0.6827,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.732394366197183e-06,
+      "loss": 0.9422,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.826291079812207e-06,
+      "loss": 0.9312,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.920187793427231e-06,
+      "loss": 0.9787,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.014084507042254e-06,
+      "loss": 0.9549,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.107981220657278e-06,
+      "loss": 0.9422,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.2018779342723e-06,
+      "loss": 0.6652,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.295774647887325e-06,
+      "loss": 0.919,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.389671361502349e-06,
+      "loss": 0.9277,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.483568075117371e-06,
+      "loss": 0.9131,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.577464788732394e-06,
+      "loss": 0.9184,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.671361502347418e-06,
+      "loss": 0.9229,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.765258215962442e-06,
+      "loss": 0.6047,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.859154929577466e-06,
+      "loss": 0.9105,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.953051643192489e-06,
+      "loss": 0.9569,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0046948356807513e-05,
+      "loss": 0.8936,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0140845070422535e-05,
+      "loss": 0.9491,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0234741784037558e-05,
+      "loss": 0.9533,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0328638497652584e-05,
+      "loss": 0.9436,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0422535211267606e-05,
+      "loss": 0.9406,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.051643192488263e-05,
+      "loss": 0.9357,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0610328638497653e-05,
+      "loss": 0.9243,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0704225352112675e-05,
+      "loss": 0.9382,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0798122065727701e-05,
+      "loss": 0.6293,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0892018779342724e-05,
+      "loss": 0.93,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0985915492957748e-05,
+      "loss": 0.9145,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.107981220657277e-05,
+      "loss": 0.6059,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1173708920187793e-05,
+      "loss": 0.9495,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1267605633802819e-05,
+      "loss": 0.934,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1361502347417841e-05,
+      "loss": 0.6188,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1455399061032865e-05,
+      "loss": 0.6399,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1549295774647888e-05,
+      "loss": 0.9333,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1643192488262912e-05,
+      "loss": 0.9552,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1737089201877936e-05,
+      "loss": 0.652,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1830985915492958e-05,
+      "loss": 0.9125,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1924882629107981e-05,
+      "loss": 0.8726,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2018779342723005e-05,
+      "loss": 0.9488,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.211267605633803e-05,
+      "loss": 0.8997,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2206572769953053e-05,
+      "loss": 0.9063,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2300469483568076e-05,
+      "loss": 0.9117,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2394366197183098e-05,
+      "loss": 0.9044,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2488262910798124e-05,
+      "loss": 0.8972,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2582159624413147e-05,
+      "loss": 0.945,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2676056338028171e-05,
+      "loss": 0.6639,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2769953051643193e-05,
+      "loss": 0.9588,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2863849765258216e-05,
+      "loss": 0.9099,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2957746478873242e-05,
+      "loss": 0.8848,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3051643192488264e-05,
+      "loss": 0.8804,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3145539906103288e-05,
+      "loss": 0.6401,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.323943661971831e-05,
+      "loss": 0.9077,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.6286,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.342723004694836e-05,
+      "loss": 0.912,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3521126760563382e-05,
+      "loss": 0.9381,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3615023474178406e-05,
+      "loss": 0.9077,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3708920187793428e-05,
+      "loss": 0.8982,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.380281690140845e-05,
+      "loss": 0.9266,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3896713615023477e-05,
+      "loss": 0.9058,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3990610328638499e-05,
+      "loss": 0.9126,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4084507042253522e-05,
+      "loss": 0.6445,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4178403755868546e-05,
+      "loss": 0.905,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4272300469483568e-05,
+      "loss": 0.9457,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4366197183098594e-05,
+      "loss": 0.9084,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4460093896713617e-05,
+      "loss": 0.9347,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4553990610328639e-05,
+      "loss": 0.9013,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4647887323943663e-05,
+      "loss": 0.8974,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4741784037558686e-05,
+      "loss": 0.9251,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4835680751173711e-05,
+      "loss": 0.8714,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4929577464788734e-05,
+      "loss": 0.8824,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5023474178403756e-05,
+      "loss": 0.6212,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.511737089201878e-05,
+      "loss": 0.6185,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5211267605633803e-05,
+      "loss": 0.6435,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5305164319248827e-05,
+      "loss": 0.8923,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.539906103286385e-05,
+      "loss": 0.9139,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5492957746478872e-05,
+      "loss": 0.892,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.55868544600939e-05,
+      "loss": 0.8549,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.568075117370892e-05,
+      "loss": 0.8982,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5774647887323945e-05,
+      "loss": 0.9028,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.586854460093897e-05,
+      "loss": 0.8886,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5962441314553993e-05,
+      "loss": 0.618,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6056338028169017e-05,
+      "loss": 0.9196,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6150234741784038e-05,
+      "loss": 0.9074,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6244131455399062e-05,
+      "loss": 0.8649,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6338028169014086e-05,
+      "loss": 0.9566,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.643192488262911e-05,
+      "loss": 0.9114,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6525821596244135e-05,
+      "loss": 0.6389,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6619718309859155e-05,
+      "loss": 0.8964,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.671361502347418e-05,
+      "loss": 0.8893,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6807511737089204e-05,
+      "loss": 0.8789,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6901408450704228e-05,
+      "loss": 0.8829,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6995305164319252e-05,
+      "loss": 0.8945,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7089201877934273e-05,
+      "loss": 0.9344,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7183098591549297e-05,
+      "loss": 0.9386,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.727699530516432e-05,
+      "loss": 0.6468,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7370892018779345e-05,
+      "loss": 0.8666,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7464788732394366e-05,
+      "loss": 0.8831,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.755868544600939e-05,
+      "loss": 0.8826,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7652582159624414e-05,
+      "loss": 0.8721,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.774647887323944e-05,
+      "loss": 0.6343,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7840375586854463e-05,
+      "loss": 0.6245,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7934272300469484e-05,
+      "loss": 0.8828,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8028169014084508e-05,
+      "loss": 0.9188,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8122065727699532e-05,
+      "loss": 0.8998,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8215962441314556e-05,
+      "loss": 0.9068,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.830985915492958e-05,
+      "loss": 0.8593,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.84037558685446e-05,
+      "loss": 0.6512,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8497652582159625e-05,
+      "loss": 0.9071,
+      "step": 197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.859154929577465e-05,
+      "loss": 0.5947,
+      "step": 198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8685446009389673e-05,
+      "loss": 0.665,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8779342723004698e-05,
+      "loss": 0.8712,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.887323943661972e-05,
+      "loss": 0.9026,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8967136150234743e-05,
+      "loss": 0.9173,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9061032863849767e-05,
+      "loss": 0.907,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9154929577464788e-05,
+      "loss": 0.89,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9248826291079815e-05,
+      "loss": 0.8987,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9342723004694836e-05,
+      "loss": 0.9289,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.943661971830986e-05,
+      "loss": 0.8653,
+      "step": 207
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9530516431924884e-05,
+      "loss": 0.8524,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.962441314553991e-05,
+      "loss": 0.8723,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9718309859154933e-05,
+      "loss": 0.6319,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9812206572769953e-05,
+      "loss": 0.9011,
+      "step": 211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9906103286384977e-05,
+      "loss": 0.9088,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.8495,
+      "step": 213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999998950147895e-05,
+      "loss": 0.8692,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999995800591792e-05,
+      "loss": 0.8494,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999990551332357e-05,
+      "loss": 0.6052,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999983202370694e-05,
+      "loss": 0.8355,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999973753708342e-05,
+      "loss": 0.8652,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999962205347286e-05,
+      "loss": 0.8872,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999948557289948e-05,
+      "loss": 0.8864,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999932809539202e-05,
+      "loss": 0.8696,
+      "step": 221
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999914962098348e-05,
+      "loss": 0.888,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999895014971136e-05,
+      "loss": 0.887,
+      "step": 223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999987296816175e-05,
+      "loss": 0.8756,
+      "step": 224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999848821674823e-05,
+      "loss": 0.8992,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999822575515426e-05,
+      "loss": 0.5803,
+      "step": 226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999979422968907e-05,
+      "loss": 0.8584,
+      "step": 227
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99997637842017e-05,
+      "loss": 0.864,
+      "step": 228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999731239059717e-05,
+      "loss": 0.8643,
+      "step": 229
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999969659426995e-05,
+      "loss": 0.9139,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999659849839674e-05,
+      "loss": 0.8715,
+      "step": 231
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999621005776607e-05,
+      "loss": 0.8215,
+      "step": 232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999580062088904e-05,
+      "loss": 0.8817,
+      "step": 233
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999953701878516e-05,
+      "loss": 0.8907,
+      "step": 234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999949187587441e-05,
+      "loss": 0.8838,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999444633366143e-05,
+      "loss": 0.8386,
+      "step": 236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999395291270266e-05,
+      "loss": 0.6394,
+      "step": 237
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999934384959715e-05,
+      "loss": 0.6417,
+      "step": 238
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999290308357587e-05,
+      "loss": 0.6795,
+      "step": 239
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999234667562827e-05,
+      "loss": 0.9457,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999917692722455e-05,
+      "loss": 0.8782,
+      "step": 241
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999117087354877e-05,
+      "loss": 0.9041,
+      "step": 242
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999055147966376e-05,
+      "loss": 0.8734,
+      "step": 243
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999899110907205e-05,
+      "loss": 0.6132,
+      "step": 244
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999892497068535e-05,
+      "loss": 0.8972,
+      "step": 245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999885673282016e-05,
+      "loss": 0.8747,
+      "step": 246
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998786395490804e-05,
+      "loss": 0.889,
+      "step": 247
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998713958712054e-05,
+      "loss": 0.9041,
+      "step": 248
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998639422499125e-05,
+      "loss": 0.864,
+      "step": 249
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999856278686766e-05,
+      "loss": 0.9227,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998484051833752e-05,
+      "loss": 0.8449,
+      "step": 251
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998403217413936e-05,
+      "loss": 0.9076,
+      "step": 252
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999832028362518e-05,
+      "loss": 0.8686,
+      "step": 253
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99982352504849e-05,
+      "loss": 0.943,
+      "step": 254
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998148118010953e-05,
+      "loss": 0.9065,
+      "step": 255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999805888622163e-05,
+      "loss": 0.9149,
+      "step": 256
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999796755513567e-05,
+      "loss": 0.8862,
+      "step": 257
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997874124772245e-05,
+      "loss": 0.8835,
+      "step": 258
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999777859515098e-05,
+      "loss": 0.8783,
+      "step": 259
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999768096629193e-05,
+      "loss": 0.8794,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997581238215593e-05,
+      "loss": 0.8661,
+      "step": 261
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999747941094291e-05,
+      "loss": 0.8781,
+      "step": 262
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999737548449526e-05,
+      "loss": 0.9124,
+      "step": 263
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999726945889447e-05,
+      "loss": 0.8417,
+      "step": 264
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997161334162794e-05,
+      "loss": 0.8322,
+      "step": 265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999705111032294e-05,
+      "loss": 0.909,
+      "step": 266
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996938787398052e-05,
+      "loss": 0.5964,
+      "step": 267
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996824365411715e-05,
+      "loss": 0.8458,
+      "step": 268
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999670784438795e-05,
+      "loss": 0.8769,
+      "step": 269
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999658922435123e-05,
+      "loss": 0.6552,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996468505326453e-05,
+      "loss": 0.8699,
+      "step": 271
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996345687338973e-05,
+      "loss": 0.8531,
+      "step": 272
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996220770414573e-05,
+      "loss": 0.8728,
+      "step": 273
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996093754579488e-05,
+      "loss": 0.8844,
+      "step": 274
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995964639860385e-05,
+      "loss": 0.8972,
+      "step": 275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995833426284373e-05,
+      "loss": 0.6254,
+      "step": 276
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995700113879e-05,
+      "loss": 0.8896,
+      "step": 277
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999556470267227e-05,
+      "loss": 0.8705,
+      "step": 278
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995427192692598e-05,
+      "loss": 0.885,
+      "step": 279
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999528758396887e-05,
+      "loss": 0.8727,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995145876530393e-05,
+      "loss": 0.8621,
+      "step": 281
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995002070406927e-05,
+      "loss": 0.8794,
+      "step": 282
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994856165628657e-05,
+      "loss": 0.8605,
+      "step": 283
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994708162226232e-05,
+      "loss": 0.8345,
+      "step": 284
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994558060230722e-05,
+      "loss": 0.6256,
+      "step": 285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999440585967364e-05,
+      "loss": 0.8159,
+      "step": 286
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999425156058695e-05,
+      "loss": 0.6082,
+      "step": 287
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994095163003046e-05,
+      "loss": 0.8853,
+      "step": 288
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993936666954773e-05,
+      "loss": 0.887,
+      "step": 289
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99937760724754e-05,
+      "loss": 0.8679,
+      "step": 290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993613379598656e-05,
+      "loss": 0.8455,
+      "step": 291
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99934485883587e-05,
+      "loss": 0.6792,
+      "step": 292
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993281698790132e-05,
+      "loss": 0.8575,
+      "step": 293
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999311271092799e-05,
+      "loss": 0.8955,
+      "step": 294
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992941624807763e-05,
+      "loss": 0.8643,
+      "step": 295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999276844046537e-05,
+      "loss": 0.8553,
+      "step": 296
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999259315793718e-05,
+      "loss": 0.8597,
+      "step": 297
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999241577725999e-05,
+      "loss": 0.9007,
+      "step": 298
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999223629847105e-05,
+      "loss": 0.6156,
+      "step": 299
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992054721608037e-05,
+      "loss": 0.6364,
+      "step": 300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999187104670909e-05,
+      "loss": 0.6362,
+      "step": 301
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991685273812768e-05,
+      "loss": 0.8867,
+      "step": 302
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991497402958074e-05,
+      "loss": 0.8531,
+      "step": 303
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991307434184463e-05,
+      "loss": 0.8616,
+      "step": 304
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991115367531814e-05,
+      "loss": 0.8398,
+      "step": 305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990921203040465e-05,
+      "loss": 0.8599,
+      "step": 306
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990724940751177e-05,
+      "loss": 0.8935,
+      "step": 307
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990526580705167e-05,
+      "loss": 0.8363,
+      "step": 308
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990326122944076e-05,
+      "loss": 0.6332,
+      "step": 309
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999012356751e-05,
+      "loss": 0.9079,
+      "step": 310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998991891444547e-05,
+      "loss": 0.8802,
+      "step": 311
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989712163793453e-05,
+      "loss": 0.8735,
+      "step": 312
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998950331559736e-05,
+      "loss": 0.6684,
+      "step": 313
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989292369901053e-05,
+      "loss": 0.8784,
+      "step": 314
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998907932674881e-05,
+      "loss": 0.8617,
+      "step": 315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988864186185376e-05,
+      "loss": 0.8768,
+      "step": 316
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988646948255915e-05,
+      "loss": 0.6308,
+      "step": 317
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998842761300605e-05,
+      "loss": 0.8543,
+      "step": 318
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988206180481822e-05,
+      "loss": 0.895,
+      "step": 319
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987982650729738e-05,
+      "loss": 0.8776,
+      "step": 320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987757023796724e-05,
+      "loss": 0.9022,
+      "step": 321
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998752929973016e-05,
+      "loss": 0.8443,
+      "step": 322
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987299478577862e-05,
+      "loss": 0.6042,
+      "step": 323
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998706756038808e-05,
+      "loss": 0.8775,
+      "step": 324
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998683354520951e-05,
+      "loss": 0.8348,
+      "step": 325
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986597433091298e-05,
+      "loss": 0.8602,
+      "step": 326
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998635922408301e-05,
+      "loss": 0.8974,
+      "step": 327
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986118918234673e-05,
+      "loss": 0.7999,
+      "step": 328
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985876515596733e-05,
+      "loss": 0.8501,
+      "step": 329
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985632016220093e-05,
+      "loss": 0.8577,
+      "step": 330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998538542015609e-05,
+      "loss": 0.9,
+      "step": 331
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99851367274565e-05,
+      "loss": 0.8806,
+      "step": 332
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984885938173545e-05,
+      "loss": 0.8036,
+      "step": 333
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984633052359882e-05,
+      "loss": 0.8593,
+      "step": 334
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984378070068606e-05,
+      "loss": 0.8488,
+      "step": 335
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984120991353262e-05,
+      "loss": 0.8461,
+      "step": 336
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983861816267823e-05,
+      "loss": 0.8685,
+      "step": 337
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983600544866713e-05,
+      "loss": 0.8546,
+      "step": 338
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983337177204788e-05,
+      "loss": 0.8238,
+      "step": 339
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998307171333735e-05,
+      "loss": 0.8759,
+      "step": 340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998280415332013e-05,
+      "loss": 0.8343,
+      "step": 341
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982534497209323e-05,
+      "loss": 0.6494,
+      "step": 342
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982262745061537e-05,
+      "loss": 0.8302,
+      "step": 343
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981988896933833e-05,
+      "loss": 0.8405,
+      "step": 344
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998171295288372e-05,
+      "loss": 0.8253,
+      "step": 345
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981434912969127e-05,
+      "loss": 0.8225,
+      "step": 346
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981154777248436e-05,
+      "loss": 0.8833,
+      "step": 347
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980872545780476e-05,
+      "loss": 0.8646,
+      "step": 348
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980588218624496e-05,
+      "loss": 0.879,
+      "step": 349
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980301795840202e-05,
+      "loss": 0.8394,
+      "step": 350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980013277487737e-05,
+      "loss": 0.8142,
+      "step": 351
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979722663627673e-05,
+      "loss": 0.6381,
+      "step": 352
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997942995432104e-05,
+      "loss": 0.6383,
+      "step": 353
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997913514962929e-05,
+      "loss": 0.6342,
+      "step": 354
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997883824961433e-05,
+      "loss": 0.88,
+      "step": 355
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978539254338497e-05,
+      "loss": 0.8438,
+      "step": 356
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978238163864567e-05,
+      "loss": 0.8281,
+      "step": 357
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997793497825577e-05,
+      "loss": 0.8666,
+      "step": 358
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977629697575755e-05,
+      "loss": 0.8648,
+      "step": 359
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997732232188863e-05,
+      "loss": 0.8473,
+      "step": 360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997701285125893e-05,
+      "loss": 0.8401,
+      "step": 361
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997670128575164e-05,
+      "loss": 0.8845,
+      "step": 362
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976387625432176e-05,
+      "loss": 0.8716,
+      "step": 363
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976071870366395e-05,
+      "loss": 0.8811,
+      "step": 364
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99757540206206e-05,
+      "loss": 0.8336,
+      "step": 365
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975434076261532e-05,
+      "loss": 0.8404,
+      "step": 366
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975112037356363e-05,
+      "loss": 0.8533,
+      "step": 367
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974787903972716e-05,
+      "loss": 0.8914,
+      "step": 368
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974461676178653e-05,
+      "loss": 0.683,
+      "step": 369
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997413335404266e-05,
+      "loss": 0.8597,
+      "step": 370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997380293763369e-05,
+      "loss": 0.8401,
+      "step": 371
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997347042702111e-05,
+      "loss": 0.8705,
+      "step": 372
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997313582227474e-05,
+      "loss": 0.8178,
+      "step": 373
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997279912346484e-05,
+      "loss": 0.8437,
+      "step": 374
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997246033066211e-05,
+      "loss": 0.8438,
+      "step": 375
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972119443937676e-05,
+      "loss": 0.8378,
+      "step": 376
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997177646336312e-05,
+      "loss": 0.8393,
+      "step": 377
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971431389010457e-05,
+      "loss": 0.6436,
+      "step": 378
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997108422095214e-05,
+      "loss": 0.8298,
+      "step": 379
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970734959261074e-05,
+      "loss": 0.8341,
+      "step": 380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997038360401058e-05,
+      "loss": 0.8823,
+      "step": 381
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970030155274443e-05,
+      "loss": 0.8892,
+      "step": 382
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969674613126867e-05,
+      "loss": 0.8272,
+      "step": 383
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969316977642513e-05,
+      "loss": 0.8162,
+      "step": 384
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968957248896475e-05,
+      "loss": 0.8119,
+      "step": 385
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968595426964277e-05,
+      "loss": 0.6517,
+      "step": 386
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99682315119219e-05,
+      "loss": 0.6245,
+      "step": 387
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996786550384575e-05,
+      "loss": 0.8031,
+      "step": 388
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967497402812678e-05,
+      "loss": 0.8615,
+      "step": 389
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967127208899976e-05,
+      "loss": 0.8669,
+      "step": 390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966754922185374e-05,
+      "loss": 0.8587,
+      "step": 391
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966380542747038e-05,
+      "loss": 0.8648,
+      "step": 392
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966004070663584e-05,
+      "loss": 0.8499,
+      "step": 393
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996562550601405e-05,
+      "loss": 0.8669,
+      "step": 394
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996524484887793e-05,
+      "loss": 0.8697,
+      "step": 395
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996486209933515e-05,
+      "loss": 0.8411,
+      "step": 396
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964477257466075e-05,
+      "loss": 0.8153,
+      "step": 397
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996409032335151e-05,
+      "loss": 0.6243,
+      "step": 398
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963701297072702e-05,
+      "loss": 0.8787,
+      "step": 399
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963310178711334e-05,
+      "loss": 0.8353,
+      "step": 400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962916968349524e-05,
+      "loss": 0.8313,
+      "step": 401
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996252166606984e-05,
+      "loss": 0.8401,
+      "step": 402
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996212427195529e-05,
+      "loss": 0.6327,
+      "step": 403
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99617247860893e-05,
+      "loss": 0.8511,
+      "step": 404
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961323208555762e-05,
+      "loss": 0.8641,
+      "step": 405
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996091953943899e-05,
+      "loss": 0.8294,
+      "step": 406
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960513778823744e-05,
+      "loss": 0.8378,
+      "step": 407
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960105926795227e-05,
+      "loss": 0.8743,
+      "step": 408
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959695983439066e-05,
+      "loss": 0.8623,
+      "step": 409
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995928394884134e-05,
+      "loss": 0.885,
+      "step": 410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995886982308857e-05,
+      "loss": 0.6327,
+      "step": 411
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958453606267705e-05,
+      "loss": 0.8302,
+      "step": 412
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995803529846614e-05,
+      "loss": 0.8275,
+      "step": 413
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957614899771704e-05,
+      "loss": 0.8664,
+      "step": 414
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995719241027267e-05,
+      "loss": 0.8258,
+      "step": 415
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995676783005775e-05,
+      "loss": 0.8618,
+      "step": 416
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956341159216095e-05,
+      "loss": 0.8507,
+      "step": 417
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995591239783729e-05,
+      "loss": 0.7922,
+      "step": 418
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995548154601136e-05,
+      "loss": 0.8092,
+      "step": 419
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955048603828777e-05,
+      "loss": 0.8462,
+      "step": 420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954613571380443e-05,
+      "loss": 0.8221,
+      "step": 421
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99541764487577e-05,
+      "loss": 0.6452,
+      "step": 422
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953737236052334e-05,
+      "loss": 0.8967,
+      "step": 423
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953295933356566e-05,
+      "loss": 0.8264,
+      "step": 424
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952852540763054e-05,
+      "loss": 0.8207,
+      "step": 425
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99524070583649e-05,
+      "loss": 0.8675,
+      "step": 426
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951959486255644e-05,
+      "loss": 0.8286,
+      "step": 427
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951509824529255e-05,
+      "loss": 0.8411,
+      "step": 428
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995105807328016e-05,
+      "loss": 0.8341,
+      "step": 429
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950604232603202e-05,
+      "loss": 0.6017,
+      "step": 430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950148302593683e-05,
+      "loss": 0.7904,
+      "step": 431
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949690283347333e-05,
+      "loss": 0.8239,
+      "step": 432
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949230174960314e-05,
+      "loss": 0.8368,
+      "step": 433
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994876797752925e-05,
+      "loss": 0.8348,
+      "step": 434
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948303691151175e-05,
+      "loss": 0.8102,
+      "step": 435
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947837315923584e-05,
+      "loss": 0.8267,
+      "step": 436
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947368851944397e-05,
+      "loss": 0.8103,
+      "step": 437
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946898299311982e-05,
+      "loss": 0.856,
+      "step": 438
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994642565812514e-05,
+      "loss": 0.8534,
+      "step": 439
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994595092848311e-05,
+      "loss": 0.5779,
+      "step": 440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945474110485568e-05,
+      "loss": 0.8469,
+      "step": 441
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944995204232636e-05,
+      "loss": 0.8488,
+      "step": 442
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944514209824874e-05,
+      "loss": 0.6421,
+      "step": 443
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944031127363268e-05,
+      "loss": 0.8532,
+      "step": 444
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943545956949258e-05,
+      "loss": 0.8578,
+      "step": 445
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943058698684712e-05,
+      "loss": 0.8607,
+      "step": 446
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942569352671942e-05,
+      "loss": 0.87,
+      "step": 447
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994207791901369e-05,
+      "loss": 0.6458,
+      "step": 448
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941584397813152e-05,
+      "loss": 0.8665,
+      "step": 449
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941088789173944e-05,
+      "loss": 0.8555,
+      "step": 450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940591093200135e-05,
+      "loss": 0.8594,
+      "step": 451
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940091309996225e-05,
+      "loss": 0.8381,
+      "step": 452
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939589439667154e-05,
+      "loss": 0.845,
+      "step": 453
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939085482318296e-05,
+      "loss": 0.8304,
+      "step": 454
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938579438055474e-05,
+      "loss": 0.8051,
+      "step": 455
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993807130698494e-05,
+      "loss": 0.81,
+      "step": 456
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993756108921338e-05,
+      "loss": 0.8578,
+      "step": 457
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937048784847932e-05,
+      "loss": 0.8277,
+      "step": 458
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936534393996163e-05,
+      "loss": 0.8446,
+      "step": 459
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993601791676608e-05,
+      "loss": 0.8196,
+      "step": 460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935499353266126e-05,
+      "loss": 0.6518,
+      "step": 461
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934978703605185e-05,
+      "loss": 0.8491,
+      "step": 462
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934455967892583e-05,
+      "loss": 0.8331,
+      "step": 463
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993393114623807e-05,
+      "loss": 0.5793,
+      "step": 464
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933404238751848e-05,
+      "loss": 0.87,
+      "step": 465
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932875245544553e-05,
+      "loss": 0.8274,
+      "step": 466
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932344166727254e-05,
+      "loss": 0.8361,
+      "step": 467
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931811002411467e-05,
+      "loss": 0.8686,
+      "step": 468
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931275752709135e-05,
+      "loss": 0.8508,
+      "step": 469
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993073841773265e-05,
+      "loss": 0.8669,
+      "step": 470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930198997594835e-05,
+      "loss": 0.6467,
+      "step": 471
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929657492408952e-05,
+      "loss": 0.6309,
+      "step": 472
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99291139022887e-05,
+      "loss": 0.85,
+      "step": 473
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992856822734821e-05,
+      "loss": 0.836,
+      "step": 474
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992802046770207e-05,
+      "loss": 0.8458,
+      "step": 475
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992747062346529e-05,
+      "loss": 0.8106,
+      "step": 476
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926918694753316e-05,
+      "loss": 0.689,
+      "step": 477
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926364681682044e-05,
+      "loss": 0.734,
+      "step": 478
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925808584367793e-05,
+      "loss": 0.862,
+      "step": 479
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925250402927332e-05,
+      "loss": 0.7934,
+      "step": 480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992469013747786e-05,
+      "loss": 0.6226,
+      "step": 481
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924127788137017e-05,
+      "loss": 0.6101,
+      "step": 482
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923563355022878e-05,
+      "loss": 0.6044,
+      "step": 483
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992299683825396e-05,
+      "loss": 0.6138,
+      "step": 484
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922428237949215e-05,
+      "loss": 0.8725,
+      "step": 485
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921857554228032e-05,
+      "loss": 0.8109,
+      "step": 486
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921284787210237e-05,
+      "loss": 0.8739,
+      "step": 487
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920709937016088e-05,
+      "loss": 0.8312,
+      "step": 488
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920133003766298e-05,
+      "loss": 0.6316,
+      "step": 489
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919553987581998e-05,
+      "loss": 0.8337,
+      "step": 490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918972888584765e-05,
+      "loss": 0.8749,
+      "step": 491
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991838970689662e-05,
+      "loss": 0.8459,
+      "step": 492
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917804442640003e-05,
+      "loss": 0.8046,
+      "step": 493
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917217095937807e-05,
+      "loss": 0.8504,
+      "step": 494
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916627666913362e-05,
+      "loss": 0.8143,
+      "step": 495
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991603615569042e-05,
+      "loss": 0.8199,
+      "step": 496
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915442562393194e-05,
+      "loss": 0.8008,
+      "step": 497
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991484688714631e-05,
+      "loss": 0.829,
+      "step": 498
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991424913007485e-05,
+      "loss": 0.8754,
+      "step": 499
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913649291304316e-05,
+      "loss": 0.8142,
+      "step": 500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913047370960667e-05,
+      "loss": 0.8335,
+      "step": 501
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991244336917028e-05,
+      "loss": 0.8325,
+      "step": 502
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911837286059986e-05,
+      "loss": 0.8153,
+      "step": 503
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911229121757035e-05,
+      "loss": 0.8072,
+      "step": 504
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991061887638913e-05,
+      "loss": 0.8176,
+      "step": 505
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910006550084403e-05,
+      "loss": 0.7864,
+      "step": 506
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909392142971424e-05,
+      "loss": 0.8594,
+      "step": 507
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99087756551792e-05,
+      "loss": 0.8741,
+      "step": 508
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908157086837173e-05,
+      "loss": 0.8401,
+      "step": 509
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990753643807523e-05,
+      "loss": 0.8223,
+      "step": 510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906913709023687e-05,
+      "loss": 0.8238,
+      "step": 511
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990628889981329e-05,
+      "loss": 0.852,
+      "step": 512
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905662010575245e-05,
+      "loss": 0.6141,
+      "step": 513
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905033041441172e-05,
+      "loss": 0.8461,
+      "step": 514
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904401992543133e-05,
+      "loss": 0.8441,
+      "step": 515
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903768864013633e-05,
+      "loss": 0.8483,
+      "step": 516
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903133655985614e-05,
+      "loss": 0.6708,
+      "step": 517
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990249636859245e-05,
+      "loss": 0.8482,
+      "step": 518
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901857001967945e-05,
+      "loss": 0.8382,
+      "step": 519
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901215556246353e-05,
+      "loss": 0.8341,
+      "step": 520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900572031562356e-05,
+      "loss": 0.6187,
+      "step": 521
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899926428051078e-05,
+      "loss": 0.8329,
+      "step": 522
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989927874584808e-05,
+      "loss": 0.8272,
+      "step": 523
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989862898508935e-05,
+      "loss": 0.8858,
+      "step": 524
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897977145911317e-05,
+      "loss": 0.8591,
+      "step": 525
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989732322845085e-05,
+      "loss": 0.8184,
+      "step": 526
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896667232845256e-05,
+      "loss": 0.871,
+      "step": 527
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896009159232276e-05,
+      "loss": 0.8158,
+      "step": 528
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895349007750077e-05,
+      "loss": 0.803,
+      "step": 529
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894686778537275e-05,
+      "loss": 0.623,
+      "step": 530
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9894022471732923e-05,
+      "loss": 0.8177,
+      "step": 531
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893356087476502e-05,
+      "loss": 0.8502,
+      "step": 532
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892687625907933e-05,
+      "loss": 0.8555,
+      "step": 533
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892017087167573e-05,
+      "loss": 0.6185,
+      "step": 534
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891344471396218e-05,
+      "loss": 0.8274,
+      "step": 535
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890669778735096e-05,
+      "loss": 0.8736,
+      "step": 536
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889993009325872e-05,
+      "loss": 0.8229,
+      "step": 537
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988931416331065e-05,
+      "loss": 0.8267,
+      "step": 538
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988863324083196e-05,
+      "loss": 0.8439,
+      "step": 539
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887950242032783e-05,
+      "loss": 0.648,
+      "step": 540
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887265167056526e-05,
+      "loss": 0.7957,
+      "step": 541
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886578016047036e-05,
+      "loss": 0.7893,
+      "step": 542
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885888789148592e-05,
+      "loss": 0.8315,
+      "step": 543
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885197486505912e-05,
+      "loss": 0.8541,
+      "step": 544
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988450410826415e-05,
+      "loss": 0.5982,
+      "step": 545
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883808654568897e-05,
+      "loss": 0.8408,
+      "step": 546
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883111125566173e-05,
+      "loss": 0.8321,
+      "step": 547
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988241152140244e-05,
+      "loss": 0.8538,
+      "step": 548
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881709842224593e-05,
+      "loss": 0.8313,
+      "step": 549
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881006088179973e-05,
+      "loss": 0.8142,
+      "step": 550
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880300259416336e-05,
+      "loss": 0.838,
+      "step": 551
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879592356081887e-05,
+      "loss": 0.6076,
+      "step": 552
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987888237832527e-05,
+      "loss": 0.8134,
+      "step": 553
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878170326295553e-05,
+      "loss": 0.8475,
+      "step": 554
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877456200142254e-05,
+      "loss": 0.6158,
+      "step": 555
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876740000015313e-05,
+      "loss": 0.6435,
+      "step": 556
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876021726065107e-05,
+      "loss": 0.8352,
+      "step": 557
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987530137844246e-05,
+      "loss": 0.8304,
+      "step": 558
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874578957298623e-05,
+      "loss": 0.8399,
+      "step": 559
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873854462785277e-05,
+      "loss": 0.8579,
+      "step": 560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987312789505455e-05,
+      "loss": 0.8168,
+      "step": 561
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872399254259e-05,
+      "loss": 0.821,
+      "step": 562
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871668540551613e-05,
+      "loss": 0.8295,
+      "step": 563
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870935754085822e-05,
+      "loss": 0.6154,
+      "step": 564
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870200895015492e-05,
+      "loss": 0.8047,
+      "step": 565
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986946396349492e-05,
+      "loss": 0.845,
+      "step": 566
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868724959678843e-05,
+      "loss": 0.8144,
+      "step": 567
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867983883722424e-05,
+      "loss": 0.7955,
+      "step": 568
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986724073578127e-05,
+      "loss": 0.8488,
+      "step": 569
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986649551601142e-05,
+      "loss": 0.8467,
+      "step": 570
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865748224569353e-05,
+      "loss": 0.8226,
+      "step": 571
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864998861611968e-05,
+      "loss": 0.652,
+      "step": 572
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864247427296616e-05,
+      "loss": 0.8392,
+      "step": 573
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863493921781075e-05,
+      "loss": 0.6286,
+      "step": 574
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862738345223554e-05,
+      "loss": 0.7923,
+      "step": 575
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861980697782712e-05,
+      "loss": 0.832,
+      "step": 576
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861220979617622e-05,
+      "loss": 0.6474,
+      "step": 577
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986045919088781e-05,
+      "loss": 0.8815,
+      "step": 578
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859695331753222e-05,
+      "loss": 0.7924,
+      "step": 579
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858929402374253e-05,
+      "loss": 0.8015,
+      "step": 580
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858161402911722e-05,
+      "loss": 0.6342,
+      "step": 581
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857391333526886e-05,
+      "loss": 0.6201,
+      "step": 582
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856619194381438e-05,
+      "loss": 0.8545,
+      "step": 583
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855844985637503e-05,
+      "loss": 0.6927,
+      "step": 584
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985506870745764e-05,
+      "loss": 0.8244,
+      "step": 585
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854290360004852e-05,
+      "loss": 0.8455,
+      "step": 586
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985350994344256e-05,
+      "loss": 0.5607,
+      "step": 587
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852727457934632e-05,
+      "loss": 0.8055,
+      "step": 588
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985194290364537e-05,
+      "loss": 0.584,
+      "step": 589
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851156280739503e-05,
+      "loss": 0.8459,
+      "step": 590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98503675893822e-05,
+      "loss": 0.8537,
+      "step": 591
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849576829739064e-05,
+      "loss": 0.7965,
+      "step": 592
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848784001976126e-05,
+      "loss": 0.8296,
+      "step": 593
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847989106259863e-05,
+      "loss": 0.6172,
+      "step": 594
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847192142757178e-05,
+      "loss": 0.8047,
+      "step": 595
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984639311163541e-05,
+      "loss": 0.8345,
+      "step": 596
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845592013062325e-05,
+      "loss": 0.5921,
+      "step": 597
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984478884720614e-05,
+      "loss": 0.7986,
+      "step": 598
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843983614235488e-05,
+      "loss": 0.8184,
+      "step": 599
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984317631431945e-05,
+      "loss": 0.842,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9842366947627534e-05,
+      "loss": 0.8049,
+      "step": 601
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984155551432968e-05,
+      "loss": 0.6108,
+      "step": 602
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840742014596264e-05,
+      "loss": 0.7908,
+      "step": 603
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839926448598106e-05,
+      "loss": 0.8227,
+      "step": 604
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839108816506442e-05,
+      "loss": 0.8285,
+      "step": 605
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838289118492953e-05,
+      "loss": 0.7792,
+      "step": 606
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983746735472975e-05,
+      "loss": 0.8058,
+      "step": 607
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983664352538938e-05,
+      "loss": 0.8107,
+      "step": 608
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835817630644826e-05,
+      "loss": 0.782,
+      "step": 609
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98349896706695e-05,
+      "loss": 0.8257,
+      "step": 610
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834159645637242e-05,
+      "loss": 0.8073,
+      "step": 611
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983332755572234e-05,
+      "loss": 0.8122,
+      "step": 612
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832493401099507e-05,
+      "loss": 0.6529,
+      "step": 613
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831657181943894e-05,
+      "loss": 0.8264,
+      "step": 614
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983081889843108e-05,
+      "loss": 0.8167,
+      "step": 615
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829978550737073e-05,
+      "loss": 0.8236,
+      "step": 616
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982913613903833e-05,
+      "loss": 0.8078,
+      "step": 617
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828291663511726e-05,
+      "loss": 0.7912,
+      "step": 618
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827445124334583e-05,
+      "loss": 0.8316,
+      "step": 619
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826596521684648e-05,
+      "loss": 0.8222,
+      "step": 620
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825745855740096e-05,
+      "loss": 0.7965,
+      "step": 621
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824893126679545e-05,
+      "loss": 0.8015,
+      "step": 622
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824038334682043e-05,
+      "loss": 0.8414,
+      "step": 623
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823181479927074e-05,
+      "loss": 0.7931,
+      "step": 624
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822322562594545e-05,
+      "loss": 0.6382,
+      "step": 625
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982146158286481e-05,
+      "loss": 0.8029,
+      "step": 626
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982059854091865e-05,
+      "loss": 0.7816,
+      "step": 627
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819733436937273e-05,
+      "loss": 0.7553,
+      "step": 628
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818866271102332e-05,
+      "loss": 0.8613,
+      "step": 629
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817997043595896e-05,
+      "loss": 0.8363,
+      "step": 630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817125754600486e-05,
+      "loss": 0.8136,
+      "step": 631
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981625240429904e-05,
+      "loss": 0.8316,
+      "step": 632
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815376992874946e-05,
+      "loss": 0.6398,
+      "step": 633
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814499520512004e-05,
+      "loss": 0.8477,
+      "step": 634
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981361998739446e-05,
+      "loss": 0.8,
+      "step": 635
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981273839370699e-05,
+      "loss": 0.8067,
+      "step": 636
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981185473963471e-05,
+      "loss": 0.8179,
+      "step": 637
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981096902536315e-05,
+      "loss": 0.7899,
+      "step": 638
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810081251078292e-05,
+      "loss": 0.8223,
+      "step": 639
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980919141696654e-05,
+      "loss": 0.8012,
+      "step": 640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808299523214727e-05,
+      "loss": 0.8065,
+      "step": 641
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980740557001013e-05,
+      "loss": 0.869,
+      "step": 642
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806509557540456e-05,
+      "loss": 0.8386,
+      "step": 643
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805611485993833e-05,
+      "loss": 0.8415,
+      "step": 644
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804711355558838e-05,
+      "loss": 0.8241,
+      "step": 645
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803809166424467e-05,
+      "loss": 0.7772,
+      "step": 646
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980290491878015e-05,
+      "loss": 0.8116,
+      "step": 647
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980199861281576e-05,
+      "loss": 0.7987,
+      "step": 648
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801090248721588e-05,
+      "loss": 0.8176,
+      "step": 649
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800179826688364e-05,
+      "loss": 0.7946,
+      "step": 650
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979926734690725e-05,
+      "loss": 0.8107,
+      "step": 651
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798352809569844e-05,
+      "loss": 0.8309,
+      "step": 652
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9797436214868172e-05,
+      "loss": 0.7802,
+      "step": 653
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796517562994684e-05,
+      "loss": 0.6254,
+      "step": 654
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795596854142277e-05,
+      "loss": 0.8107,
+      "step": 655
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979467408850427e-05,
+      "loss": 0.8135,
+      "step": 656
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9793749266274416e-05,
+      "loss": 0.8493,
+      "step": 657
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97928223876469e-05,
+      "loss": 0.8336,
+      "step": 658
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791893452816343e-05,
+      "loss": 0.8608,
+      "step": 659
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9790962461977787e-05,
+      "loss": 0.8086,
+      "step": 660
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979002941532672e-05,
+      "loss": 0.8106,
+      "step": 661
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789094313059048e-05,
+      "loss": 0.8237,
+      "step": 662
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978815715537112e-05,
+      "loss": 0.6581,
+      "step": 663
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9787217942459705e-05,
+      "loss": 0.8378,
+      "step": 664
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786276674522014e-05,
+      "loss": 0.8025,
+      "step": 665
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978533335175569e-05,
+      "loss": 0.7765,
+      "step": 666
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784387974358795e-05,
+      "loss": 0.7851,
+      "step": 667
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978344054252983e-05,
+      "loss": 0.8422,
+      "step": 668
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782491056467737e-05,
+      "loss": 0.8306,
+      "step": 669
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9781539516371868e-05,
+      "loss": 0.7933,
+      "step": 670
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978058592244203e-05,
+      "loss": 0.8239,
+      "step": 671
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9779630274878442e-05,
+      "loss": 0.6691,
+      "step": 672
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977867257388176e-05,
+      "loss": 0.8408,
+      "step": 673
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977771281965308e-05,
+      "loss": 0.8101,
+      "step": 674
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9776751012393918e-05,
+      "loss": 0.8353,
+      "step": 675
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775787152306223e-05,
+      "loss": 0.8623,
+      "step": 676
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774821239592383e-05,
+      "loss": 0.8124,
+      "step": 677
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773853274455204e-05,
+      "loss": 0.8126,
+      "step": 678
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9772883257097937e-05,
+      "loss": 0.6332,
+      "step": 679
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771911187724253e-05,
+      "loss": 0.7823,
+      "step": 680
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977093706653826e-05,
+      "loss": 0.7921,
+      "step": 681
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976996089374449e-05,
+      "loss": 0.8018,
+      "step": 682
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768982669547914e-05,
+      "loss": 0.5832,
+      "step": 683
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768002394153927e-05,
+      "loss": 0.8363,
+      "step": 684
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767020067768365e-05,
+      "loss": 0.8041,
+      "step": 685
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9766035690597484e-05,
+      "loss": 0.8307,
+      "step": 686
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976504926284797e-05,
+      "loss": 0.7973,
+      "step": 687
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9764060784726945e-05,
+      "loss": 0.6524,
+      "step": 688
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9763070256441968e-05,
+      "loss": 0.6505,
+      "step": 689
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976207767820101e-05,
+      "loss": 0.7975,
+      "step": 690
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976108305021249e-05,
+      "loss": 0.6389,
+      "step": 691
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760086372685246e-05,
+      "loss": 0.8646,
+      "step": 692
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759087645828554e-05,
+      "loss": 0.8141,
+      "step": 693
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9758086869852115e-05,
+      "loss": 0.8366,
+      "step": 694
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757084044966063e-05,
+      "loss": 0.8394,
+      "step": 695
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9756079171380964e-05,
+      "loss": 0.8581,
+      "step": 696
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755072249307807e-05,
+      "loss": 0.8581,
+      "step": 697
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975406327895802e-05,
+      "loss": 0.792,
+      "step": 698
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9753052260543456e-05,
+      "loss": 0.7986,
+      "step": 699
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9752039194276395e-05,
+      "loss": 0.7731,
+      "step": 700
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975102408036956e-05,
+      "loss": 0.8321,
+      "step": 701
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9750006919036083e-05,
+      "loss": 0.7722,
+      "step": 702
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974898771048955e-05,
+      "loss": 0.8172,
+      "step": 703
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747966454943956e-05,
+      "loss": 0.8431,
+      "step": 704
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9746943152613734e-05,
+      "loss": 0.8387,
+      "step": 705
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745917803713754e-05,
+      "loss": 0.8191,
+      "step": 706
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744890408459305e-05,
+      "loss": 0.8099,
+      "step": 707
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743860967066108e-05,
+      "loss": 0.811,
+      "step": 708
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974282947975032e-05,
+      "loss": 0.8278,
+      "step": 709
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974179594672852e-05,
+      "loss": 0.5917,
+      "step": 710
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740760368217714e-05,
+      "loss": 0.8559,
+      "step": 711
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9739722744435353e-05,
+      "loss": 0.6037,
+      "step": 712
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97386830755993e-05,
+      "loss": 0.6059,
+      "step": 713
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9737641361927862e-05,
+      "loss": 0.8023,
+      "step": 714
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736597603639756e-05,
+      "loss": 0.6288,
+      "step": 715
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9735551800954152e-05,
+      "loss": 0.8275,
+      "step": 716
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9734503954090634e-05,
+      "loss": 0.8025,
+      "step": 717
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973345406326922e-05,
+      "loss": 0.8171,
+      "step": 718
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973240212871035e-05,
+      "loss": 0.7923,
+      "step": 719
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9731348150634903e-05,
+      "loss": 0.7767,
+      "step": 720
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9730292129264185e-05,
+      "loss": 0.7988,
+      "step": 721
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9729234064819925e-05,
+      "loss": 0.8168,
+      "step": 722
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972817395752429e-05,
+      "loss": 0.8137,
+      "step": 723
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972711180759987e-05,
+      "loss": 0.6142,
+      "step": 724
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726047615269682e-05,
+      "loss": 0.8381,
+      "step": 725
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972498138075718e-05,
+      "loss": 0.6253,
+      "step": 726
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723913104286237e-05,
+      "loss": 0.7934,
+      "step": 727
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9722842786081163e-05,
+      "loss": 0.6504,
+      "step": 728
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9721770426366688e-05,
+      "loss": 0.647,
+      "step": 729
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972069602536798e-05,
+      "loss": 0.8221,
+      "step": 730
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9719619583310634e-05,
+      "loss": 0.8078,
+      "step": 731
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971854110042067e-05,
+      "loss": 0.7932,
+      "step": 732
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717460576924528e-05,
+      "loss": 0.8377,
+      "step": 733
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9716378013049094e-05,
+      "loss": 0.8358,
+      "step": 734
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715293409021678e-05,
+      "loss": 0.8467,
+      "step": 735
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971420676507001e-05,
+      "loss": 0.811,
+      "step": 736
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971311808142225e-05,
+      "loss": 0.6318,
+      "step": 737
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9712027358306993e-05,
+      "loss": 0.7945,
+      "step": 738
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971093459595326e-05,
+      "loss": 0.8066,
+      "step": 739
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9709839794590497e-05,
+      "loss": 0.792,
+      "step": 740
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9708742954448578e-05,
+      "loss": 0.8359,
+      "step": 741
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970764407575781e-05,
+      "loss": 0.7913,
+      "step": 742
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970654315874893e-05,
+      "loss": 0.8019,
+      "step": 743
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9705440203653086e-05,
+      "loss": 0.6118,
+      "step": 744
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970433521070187e-05,
+      "loss": 0.6123,
+      "step": 745
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9703228180127304e-05,
+      "loss": 0.8182,
+      "step": 746
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9702119112161827e-05,
+      "loss": 0.8385,
+      "step": 747
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970100800703831e-05,
+      "loss": 0.8533,
+      "step": 748
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9699894864990054e-05,
+      "loss": 0.8366,
+      "step": 749
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9698779686250784e-05,
+      "loss": 0.7817,
+      "step": 750
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969766247105466e-05,
+      "loss": 0.8016,
+      "step": 751
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696543219636254e-05,
+      "loss": 0.8281,
+      "step": 752
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9695421932230583e-05,
+      "loss": 0.8182,
+      "step": 753
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9694298609073085e-05,
+      "loss": 0.8003,
+      "step": 754
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9693173250399618e-05,
+      "loss": 0.8056,
+      "step": 755
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9692045856446482e-05,
+      "loss": 0.632,
+      "step": 756
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969091642745039e-05,
+      "loss": 0.7967,
+      "step": 757
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9689784963648495e-05,
+      "loss": 0.826,
+      "step": 758
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688651465278365e-05,
+      "loss": 0.7604,
+      "step": 759
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9687515932578005e-05,
+      "loss": 0.798,
+      "step": 760
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968637836578584e-05,
+      "loss": 0.6284,
+      "step": 761
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685238765140728e-05,
+      "loss": 0.8105,
+      "step": 762
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968409713088195e-05,
+      "loss": 0.8126,
+      "step": 763
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9682953463249216e-05,
+      "loss": 0.8296,
+      "step": 764
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9681807762482665e-05,
+      "loss": 0.8119,
+      "step": 765
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9680660028822854e-05,
+      "loss": 0.7916,
+      "step": 766
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9679510262510778e-05,
+      "loss": 0.8448,
+      "step": 767
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9678358463787855e-05,
+      "loss": 0.6168,
+      "step": 768
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9677204632895926e-05,
+      "loss": 0.8073,
+      "step": 769
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676048770077257e-05,
+      "loss": 0.8233,
+      "step": 770
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9674890875574552e-05,
+      "loss": 0.7926,
+      "step": 771
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9673730949630934e-05,
+      "loss": 0.8432,
+      "step": 772
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9672568992489952e-05,
+      "loss": 0.6678,
+      "step": 773
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967140500439558e-05,
+      "loss": 0.822,
+      "step": 774
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9670238985592224e-05,
+      "loss": 0.811,
+      "step": 775
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9669070936324713e-05,
+      "loss": 0.8284,
+      "step": 776
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96679008568383e-05,
+      "loss": 0.8081,
+      "step": 777
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9666728747378673e-05,
+      "loss": 0.7576,
+      "step": 778
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9665554608191934e-05,
+      "loss": 0.7893,
+      "step": 779
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966437843952462e-05,
+      "loss": 0.8106,
+      "step": 780
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9663200241623695e-05,
+      "loss": 0.8152,
+      "step": 781
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966202001473654e-05,
+      "loss": 0.6438,
+      "step": 782
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9660837759110972e-05,
+      "loss": 0.6213,
+      "step": 783
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965965347499523e-05,
+      "loss": 0.8089,
+      "step": 784
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965846716263797e-05,
+      "loss": 0.7996,
+      "step": 785
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9657278822288295e-05,
+      "loss": 0.8032,
+      "step": 786
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9656088454195716e-05,
+      "loss": 0.882,
+      "step": 787
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965489605861017e-05,
+      "loss": 0.7578,
+      "step": 788
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9653701635782033e-05,
+      "loss": 0.8531,
+      "step": 789
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9652505185962095e-05,
+      "loss": 0.833,
+      "step": 790
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965130670940157e-05,
+      "loss": 0.7871,
+      "step": 791
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9650106206352113e-05,
+      "loss": 0.5783,
+      "step": 792
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9648903677065787e-05,
+      "loss": 0.8012,
+      "step": 793
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964769912179509e-05,
+      "loss": 0.8187,
+      "step": 794
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9646492540792943e-05,
+      "loss": 0.7957,
+      "step": 795
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9645283934312693e-05,
+      "loss": 0.6205,
+      "step": 796
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9644073302608104e-05,
+      "loss": 0.7805,
+      "step": 797
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9642860645933385e-05,
+      "loss": 0.836,
+      "step": 798
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9641645964543148e-05,
+      "loss": 0.8297,
+      "step": 799
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9640429258692446e-05,
+      "loss": 0.6179,
+      "step": 800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963921052863675e-05,
+      "loss": 0.8186,
+      "step": 801
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9637989774631956e-05,
+      "loss": 0.6403,
+      "step": 802
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963676699693439e-05,
+      "loss": 0.8041,
+      "step": 803
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9635542195800794e-05,
+      "loss": 0.817,
+      "step": 804
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963431537148834e-05,
+      "loss": 0.6547,
+      "step": 805
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963308652425463e-05,
+      "loss": 0.8336,
+      "step": 806
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963185565435768e-05,
+      "loss": 0.8174,
+      "step": 807
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963062276205594e-05,
+      "loss": 0.6234,
+      "step": 808
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9629387847608282e-05,
+      "loss": 0.8527,
+      "step": 809
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9628150911273996e-05,
+      "loss": 0.8271,
+      "step": 810
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9626911953312807e-05,
+      "loss": 0.8034,
+      "step": 811
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962567097398486e-05,
+      "loss": 0.794,
+      "step": 812
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9624427973550717e-05,
+      "loss": 0.6563,
+      "step": 813
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962318295227138e-05,
+      "loss": 0.7947,
+      "step": 814
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962193591040826e-05,
+      "loss": 0.6141,
+      "step": 815
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9620686848223203e-05,
+      "loss": 0.7936,
+      "step": 816
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961943576597847e-05,
+      "loss": 0.817,
+      "step": 817
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9618182663936762e-05,
+      "loss": 0.5891,
+      "step": 818
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9616927542361185e-05,
+      "loss": 0.8305,
+      "step": 819
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615670401515274e-05,
+      "loss": 0.8207,
+      "step": 820
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9614411241663e-05,
+      "loss": 0.8161,
+      "step": 821
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9613150063068746e-05,
+      "loss": 0.8227,
+      "step": 822
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961188686599732e-05,
+      "loss": 0.8141,
+      "step": 823
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961062165071396e-05,
+      "loss": 0.764,
+      "step": 824
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9609354417484315e-05,
+      "loss": 0.8216,
+      "step": 825
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960808516657448e-05,
+      "loss": 0.8238,
+      "step": 826
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9606813898250948e-05,
+      "loss": 0.8092,
+      "step": 827
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9605540612780656e-05,
+      "loss": 0.8164,
+      "step": 828
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9604265310430954e-05,
+      "loss": 0.7899,
+      "step": 829
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9602987991469615e-05,
+      "loss": 0.7749,
+      "step": 830
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960170865616484e-05,
+      "loss": 0.8636,
+      "step": 831
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9600427304785253e-05,
+      "loss": 0.8128,
+      "step": 832
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95991439375999e-05,
+      "loss": 0.8137,
+      "step": 833
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959785855487825e-05,
+      "loss": 0.7894,
+      "step": 834
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959657115689019e-05,
+      "loss": 0.7456,
+      "step": 835
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9595281743906042e-05,
+      "loss": 0.7976,
+      "step": 836
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959399031619654e-05,
+      "loss": 0.8041,
+      "step": 837
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959269687403285e-05,
+      "loss": 0.7957,
+      "step": 838
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9591401417686552e-05,
+      "loss": 0.7509,
+      "step": 839
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9590103947429657e-05,
+      "loss": 0.7867,
+      "step": 840
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9588804463534595e-05,
+      "loss": 0.8177,
+      "step": 841
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9587502966274218e-05,
+      "loss": 0.8585,
+      "step": 842
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9586199455921802e-05,
+      "loss": 0.629,
+      "step": 843
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958489393275105e-05,
+      "loss": 0.7858,
+      "step": 844
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9583586397036077e-05,
+      "loss": 0.6488,
+      "step": 845
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958227684905143e-05,
+      "loss": 0.8074,
+      "step": 846
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958096528907207e-05,
+      "loss": 0.7855,
+      "step": 847
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9579651717373393e-05,
+      "loss": 0.7767,
+      "step": 848
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957833613423121e-05,
+      "loss": 0.6482,
+      "step": 849
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9577018539921747e-05,
+      "loss": 0.8062,
+      "step": 850
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9575698934721668e-05,
+      "loss": 0.8072,
+      "step": 851
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9574377318908048e-05,
+      "loss": 0.8087,
+      "step": 852
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9573053692758387e-05,
+      "loss": 0.6411,
+      "step": 853
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9571728056550604e-05,
+      "loss": 0.7811,
+      "step": 854
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9570400410563054e-05,
+      "loss": 0.8104,
+      "step": 855
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956907075507449e-05,
+      "loss": 0.8388,
+      "step": 856
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956773909036411e-05,
+      "loss": 0.8146,
+      "step": 857
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956640541671152e-05,
+      "loss": 0.7964,
+      "step": 858
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9565069734396754e-05,
+      "loss": 0.7723,
+      "step": 859
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956373204370026e-05,
+      "loss": 0.8224,
+      "step": 860
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956239234490292e-05,
+      "loss": 0.8132,
+      "step": 861
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9561050638286036e-05,
+      "loss": 0.8003,
+      "step": 862
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955970692413131e-05,
+      "loss": 0.8166,
+      "step": 863
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95583612027209e-05,
+      "loss": 0.7983,
+      "step": 864
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9557013474337358e-05,
+      "loss": 0.8063,
+      "step": 865
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9555663739263667e-05,
+      "loss": 0.5879,
+      "step": 866
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955431199778323e-05,
+      "loss": 0.8209,
+      "step": 867
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955295825017988e-05,
+      "loss": 0.7969,
+      "step": 868
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955160249673786e-05,
+      "loss": 0.791,
+      "step": 869
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955024473774184e-05,
+      "loss": 0.7963,
+      "step": 870
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9548884973476902e-05,
+      "loss": 0.8048,
+      "step": 871
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9547523204228566e-05,
+      "loss": 0.8137,
+      "step": 872
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954615943028276e-05,
+      "loss": 0.8579,
+      "step": 873
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954479365192583e-05,
+      "loss": 0.6296,
+      "step": 874
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954342586944456e-05,
+      "loss": 0.785,
+      "step": 875
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9542056083126136e-05,
+      "loss": 0.8209,
+      "step": 876
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9540684293258172e-05,
+      "loss": 0.8009,
+      "step": 877
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9539310500128708e-05,
+      "loss": 0.6196,
+      "step": 878
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9537934704026197e-05,
+      "loss": 0.8185,
+      "step": 879
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9536556905239518e-05,
+      "loss": 0.7681,
+      "step": 880
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9535177104057966e-05,
+      "loss": 0.7606,
+      "step": 881
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9533795300771258e-05,
+      "loss": 0.8252,
+      "step": 882
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9532411495669535e-05,
+      "loss": 0.8286,
+      "step": 883
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9531025689043347e-05,
+      "loss": 0.6163,
+      "step": 884
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9529637881183684e-05,
+      "loss": 0.7734,
+      "step": 885
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9528248072381935e-05,
+      "loss": 0.5787,
+      "step": 886
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9526856262929924e-05,
+      "loss": 0.7855,
+      "step": 887
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952546245311989e-05,
+      "loss": 0.8408,
+      "step": 888
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952406664324449e-05,
+      "loss": 0.8128,
+      "step": 889
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95226688335968e-05,
+      "loss": 0.7936,
+      "step": 890
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9521269024470325e-05,
+      "loss": 0.6131,
+      "step": 891
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9519867216158975e-05,
+      "loss": 0.7636,
+      "step": 892
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9518463408957097e-05,
+      "loss": 0.7781,
+      "step": 893
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9517057603159445e-05,
+      "loss": 0.822,
+      "step": 894
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95156497990612e-05,
+      "loss": 0.8082,
+      "step": 895
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9514239996957952e-05,
+      "loss": 0.8156,
+      "step": 896
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9512828197145723e-05,
+      "loss": 0.5654,
+      "step": 897
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951141439992095e-05,
+      "loss": 0.8303,
+      "step": 898
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9509998605580486e-05,
+      "loss": 0.8077,
+      "step": 899
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9508580814421606e-05,
+      "loss": 0.7941,
+      "step": 900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9507161026742004e-05,
+      "loss": 0.8007,
+      "step": 901
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9505739242839798e-05,
+      "loss": 0.5917,
+      "step": 902
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950431546301351e-05,
+      "loss": 0.856,
+      "step": 903
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9502889687562104e-05,
+      "loss": 0.8068,
+      "step": 904
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9501461916784944e-05,
+      "loss": 0.7513,
+      "step": 905
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950003215098182e-05,
+      "loss": 0.789,
+      "step": 906
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9498600390452946e-05,
+      "loss": 0.6681,
+      "step": 907
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9497166635498938e-05,
+      "loss": 0.8351,
+      "step": 908
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9495730886420853e-05,
+      "loss": 0.6398,
+      "step": 909
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949429314352015e-05,
+      "loss": 0.7696,
+      "step": 910
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9492853407098718e-05,
+      "loss": 0.7937,
+      "step": 911
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949141167745885e-05,
+      "loss": 0.6114,
+      "step": 912
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9489967954903272e-05,
+      "loss": 0.818,
+      "step": 913
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9488522239735127e-05,
+      "loss": 0.8165,
+      "step": 914
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9487074532257968e-05,
+      "loss": 0.6524,
+      "step": 915
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9485624832775768e-05,
+      "loss": 0.8182,
+      "step": 916
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9484173141592926e-05,
+      "loss": 0.8062,
+      "step": 917
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9482719459014253e-05,
+      "loss": 0.7971,
+      "step": 918
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9481263785344976e-05,
+      "loss": 0.8076,
+      "step": 919
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9479806120890747e-05,
+      "loss": 0.7857,
+      "step": 920
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947834646595763e-05,
+      "loss": 0.7926,
+      "step": 921
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9476884820852117e-05,
+      "loss": 0.7918,
+      "step": 922
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94754211858811e-05,
+      "loss": 0.608,
+      "step": 923
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94739555613519e-05,
+      "loss": 0.7706,
+      "step": 924
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947248794757226e-05,
+      "loss": 0.8148,
+      "step": 925
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9471018344850338e-05,
+      "loss": 0.632,
+      "step": 926
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9469546753494695e-05,
+      "loss": 0.7828,
+      "step": 927
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9468073173814336e-05,
+      "loss": 0.7669,
+      "step": 928
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9466597606118657e-05,
+      "loss": 0.7799,
+      "step": 929
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9465120050717493e-05,
+      "loss": 0.798,
+      "step": 930
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946364050792108e-05,
+      "loss": 0.7728,
+      "step": 931
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946215897804008e-05,
+      "loss": 0.8298,
+      "step": 932
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9460675461385575e-05,
+      "loss": 0.7761,
+      "step": 933
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9459189958269056e-05,
+      "loss": 0.8016,
+      "step": 934
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945770246900243e-05,
+      "loss": 0.8109,
+      "step": 935
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9456212993898036e-05,
+      "loss": 0.7809,
+      "step": 936
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9454721533268612e-05,
+      "loss": 0.7694,
+      "step": 937
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9453228087427324e-05,
+      "loss": 0.8072,
+      "step": 938
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945173265668775e-05,
+      "loss": 0.7893,
+      "step": 939
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9450235241363886e-05,
+      "loss": 0.836,
+      "step": 940
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9448735841770147e-05,
+      "loss": 0.793,
+      "step": 941
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9447234458221357e-05,
+      "loss": 0.8251,
+      "step": 942
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944573109103277e-05,
+      "loss": 0.7631,
+      "step": 943
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9444225740520043e-05,
+      "loss": 0.7873,
+      "step": 944
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944271840699926e-05,
+      "loss": 0.7965,
+      "step": 945
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944120909078691e-05,
+      "loss": 0.6407,
+      "step": 946
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9439697792199908e-05,
+      "loss": 0.8583,
+      "step": 947
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9438184511555583e-05,
+      "loss": 0.8372,
+      "step": 948
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9436669249171677e-05,
+      "loss": 0.8205,
+      "step": 949
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9435152005366355e-05,
+      "loss": 0.7775,
+      "step": 950
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9433632780458188e-05,
+      "loss": 0.7728,
+      "step": 951
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943211157476617e-05,
+      "loss": 0.8038,
+      "step": 952
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943058838860971e-05,
+      "loss": 0.8074,
+      "step": 953
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9429063222308634e-05,
+      "loss": 0.6103,
+      "step": 954
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942753607618317e-05,
+      "loss": 0.8262,
+      "step": 955
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9426006950553993e-05,
+      "loss": 0.64,
+      "step": 956
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942447584574216e-05,
+      "loss": 0.7918,
+      "step": 957
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942294276206916e-05,
+      "loss": 0.7975,
+      "step": 958
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9421407699856903e-05,
+      "loss": 0.792,
+      "step": 959
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9419870659427694e-05,
+      "loss": 0.5922,
+      "step": 960
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9418331641104274e-05,
+      "loss": 0.8167,
+      "step": 961
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941679064520979e-05,
+      "loss": 0.5905,
+      "step": 962
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9415247672067808e-05,
+      "loss": 0.8289,
+      "step": 963
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.94137027220023e-05,
+      "loss": 0.8309,
+      "step": 964
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941215579533767e-05,
+      "loss": 0.8148,
+      "step": 965
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9410606892398717e-05,
+      "loss": 0.8146,
+      "step": 966
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9409056013510667e-05,
+      "loss": 0.8684,
+      "step": 967
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9407503158999165e-05,
+      "loss": 0.8017,
+      "step": 968
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9405948329190252e-05,
+      "loss": 0.7972,
+      "step": 969
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9404391524410412e-05,
+      "loss": 0.7898,
+      "step": 970
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9402832744986518e-05,
+      "loss": 0.7864,
+      "step": 971
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9401271991245867e-05,
+      "loss": 0.7482,
+      "step": 972
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9399709263516176e-05,
+      "loss": 0.8365,
+      "step": 973
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939814456212557e-05,
+      "loss": 0.7815,
+      "step": 974
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939657788740259e-05,
+      "loss": 0.7793,
+      "step": 975
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9395009239676186e-05,
+      "loss": 0.8057,
+      "step": 976
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939343861927574e-05,
+      "loss": 0.7947,
+      "step": 977
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939186602653102e-05,
+      "loss": 0.7806,
+      "step": 978
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9390291461772234e-05,
+      "loss": 0.7692,
+      "step": 979
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9388714925329997e-05,
+      "loss": 0.8019,
+      "step": 980
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9387136417535325e-05,
+      "loss": 0.7777,
+      "step": 981
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9385555938719666e-05,
+      "loss": 0.8272,
+      "step": 982
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9383973489214872e-05,
+      "loss": 0.7676,
+      "step": 983
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9382389069353207e-05,
+      "loss": 0.8131,
+      "step": 984
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9380802679467355e-05,
+      "loss": 0.8104,
+      "step": 985
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9379214319890413e-05,
+      "loss": 0.7798,
+      "step": 986
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9377623990955885e-05,
+      "loss": 0.7923,
+      "step": 987
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9376031692997695e-05,
+      "loss": 0.791,
+      "step": 988
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937443742635018e-05,
+      "loss": 0.7886,
+      "step": 989
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9372841191348088e-05,
+      "loss": 0.7893,
+      "step": 990
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937124298832658e-05,
+      "loss": 0.6532,
+      "step": 991
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9369642817621232e-05,
+      "loss": 0.8349,
+      "step": 992
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9368040679568033e-05,
+      "loss": 0.8127,
+      "step": 993
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9366436574503383e-05,
+      "loss": 0.8198,
+      "step": 994
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9364830502764104e-05,
+      "loss": 0.7729,
+      "step": 995
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936322246468741e-05,
+      "loss": 0.7693,
+      "step": 996
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936161246061095e-05,
+      "loss": 0.7787,
+      "step": 997
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936000049087278e-05,
+      "loss": 0.6489,
+      "step": 998
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935838655581136e-05,
+      "loss": 0.8276,
+      "step": 999
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9356770655765568e-05,
+      "loss": 0.8141,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9355152791074704e-05,
+      "loss": 0.7779,
+      "step": 1001
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935353296207846e-05,
+      "loss": 0.815,
+      "step": 1002
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9351911169116962e-05,
+      "loss": 0.7856,
+      "step": 1003
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9350287412530732e-05,
+      "loss": 0.7236,
+      "step": 1004
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934866169266071e-05,
+      "loss": 0.7926,
+      "step": 1005
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9347034009848258e-05,
+      "loss": 0.7878,
+      "step": 1006
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9345404364435137e-05,
+      "loss": 0.813,
+      "step": 1007
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9343772756763517e-05,
+      "loss": 0.6269,
+      "step": 1008
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9342139187175998e-05,
+      "loss": 0.6099,
+      "step": 1009
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9340503656015573e-05,
+      "loss": 0.7637,
+      "step": 1010
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933886616362566e-05,
+      "loss": 0.7893,
+      "step": 1011
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933722671035008e-05,
+      "loss": 0.8196,
+      "step": 1012
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9335585296533076e-05,
+      "loss": 0.7946,
+      "step": 1013
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9333941922519294e-05,
+      "loss": 0.7791,
+      "step": 1014
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933229658865379e-05,
+      "loss": 0.6206,
+      "step": 1015
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9330649295282036e-05,
+      "loss": 0.6373,
+      "step": 1016
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9329000042749923e-05,
+      "loss": 0.7687,
+      "step": 1017
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9327348831403738e-05,
+      "loss": 0.8225,
+      "step": 1018
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9325695661590184e-05,
+      "loss": 0.7755,
+      "step": 1019
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9324040533656384e-05,
+      "loss": 0.7897,
+      "step": 1020
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932238344794986e-05,
+      "loss": 0.8162,
+      "step": 1021
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932072440481856e-05,
+      "loss": 0.7858,
+      "step": 1022
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9319063404610822e-05,
+      "loss": 0.7695,
+      "step": 1023
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9317400447675418e-05,
+      "loss": 0.6306,
+      "step": 1024
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9315735534361514e-05,
+      "loss": 0.6105,
+      "step": 1025
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9314068665018694e-05,
+      "loss": 0.6216,
+      "step": 1026
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9312399839996952e-05,
+      "loss": 0.8046,
+      "step": 1027
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931072905964669e-05,
+      "loss": 0.8183,
+      "step": 1028
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930905632431872e-05,
+      "loss": 0.8076,
+      "step": 1029
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9307381634364274e-05,
+      "loss": 0.8085,
+      "step": 1030
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9305704990134984e-05,
+      "loss": 0.764,
+      "step": 1031
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9304026391982895e-05,
+      "loss": 0.8077,
+      "step": 1032
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930234584026046e-05,
+      "loss": 0.7925,
+      "step": 1033
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9300663335320552e-05,
+      "loss": 0.8135,
+      "step": 1034
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9298978877516442e-05,
+      "loss": 0.8189,
+      "step": 1035
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929729246720182e-05,
+      "loss": 0.6159,
+      "step": 1036
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9295604104730775e-05,
+      "loss": 0.8015,
+      "step": 1037
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9293913790457825e-05,
+      "loss": 0.7953,
+      "step": 1038
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9292221524737872e-05,
+      "loss": 0.7794,
+      "step": 1039
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9290527307926256e-05,
+      "loss": 0.5759,
+      "step": 1040
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9288831140378705e-05,
+      "loss": 0.7822,
+      "step": 1041
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9287133022451358e-05,
+      "loss": 0.5895,
+      "step": 1042
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9285432954500782e-05,
+      "loss": 0.7921,
+      "step": 1043
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9283730936883934e-05,
+      "loss": 0.7887,
+      "step": 1044
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9282026969958187e-05,
+      "loss": 0.5649,
+      "step": 1045
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928032105408133e-05,
+      "loss": 0.8124,
+      "step": 1046
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9278613189611545e-05,
+      "loss": 0.8256,
+      "step": 1047
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9276903376907442e-05,
+      "loss": 0.7873,
+      "step": 1048
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927519161632802e-05,
+      "loss": 0.6114,
+      "step": 1049
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9273477908232712e-05,
+      "loss": 0.6228,
+      "step": 1050
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9271762252981338e-05,
+      "loss": 0.5892,
+      "step": 1051
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9270044650934134e-05,
+      "loss": 0.8034,
+      "step": 1052
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9268325102451747e-05,
+      "loss": 0.7963,
+      "step": 1053
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926660360789524e-05,
+      "loss": 0.7751,
+      "step": 1054
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926488016762606e-05,
+      "loss": 0.792,
+      "step": 1055
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926315478200609e-05,
+      "loss": 0.7963,
+      "step": 1056
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9261427451397602e-05,
+      "loss": 0.8093,
+      "step": 1057
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9259698176163292e-05,
+      "loss": 0.64,
+      "step": 1058
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9257966956666252e-05,
+      "loss": 0.8027,
+      "step": 1059
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9256233793269988e-05,
+      "loss": 0.7768,
+      "step": 1060
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9254498686338415e-05,
+      "loss": 0.7952,
+      "step": 1061
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9252761636235852e-05,
+      "loss": 0.7976,
+      "step": 1062
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925102264332703e-05,
+      "loss": 0.8292,
+      "step": 1063
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9249281707977083e-05,
+      "loss": 0.7634,
+      "step": 1064
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924753883055156e-05,
+      "loss": 0.5943,
+      "step": 1065
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9245794011416406e-05,
+      "loss": 0.7822,
+      "step": 1066
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9244047250937992e-05,
+      "loss": 0.7652,
+      "step": 1067
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9242298549483078e-05,
+      "loss": 0.8217,
+      "step": 1068
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9240547907418843e-05,
+      "loss": 0.8321,
+      "step": 1069
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9238795325112867e-05,
+      "loss": 0.6391,
+      "step": 1070
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9237040802933148e-05,
+      "loss": 0.8135,
+      "step": 1071
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9235284341248075e-05,
+      "loss": 0.7742,
+      "step": 1072
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9233525940426458e-05,
+      "loss": 0.7645,
+      "step": 1073
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9231765600837507e-05,
+      "loss": 0.7638,
+      "step": 1074
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9230003322850842e-05,
+      "loss": 0.7654,
+      "step": 1075
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922823910683649e-05,
+      "loss": 0.8321,
+      "step": 1076
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9226472953164884e-05,
+      "loss": 0.8086,
+      "step": 1077
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922470486220686e-05,
+      "loss": 0.7917,
+      "step": 1078
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9222934834333672e-05,
+      "loss": 0.7961,
+      "step": 1079
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9221162869916966e-05,
+      "loss": 0.7997,
+      "step": 1080
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921938896932881e-05,
+      "loss": 0.6051,
+      "step": 1081
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9217613132941665e-05,
+      "loss": 0.7941,
+      "step": 1082
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9215835361128407e-05,
+      "loss": 0.5987,
+      "step": 1083
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921405565426231e-05,
+      "loss": 0.7952,
+      "step": 1084
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9212274012717066e-05,
+      "loss": 0.824,
+      "step": 1085
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9210490436866768e-05,
+      "loss": 0.763,
+      "step": 1086
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9208704927085908e-05,
+      "loss": 0.5774,
+      "step": 1087
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9206917483749394e-05,
+      "loss": 0.7604,
+      "step": 1088
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9205128107232535e-05,
+      "loss": 0.7974,
+      "step": 1089
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9203336797911048e-05,
+      "loss": 0.7972,
+      "step": 1090
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9201543556161054e-05,
+      "loss": 0.5774,
+      "step": 1091
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919974838235908e-05,
+      "loss": 0.7815,
+      "step": 1092
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9197951276882068e-05,
+      "loss": 0.7772,
+      "step": 1093
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9196152240107345e-05,
+      "loss": 0.7998,
+      "step": 1094
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919435127241266e-05,
+      "loss": 0.8371,
+      "step": 1095
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.919254837417616e-05,
+      "loss": 0.6294,
+      "step": 1096
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.919074354577641e-05,
+      "loss": 0.771,
+      "step": 1097
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9188936787592363e-05,
+      "loss": 0.7811,
+      "step": 1098
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9187128100003384e-05,
+      "loss": 0.7981,
+      "step": 1099
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918531748338925e-05,
+      "loss": 0.8037,
+      "step": 1100
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9183504938130126e-05,
+      "loss": 0.778,
+      "step": 1101
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9181690464606603e-05,
+      "loss": 0.8151,
+      "step": 1102
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9179874063199664e-05,
+      "loss": 0.7439,
+      "step": 1103
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91780557342907e-05,
+      "loss": 0.8423,
+      "step": 1104
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91762354782615e-05,
+      "loss": 0.8283,
+      "step": 1105
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9174413295494277e-05,
+      "loss": 0.7445,
+      "step": 1106
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9172589186371623e-05,
+      "loss": 0.7878,
+      "step": 1107
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9170763151276552e-05,
+      "loss": 0.7482,
+      "step": 1108
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9168935190592477e-05,
+      "loss": 0.7821,
+      "step": 1109
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9167105304703218e-05,
+      "loss": 0.781,
+      "step": 1110
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9165273493992993e-05,
+      "loss": 0.6365,
+      "step": 1111
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916343975884643e-05,
+      "loss": 0.7964,
+      "step": 1112
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9161604099648555e-05,
+      "loss": 0.6368,
+      "step": 1113
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9159766516784812e-05,
+      "loss": 0.7816,
+      "step": 1114
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915792701064103e-05,
+      "loss": 0.7694,
+      "step": 1115
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9156085581603454e-05,
+      "loss": 0.6439,
+      "step": 1116
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9154242230058728e-05,
+      "loss": 0.8122,
+      "step": 1117
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915239695639391e-05,
+      "loss": 0.7651,
+      "step": 1118
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9150549760996436e-05,
+      "loss": 0.5901,
+      "step": 1119
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9148700644254176e-05,
+      "loss": 0.788,
+      "step": 1120
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914684960655539e-05,
+      "loss": 0.8135,
+      "step": 1121
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914499664828873e-05,
+      "loss": 0.789,
+      "step": 1122
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9143141769843274e-05,
+      "loss": 0.8022,
+      "step": 1123
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9141284971608487e-05,
+      "loss": 0.7483,
+      "step": 1124
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9139426253974243e-05,
+      "loss": 0.6286,
+      "step": 1125
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9137565617330814e-05,
+      "loss": 0.644,
+      "step": 1126
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913570306206888e-05,
+      "loss": 0.806,
+      "step": 1127
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9133838588579526e-05,
+      "loss": 0.7731,
+      "step": 1128
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9131972197254233e-05,
+      "loss": 0.7721,
+      "step": 1129
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913010388848489e-05,
+      "loss": 0.6013,
+      "step": 1130
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9128233662663786e-05,
+      "loss": 0.7746,
+      "step": 1131
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912636152018361e-05,
+      "loss": 0.6618,
+      "step": 1132
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9124487461437465e-05,
+      "loss": 0.7983,
+      "step": 1133
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912261148681884e-05,
+      "loss": 0.7956,
+      "step": 1134
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912073359672163e-05,
+      "loss": 0.8005,
+      "step": 1135
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911885379154015e-05,
+      "loss": 0.76,
+      "step": 1136
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9116972071669094e-05,
+      "loss": 0.7564,
+      "step": 1137
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911508843750357e-05,
+      "loss": 0.7717,
+      "step": 1138
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9113202889439083e-05,
+      "loss": 0.7736,
+      "step": 1139
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9111315427871547e-05,
+      "loss": 0.7945,
+      "step": 1140
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910942605319727e-05,
+      "loss": 0.7914,
+      "step": 1141
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9107534765812963e-05,
+      "loss": 0.7847,
+      "step": 1142
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9105641566115745e-05,
+      "loss": 0.7676,
+      "step": 1143
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9103746454503128e-05,
+      "loss": 0.6112,
+      "step": 1144
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9101849431373034e-05,
+      "loss": 0.8183,
+      "step": 1145
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909995049712378e-05,
+      "loss": 0.77,
+      "step": 1146
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909804965215408e-05,
+      "loss": 0.7777,
+      "step": 1147
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9096146896863063e-05,
+      "loss": 0.8037,
+      "step": 1148
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9094242231650246e-05,
+      "loss": 0.8118,
+      "step": 1149
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9092335656915558e-05,
+      "loss": 0.7522,
+      "step": 1150
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9090427173059317e-05,
+      "loss": 0.8399,
+      "step": 1151
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9088516780482252e-05,
+      "loss": 0.7932,
+      "step": 1152
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908660447958549e-05,
+      "loss": 0.788,
+      "step": 1153
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9084690270770558e-05,
+      "loss": 0.7723,
+      "step": 1154
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9082774154439377e-05,
+      "loss": 0.6389,
+      "step": 1155
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9080856130994277e-05,
+      "loss": 0.7465,
+      "step": 1156
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9078936200837993e-05,
+      "loss": 0.6584,
+      "step": 1157
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9077014364373644e-05,
+      "loss": 0.8153,
+      "step": 1158
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9075090622004766e-05,
+      "loss": 0.6388,
+      "step": 1159
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9073164974135286e-05,
+      "loss": 0.7659,
+      "step": 1160
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9071237421169534e-05,
+      "loss": 0.6029,
+      "step": 1161
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9069307963512232e-05,
+      "loss": 0.8295,
+      "step": 1162
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9067376601568516e-05,
+      "loss": 0.7752,
+      "step": 1163
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9065443335743917e-05,
+      "loss": 0.5861,
+      "step": 1164
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906350816644436e-05,
+      "loss": 0.809,
+      "step": 1165
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906157109407617e-05,
+      "loss": 0.7804,
+      "step": 1166
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905963211904608e-05,
+      "loss": 0.7582,
+      "step": 1167
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9057691241761212e-05,
+      "loss": 0.7778,
+      "step": 1168
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9055748462629103e-05,
+      "loss": 0.7789,
+      "step": 1169
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905380378205767e-05,
+      "loss": 0.7726,
+      "step": 1170
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905185720045524e-05,
+      "loss": 0.7972,
+      "step": 1171
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904990871823054e-05,
+      "loss": 0.7861,
+      "step": 1172
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9047958335792693e-05,
+      "loss": 0.8085,
+      "step": 1173
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904600605355122e-05,
+      "loss": 0.7612,
+      "step": 1174
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9044051871916043e-05,
+      "loss": 0.7652,
+      "step": 1175
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9042095791297483e-05,
+      "loss": 0.7578,
+      "step": 1176
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9040137812106257e-05,
+      "loss": 0.7842,
+      "step": 1177
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903817793475349e-05,
+      "loss": 0.7928,
+      "step": 1178
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9036216159650687e-05,
+      "loss": 0.8013,
+      "step": 1179
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9034252487209767e-05,
+      "loss": 0.7555,
+      "step": 1180
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903228691784305e-05,
+      "loss": 0.7668,
+      "step": 1181
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903031945196324e-05,
+      "loss": 0.6181,
+      "step": 1182
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902835008998345e-05,
+      "loss": 0.7693,
+      "step": 1183
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9026378832317188e-05,
+      "loss": 0.7977,
+      "step": 1184
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9024405679378355e-05,
+      "loss": 0.7832,
+      "step": 1185
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902243063158126e-05,
+      "loss": 0.7898,
+      "step": 1186
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90204536893406e-05,
+      "loss": 0.783,
+      "step": 1187
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901847485307148e-05,
+      "loss": 0.8028,
+      "step": 1188
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9016494123189396e-05,
+      "loss": 0.7707,
+      "step": 1189
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9014511500110237e-05,
+      "loss": 0.8234,
+      "step": 1190
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90125269842503e-05,
+      "loss": 0.8115,
+      "step": 1191
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9010540576026273e-05,
+      "loss": 0.8109,
+      "step": 1192
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9008552275855244e-05,
+      "loss": 0.7892,
+      "step": 1193
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90065620841547e-05,
+      "loss": 0.6101,
+      "step": 1194
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9004570001342516e-05,
+      "loss": 0.7886,
+      "step": 1195
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9002576027836976e-05,
+      "loss": 0.7909,
+      "step": 1196
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900058016405675e-05,
+      "loss": 0.7643,
+      "step": 1197
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8998582410420916e-05,
+      "loss": 0.7986,
+      "step": 1198
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899658276734894e-05,
+      "loss": 0.8018,
+      "step": 1199
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8994581235260688e-05,
+      "loss": 0.8145,
+      "step": 1200
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8992577814576426e-05,
+      "loss": 0.8068,
+      "step": 1201
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899057250571681e-05,
+      "loss": 0.7934,
+      "step": 1202
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8988565309102893e-05,
+      "loss": 0.7713,
+      "step": 1203
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898655622515613e-05,
+      "loss": 0.7971,
+      "step": 1204
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8984545254298373e-05,
+      "loss": 0.7827,
+      "step": 1205
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8982532396951858e-05,
+      "loss": 0.823,
+      "step": 1206
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8980517653539233e-05,
+      "loss": 0.6127,
+      "step": 1207
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8978501024483525e-05,
+      "loss": 0.778,
+      "step": 1208
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897648251020818e-05,
+      "loss": 0.7422,
+      "step": 1209
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8974462111137017e-05,
+      "loss": 0.7901,
+      "step": 1210
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897243982769426e-05,
+      "loss": 0.7508,
+      "step": 1211
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8970415660304536e-05,
+      "loss": 0.7921,
+      "step": 1212
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896838960939285e-05,
+      "loss": 0.7743,
+      "step": 1213
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896636167538462e-05,
+      "loss": 0.7853,
+      "step": 1214
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896433185870565e-05,
+      "loss": 0.8164,
+      "step": 1215
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896230015978214e-05,
+      "loss": 0.5959,
+      "step": 1216
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896026657904069e-05,
+      "loss": 0.7673,
+      "step": 1217
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8958231116908293e-05,
+      "loss": 0.7935,
+      "step": 1218
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8956193773812327e-05,
+      "loss": 0.8038,
+      "step": 1219
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8954154550180584e-05,
+      "loss": 0.7953,
+      "step": 1220
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8952113446441234e-05,
+      "loss": 0.801,
+      "step": 1221
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895007046302285e-05,
+      "loss": 0.7842,
+      "step": 1222
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8948025600354402e-05,
+      "loss": 0.7564,
+      "step": 1223
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8945978858865243e-05,
+      "loss": 0.7435,
+      "step": 1224
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8943930238985134e-05,
+      "loss": 0.8187,
+      "step": 1225
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8941879741144225e-05,
+      "loss": 0.6396,
+      "step": 1226
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8939827365773057e-05,
+      "loss": 0.7615,
+      "step": 1227
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8937773113302572e-05,
+      "loss": 0.7995,
+      "step": 1228
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8935716984164098e-05,
+      "loss": 0.7444,
+      "step": 1229
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893365897878936e-05,
+      "loss": 0.7837,
+      "step": 1230
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8931599097610485e-05,
+      "loss": 0.7941,
+      "step": 1231
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892953734105998e-05,
+      "loss": 0.8062,
+      "step": 1232
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892747370957076e-05,
+      "loss": 0.7777,
+      "step": 1233
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8925408203576117e-05,
+      "loss": 0.812,
+      "step": 1234
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8923340823509756e-05,
+      "loss": 0.8102,
+      "step": 1235
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892127156980576e-05,
+      "loss": 0.8037,
+      "step": 1236
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8919200442898617e-05,
+      "loss": 0.7548,
+      "step": 1237
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8917127443223193e-05,
+      "loss": 0.7703,
+      "step": 1238
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891505257121476e-05,
+      "loss": 0.7666,
+      "step": 1239
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8912975827308987e-05,
+      "loss": 0.7828,
+      "step": 1240
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8910897211941915e-05,
+      "loss": 0.7499,
+      "step": 1241
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8908816725550006e-05,
+      "loss": 0.7453,
+      "step": 1242
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8906734368570093e-05,
+      "loss": 0.6511,
+      "step": 1243
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890465014143941e-05,
+      "loss": 0.7597,
+      "step": 1244
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890256404459559e-05,
+      "loss": 0.7547,
+      "step": 1245
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890047607847664e-05,
+      "loss": 0.784,
+      "step": 1246
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8898386243520973e-05,
+      "loss": 0.8408,
+      "step": 1247
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8896294540167402e-05,
+      "loss": 0.7839,
+      "step": 1248
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8894200968855116e-05,
+      "loss": 0.5857,
+      "step": 1249
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8892105530023703e-05,
+      "loss": 0.7976,
+      "step": 1250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8890008224113146e-05,
+      "loss": 0.8224,
+      "step": 1251
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8887909051563812e-05,
+      "loss": 0.6075,
+      "step": 1252
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8885808012816472e-05,
+      "loss": 0.7898,
+      "step": 1253
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8883705108312275e-05,
+      "loss": 0.7638,
+      "step": 1254
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8881600338492772e-05,
+      "loss": 0.6307,
+      "step": 1255
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8879493703799908e-05,
+      "loss": 0.7557,
+      "step": 1256
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8877385204676007e-05,
+      "loss": 0.7974,
+      "step": 1257
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8875274841563792e-05,
+      "loss": 0.6301,
+      "step": 1258
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8873162614906374e-05,
+      "loss": 0.7877,
+      "step": 1259
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887104852514727e-05,
+      "loss": 0.8098,
+      "step": 1260
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8868932572730363e-05,
+      "loss": 0.582,
+      "step": 1261
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8866814758099948e-05,
+      "loss": 0.7587,
+      "step": 1262
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88646950817007e-05,
+      "loss": 0.8331,
+      "step": 1263
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8862573543977692e-05,
+      "loss": 0.8166,
+      "step": 1264
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8860450145376377e-05,
+      "loss": 0.7501,
+      "step": 1265
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8858324886342612e-05,
+      "loss": 0.792,
+      "step": 1266
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885619776732264e-05,
+      "loss": 0.7612,
+      "step": 1267
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885406878876309e-05,
+      "loss": 0.7474,
+      "step": 1268
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8851937951110985e-05,
+      "loss": 0.7678,
+      "step": 1269
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8849805254813735e-05,
+      "loss": 0.8244,
+      "step": 1270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8847670700319147e-05,
+      "loss": 0.7808,
+      "step": 1271
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8845534288075415e-05,
+      "loss": 0.6396,
+      "step": 1272
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8843396018531118e-05,
+      "loss": 0.7358,
+      "step": 1273
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884125589213523e-05,
+      "loss": 0.7632,
+      "step": 1274
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8839113909337123e-05,
+      "loss": 0.5807,
+      "step": 1275
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883697007058654e-05,
+      "loss": 0.7455,
+      "step": 1276
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8834824376333625e-05,
+      "loss": 0.589,
+      "step": 1277
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8832676827028913e-05,
+      "loss": 0.7786,
+      "step": 1278
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8830527423123324e-05,
+      "loss": 0.6089,
+      "step": 1279
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8828376165068173e-05,
+      "loss": 0.8055,
+      "step": 1280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8826223053315156e-05,
+      "loss": 0.791,
+      "step": 1281
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8824068088316365e-05,
+      "loss": 0.7986,
+      "step": 1282
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8821911270524282e-05,
+      "loss": 0.7698,
+      "step": 1283
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8819752600391764e-05,
+      "loss": 0.6137,
+      "step": 1284
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881759207837208e-05,
+      "loss": 0.7319,
+      "step": 1285
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881542970491887e-05,
+      "loss": 0.767,
+      "step": 1286
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881326548048617e-05,
+      "loss": 0.7697,
+      "step": 1287
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8811099405528405e-05,
+      "loss": 0.7342,
+      "step": 1288
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880893148050038e-05,
+      "loss": 0.8085,
+      "step": 1289
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8806761705857303e-05,
+      "loss": 0.7527,
+      "step": 1290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8804590082054757e-05,
+      "loss": 0.789,
+      "step": 1291
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8802416609548723e-05,
+      "loss": 0.7487,
+      "step": 1292
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8800241288795557e-05,
+      "loss": 0.7598,
+      "step": 1293
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8798064120252024e-05,
+      "loss": 0.784,
+      "step": 1294
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879588510437526e-05,
+      "loss": 0.7939,
+      "step": 1295
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8793704241622796e-05,
+      "loss": 0.7686,
+      "step": 1296
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879152153245254e-05,
+      "loss": 0.7514,
+      "step": 1297
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8789336977322805e-05,
+      "loss": 0.7807,
+      "step": 1298
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878715057669228e-05,
+      "loss": 0.774,
+      "step": 1299
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8784962331020045e-05,
+      "loss": 0.7688,
+      "step": 1300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878277224076557e-05,
+      "loss": 0.7955,
+      "step": 1301
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8780580306388707e-05,
+      "loss": 0.7882,
+      "step": 1302
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877838652834969e-05,
+      "loss": 0.7661,
+      "step": 1303
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8776190907109156e-05,
+      "loss": 0.6371,
+      "step": 1304
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8773993443128123e-05,
+      "loss": 0.8495,
+      "step": 1305
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8771794136867984e-05,
+      "loss": 0.7401,
+      "step": 1306
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8769592988790534e-05,
+      "loss": 0.7564,
+      "step": 1307
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8767389999357948e-05,
+      "loss": 0.7781,
+      "step": 1308
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876518516903279e-05,
+      "loss": 0.8103,
+      "step": 1309
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8762978498278007e-05,
+      "loss": 0.7575,
+      "step": 1310
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8760769987556937e-05,
+      "loss": 0.7748,
+      "step": 1311
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87585596373333e-05,
+      "loss": 0.7696,
+      "step": 1312
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8756347448071204e-05,
+      "loss": 0.5748,
+      "step": 1313
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8754133420235147e-05,
+      "loss": 0.7749,
+      "step": 1314
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875191755429e-05,
+      "loss": 0.7904,
+      "step": 1315
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874969985070104e-05,
+      "loss": 0.6348,
+      "step": 1316
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874748030993391e-05,
+      "loss": 0.7524,
+      "step": 1317
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8745258932454658e-05,
+      "loss": 0.8023,
+      "step": 1318
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87430357187297e-05,
+      "loss": 0.8042,
+      "step": 1319
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8740810669225847e-05,
+      "loss": 0.63,
+      "step": 1320
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8738583784410294e-05,
+      "loss": 0.6184,
+      "step": 1321
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8736355064750623e-05,
+      "loss": 0.8473,
+      "step": 1322
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8734124510714795e-05,
+      "loss": 0.5986,
+      "step": 1323
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8731892122771165e-05,
+      "loss": 0.659,
+      "step": 1324
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872965790138846e-05,
+      "loss": 0.7915,
+      "step": 1325
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872742184703581e-05,
+      "loss": 0.7431,
+      "step": 1326
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872518396018272e-05,
+      "loss": 0.8034,
+      "step": 1327
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8722944241299074e-05,
+      "loss": 0.7999,
+      "step": 1328
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872070269085515e-05,
+      "loss": 0.7816,
+      "step": 1329
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87184593093216e-05,
+      "loss": 0.7805,
+      "step": 1330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8716214097169482e-05,
+      "loss": 0.7584,
+      "step": 1331
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8713967054870212e-05,
+      "loss": 0.7564,
+      "step": 1332
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871171818289561e-05,
+      "loss": 0.7672,
+      "step": 1333
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870946748171787e-05,
+      "loss": 0.7576,
+      "step": 1334
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870721495180957e-05,
+      "loss": 0.7969,
+      "step": 1335
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8704960593643677e-05,
+      "loss": 0.7199,
+      "step": 1336
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870270440769354e-05,
+      "loss": 0.7471,
+      "step": 1337
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8700446394432893e-05,
+      "loss": 0.7753,
+      "step": 1338
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8698186554335845e-05,
+      "loss": 0.7638,
+      "step": 1339
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86959248878769e-05,
+      "loss": 0.7261,
+      "step": 1340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8693661395530946e-05,
+      "loss": 0.6234,
+      "step": 1341
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869139607777324e-05,
+      "loss": 0.7722,
+      "step": 1342
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868912893507944e-05,
+      "loss": 0.7393,
+      "step": 1343
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8686859967925576e-05,
+      "loss": 0.7822,
+      "step": 1344
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8684589176788062e-05,
+      "loss": 0.7641,
+      "step": 1345
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8682316562143698e-05,
+      "loss": 0.6262,
+      "step": 1346
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8680042124469668e-05,
+      "loss": 0.6638,
+      "step": 1347
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8677765864243532e-05,
+      "loss": 0.8155,
+      "step": 1348
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867548778194324e-05,
+      "loss": 0.7679,
+      "step": 1349
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8673207878047125e-05,
+      "loss": 0.7755,
+      "step": 1350
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8670926153033895e-05,
+      "loss": 0.7735,
+      "step": 1351
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8668642607382645e-05,
+      "loss": 0.7628,
+      "step": 1352
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8666357241572855e-05,
+      "loss": 0.8202,
+      "step": 1353
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8664070056084384e-05,
+      "loss": 0.8094,
+      "step": 1354
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866178105139747e-05,
+      "loss": 0.7777,
+      "step": 1355
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8659490227992737e-05,
+      "loss": 0.7605,
+      "step": 1356
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8657197586351192e-05,
+      "loss": 0.7766,
+      "step": 1357
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865490312695422e-05,
+      "loss": 0.787,
+      "step": 1358
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8652606850283593e-05,
+      "loss": 0.6502,
+      "step": 1359
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8650308756821457e-05,
+      "loss": 0.7757,
+      "step": 1360
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864800884705035e-05,
+      "loss": 0.6016,
+      "step": 1361
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8645707121453177e-05,
+      "loss": 0.7695,
+      "step": 1362
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8643403580513234e-05,
+      "loss": 0.8162,
+      "step": 1363
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86410982247142e-05,
+      "loss": 0.7389,
+      "step": 1364
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863879105454013e-05,
+      "loss": 0.6306,
+      "step": 1365
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8636482070475463e-05,
+      "loss": 0.6316,
+      "step": 1366
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8634171273005017e-05,
+      "loss": 0.7524,
+      "step": 1367
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8631858662613985e-05,
+      "loss": 0.5667,
+      "step": 1368
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8629544239787955e-05,
+      "loss": 0.7519,
+      "step": 1369
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8627228005012883e-05,
+      "loss": 0.7936,
+      "step": 1370
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862490995877511e-05,
+      "loss": 0.8034,
+      "step": 1371
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862259010156136e-05,
+      "loss": 0.7977,
+      "step": 1372
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862026843385873e-05,
+      "loss": 0.7821,
+      "step": 1373
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8617944956154705e-05,
+      "loss": 0.7681,
+      "step": 1374
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8615619668937145e-05,
+      "loss": 0.754,
+      "step": 1375
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8613292572694295e-05,
+      "loss": 0.8192,
+      "step": 1376
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861096366791477e-05,
+      "loss": 0.798,
+      "step": 1377
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8608632955087576e-05,
+      "loss": 0.7773,
+      "step": 1378
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860630043470209e-05,
+      "loss": 0.7863,
+      "step": 1379
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8603966107248077e-05,
+      "loss": 0.81,
+      "step": 1380
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8601629973215673e-05,
+      "loss": 0.7685,
+      "step": 1381
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8599292033095398e-05,
+      "loss": 0.6329,
+      "step": 1382
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859695228737815e-05,
+      "loss": 0.7752,
+      "step": 1383
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859461073655521e-05,
+      "loss": 0.6303,
+      "step": 1384
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859226738111823e-05,
+      "loss": 0.78,
+      "step": 1385
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8589922221559248e-05,
+      "loss": 0.7598,
+      "step": 1386
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8587575258370673e-05,
+      "loss": 0.6496,
+      "step": 1387
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8585226492045306e-05,
+      "loss": 0.7837,
+      "step": 1388
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8582875923076312e-05,
+      "loss": 0.7537,
+      "step": 1389
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8580523551957242e-05,
+      "loss": 0.7613,
+      "step": 1390
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8578169379182027e-05,
+      "loss": 0.753,
+      "step": 1391
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857581340524497e-05,
+      "loss": 0.7985,
+      "step": 1392
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857345563064076e-05,
+      "loss": 0.8045,
+      "step": 1393
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857109605586446e-05,
+      "loss": 0.819,
+      "step": 1394
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8568734681411508e-05,
+      "loss": 0.7987,
+      "step": 1395
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856637150777772e-05,
+      "loss": 0.8008,
+      "step": 1396
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8564006535459298e-05,
+      "loss": 0.7839,
+      "step": 1397
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8561639764952814e-05,
+      "loss": 0.7955,
+      "step": 1398
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855927119675522e-05,
+      "loss": 0.8022,
+      "step": 1399
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8556900831363846e-05,
+      "loss": 0.774,
+      "step": 1400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85545286692764e-05,
+      "loss": 0.7935,
+      "step": 1401
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855215471099096e-05,
+      "loss": 0.6298,
+      "step": 1402
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8549778957005996e-05,
+      "loss": 0.772,
+      "step": 1403
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8547401407820337e-05,
+      "loss": 0.7914,
+      "step": 1404
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8545022063933206e-05,
+      "loss": 0.6526,
+      "step": 1405
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8542640925844186e-05,
+      "loss": 0.7453,
+      "step": 1406
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8540257994053254e-05,
+      "loss": 0.7896,
+      "step": 1407
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8537873269060748e-05,
+      "loss": 0.783,
+      "step": 1408
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8535486751367397e-05,
+      "loss": 0.6347,
+      "step": 1409
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8533098441474292e-05,
+      "loss": 0.8057,
+      "step": 1410
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8530708339882914e-05,
+      "loss": 0.7333,
+      "step": 1411
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852831644709511e-05,
+      "loss": 0.8163,
+      "step": 1412
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8525922763613106e-05,
+      "loss": 0.7909,
+      "step": 1413
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852352728993951e-05,
+      "loss": 0.7478,
+      "step": 1414
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852113002657729e-05,
+      "loss": 0.7508,
+      "step": 1415
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851873097402981e-05,
+      "loss": 0.7597,
+      "step": 1416
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85163301328008e-05,
+      "loss": 0.8043,
+      "step": 1417
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851392750339436e-05,
+      "loss": 0.773,
+      "step": 1418
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851152308631497e-05,
+      "loss": 0.7996,
+      "step": 1419
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85091168820675e-05,
+      "loss": 0.7904,
+      "step": 1420
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8506708891157162e-05,
+      "loss": 0.7532,
+      "step": 1421
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850429911408958e-05,
+      "loss": 0.6191,
+      "step": 1422
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8501887551370728e-05,
+      "loss": 0.6435,
+      "step": 1423
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849947420350696e-05,
+      "loss": 0.7457,
+      "step": 1424
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8497059071005014e-05,
+      "loss": 0.7568,
+      "step": 1425
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8494642154371994e-05,
+      "loss": 0.7825,
+      "step": 1426
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8492223454115378e-05,
+      "loss": 0.7541,
+      "step": 1427
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8489802970743027e-05,
+      "loss": 0.7343,
+      "step": 1428
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8487380704763166e-05,
+      "loss": 0.6593,
+      "step": 1429
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8484956656684403e-05,
+      "loss": 0.7707,
+      "step": 1430
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8482530827015716e-05,
+      "loss": 0.7697,
+      "step": 1431
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8480103216266456e-05,
+      "loss": 0.7976,
+      "step": 1432
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8477673824946343e-05,
+      "loss": 0.7641,
+      "step": 1433
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8475242653565493e-05,
+      "loss": 0.6692,
+      "step": 1434
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8472809702634364e-05,
+      "loss": 0.7947,
+      "step": 1435
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8470374972663814e-05,
+      "loss": 0.7669,
+      "step": 1436
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8467938464165063e-05,
+      "loss": 0.7904,
+      "step": 1437
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8465500177649702e-05,
+      "loss": 0.7551,
+      "step": 1438
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.84630601136297e-05,
+      "loss": 0.7584,
+      "step": 1439
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8460618272617398e-05,
+      "loss": 0.6388,
+      "step": 1440
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8458174655125515e-05,
+      "loss": 0.8039,
+      "step": 1441
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8455729261667135e-05,
+      "loss": 0.6044,
+      "step": 1442
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8453282092755714e-05,
+      "loss": 0.7926,
+      "step": 1443
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8450833148905092e-05,
+      "loss": 0.7714,
+      "step": 1444
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8448382430629475e-05,
+      "loss": 0.7984,
+      "step": 1445
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8445929938443433e-05,
+      "loss": 0.7665,
+      "step": 1446
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8443475672861925e-05,
+      "loss": 0.7586,
+      "step": 1447
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844101963440027e-05,
+      "loss": 0.7424,
+      "step": 1448
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8438561823574166e-05,
+      "loss": 0.7462,
+      "step": 1449
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843610224089968e-05,
+      "loss": 0.7592,
+      "step": 1450
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8433640886893252e-05,
+      "loss": 0.7588,
+      "step": 1451
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8431177762071688e-05,
+      "loss": 0.617,
+      "step": 1452
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8428712866952182e-05,
+      "loss": 0.7992,
+      "step": 1453
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8426246202052275e-05,
+      "loss": 0.7364,
+      "step": 1454
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842377776788991e-05,
+      "loss": 0.7713,
+      "step": 1455
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842130756498337e-05,
+      "loss": 0.7745,
+      "step": 1456
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8418835593851336e-05,
+      "loss": 0.78,
+      "step": 1457
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8416361855012843e-05,
+      "loss": 0.7839,
+      "step": 1458
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8413886348987303e-05,
+      "loss": 0.7563,
+      "step": 1459
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8411409076294503e-05,
+      "loss": 0.7431,
+      "step": 1460
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8408930037454592e-05,
+      "loss": 0.7942,
+      "step": 1461
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8406449232988096e-05,
+      "loss": 0.7599,
+      "step": 1462
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8403966663415912e-05,
+      "loss": 0.7705,
+      "step": 1463
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840148232925931e-05,
+      "loss": 0.7854,
+      "step": 1464
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8398996231039916e-05,
+      "loss": 0.745,
+      "step": 1465
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839650836927975e-05,
+      "loss": 0.7998,
+      "step": 1466
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8394018744501177e-05,
+      "loss": 0.7579,
+      "step": 1467
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8391527357226952e-05,
+      "loss": 0.7798,
+      "step": 1468
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8389034207980195e-05,
+      "loss": 0.7809,
+      "step": 1469
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8386539297284388e-05,
+      "loss": 0.7584,
+      "step": 1470
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8384042625663387e-05,
+      "loss": 0.7559,
+      "step": 1471
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8381544193641426e-05,
+      "loss": 0.7281,
+      "step": 1472
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8379044001743094e-05,
+      "loss": 0.7934,
+      "step": 1473
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8376542050493363e-05,
+      "loss": 0.7952,
+      "step": 1474
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837403834041757e-05,
+      "loss": 0.8015,
+      "step": 1475
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8371532872041418e-05,
+      "loss": 0.7613,
+      "step": 1476
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8369025645890976e-05,
+      "loss": 0.6184,
+      "step": 1477
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8366516662492696e-05,
+      "loss": 0.6175,
+      "step": 1478
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8364005922373383e-05,
+      "loss": 0.7536,
+      "step": 1479
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8361493426060227e-05,
+      "loss": 0.7468,
+      "step": 1480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8358979174080765e-05,
+      "loss": 0.7864,
+      "step": 1481
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8356463166962928e-05,
+      "loss": 0.6117,
+      "step": 1482
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8353945405234996e-05,
+      "loss": 0.7715,
+      "step": 1483
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835142588942563e-05,
+      "loss": 0.7599,
+      "step": 1484
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8348904620063843e-05,
+      "loss": 0.8218,
+      "step": 1485
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834638159767904e-05,
+      "loss": 0.6231,
+      "step": 1486
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8343856822800973e-05,
+      "loss": 0.6097,
+      "step": 1487
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8341330295959772e-05,
+      "loss": 0.7557,
+      "step": 1488
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8338802017685933e-05,
+      "loss": 0.776,
+      "step": 1489
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8336271988510323e-05,
+      "loss": 0.7748,
+      "step": 1490
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833374020896417e-05,
+      "loss": 0.7544,
+      "step": 1491
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833120667957907e-05,
+      "loss": 0.802,
+      "step": 1492
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8328671400886993e-05,
+      "loss": 0.6019,
+      "step": 1493
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832613437342027e-05,
+      "loss": 0.576,
+      "step": 1494
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8323595597711606e-05,
+      "loss": 0.5838,
+      "step": 1495
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8321055074294066e-05,
+      "loss": 0.7794,
+      "step": 1496
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8318512803701087e-05,
+      "loss": 0.7737,
+      "step": 1497
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8315968786466464e-05,
+      "loss": 0.7362,
+      "step": 1498
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8313423023124375e-05,
+      "loss": 0.7683,
+      "step": 1499
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8310875514209348e-05,
+      "loss": 0.7903,
+      "step": 1500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8308326260256287e-05,
+      "loss": 0.7528,
+      "step": 1501
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8305775261800457e-05,
+      "loss": 0.7788,
+      "step": 1502
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8303222519377498e-05,
+      "loss": 0.7433,
+      "step": 1503
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8300668033523404e-05,
+      "loss": 0.6316,
+      "step": 1504
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829811180477455e-05,
+      "loss": 0.6348,
+      "step": 1505
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8295553833667658e-05,
+      "loss": 0.7473,
+      "step": 1506
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8292994120739833e-05,
+      "loss": 0.5921,
+      "step": 1507
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8290432666528537e-05,
+      "loss": 0.7749,
+      "step": 1508
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.82878694715716e-05,
+      "loss": 0.815,
+      "step": 1509
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828530453640722e-05,
+      "loss": 0.7822,
+      "step": 1510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828273786157395e-05,
+      "loss": 0.7788,
+      "step": 1511
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8280169447610722e-05,
+      "loss": 0.7984,
+      "step": 1512
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8277599295056826e-05,
+      "loss": 0.7517,
+      "step": 1513
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8275027404451918e-05,
+      "loss": 0.6765,
+      "step": 1514
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8272453776336016e-05,
+      "loss": 0.7933,
+      "step": 1515
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826987841124951e-05,
+      "loss": 0.7806,
+      "step": 1516
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826730130973315e-05,
+      "loss": 0.7575,
+      "step": 1517
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8264722472328043e-05,
+      "loss": 0.5853,
+      "step": 1518
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8262141899575683e-05,
+      "loss": 0.7703,
+      "step": 1519
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8259559592017905e-05,
+      "loss": 0.7826,
+      "step": 1520
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8256975550196915e-05,
+      "loss": 0.5862,
+      "step": 1521
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825438977465529e-05,
+      "loss": 0.8007,
+      "step": 1522
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8251802265935964e-05,
+      "loss": 0.7719,
+      "step": 1523
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824921302458224e-05,
+      "loss": 0.7709,
+      "step": 1524
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824662205113778e-05,
+      "loss": 0.7991,
+      "step": 1525
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8244029346146613e-05,
+      "loss": 0.7663,
+      "step": 1526
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8241434910153132e-05,
+      "loss": 0.7044,
+      "step": 1527
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8238838743702086e-05,
+      "loss": 0.7706,
+      "step": 1528
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8236240847338597e-05,
+      "loss": 0.7524,
+      "step": 1529
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8233641221608146e-05,
+      "loss": 0.7941,
+      "step": 1530
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8231039867056578e-05,
+      "loss": 0.7913,
+      "step": 1531
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8228436784230104e-05,
+      "loss": 0.7687,
+      "step": 1532
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8225831973675285e-05,
+      "loss": 0.7963,
+      "step": 1533
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8223225435939063e-05,
+      "loss": 0.7861,
+      "step": 1534
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822061717156873e-05,
+      "loss": 0.7553,
+      "step": 1535
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8218007181111944e-05,
+      "loss": 0.6029,
+      "step": 1536
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8215395465116726e-05,
+      "loss": 0.5953,
+      "step": 1537
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8212782024131457e-05,
+      "loss": 0.7531,
+      "step": 1538
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8210166858704893e-05,
+      "loss": 0.7559,
+      "step": 1539
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8207549969386127e-05,
+      "loss": 0.7624,
+      "step": 1540
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8204931356724636e-05,
+      "loss": 0.7582,
+      "step": 1541
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8202311021270248e-05,
+      "loss": 0.7673,
+      "step": 1542
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8199688963573163e-05,
+      "loss": 0.7913,
+      "step": 1543
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819706518418393e-05,
+      "loss": 0.6124,
+      "step": 1544
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819443968365346e-05,
+      "loss": 0.7628,
+      "step": 1545
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819181246253304e-05,
+      "loss": 0.7847,
+      "step": 1546
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8189183521374304e-05,
+      "loss": 0.7695,
+      "step": 1547
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8186552860729255e-05,
+      "loss": 0.7811,
+      "step": 1548
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8183920481150247e-05,
+      "loss": 0.7957,
+      "step": 1549
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818128638319001e-05,
+      "loss": 0.7549,
+      "step": 1550
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8178650567401624e-05,
+      "loss": 0.7765,
+      "step": 1551
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817601303433853e-05,
+      "loss": 0.7274,
+      "step": 1552
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817337378455453e-05,
+      "loss": 0.6249,
+      "step": 1553
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8170732818603797e-05,
+      "loss": 0.7699,
+      "step": 1554
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8168090137040846e-05,
+      "loss": 0.7842,
+      "step": 1555
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816544574042057e-05,
+      "loss": 0.7431,
+      "step": 1556
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8162799629298207e-05,
+      "loss": 0.7748,
+      "step": 1557
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816015180422937e-05,
+      "loss": 0.6099,
+      "step": 1558
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8157502265770017e-05,
+      "loss": 0.7908,
+      "step": 1559
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8154851014476475e-05,
+      "loss": 0.5977,
+      "step": 1560
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8152198050905426e-05,
+      "loss": 0.8238,
+      "step": 1561
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8149543375613922e-05,
+      "loss": 0.624,
+      "step": 1562
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814688698915936e-05,
+      "loss": 0.7576,
+      "step": 1563
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81442288920995e-05,
+      "loss": 0.7779,
+      "step": 1564
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8141569084992466e-05,
+      "loss": 0.7557,
+      "step": 1565
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8138907568396746e-05,
+      "loss": 0.7519,
+      "step": 1566
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8136244342871166e-05,
+      "loss": 0.7852,
+      "step": 1567
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813357940897494e-05,
+      "loss": 0.7603,
+      "step": 1568
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8130912767267613e-05,
+      "loss": 0.7501,
+      "step": 1569
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8128244418309102e-05,
+      "loss": 0.8071,
+      "step": 1570
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812557436265969e-05,
+      "loss": 0.7947,
+      "step": 1571
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8122902600880007e-05,
+      "loss": 0.7956,
+      "step": 1572
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8120229133531037e-05,
+      "loss": 0.7817,
+      "step": 1573
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8117553961174132e-05,
+      "loss": 0.7604,
+      "step": 1574
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8114877084371002e-05,
+      "loss": 0.7812,
+      "step": 1575
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8112198503683714e-05,
+      "loss": 0.7805,
+      "step": 1576
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810951821967468e-05,
+      "loss": 0.7562,
+      "step": 1577
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8106836232906694e-05,
+      "loss": 0.8091,
+      "step": 1578
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810415254394289e-05,
+      "loss": 0.7439,
+      "step": 1579
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8101467153346756e-05,
+      "loss": 0.768,
+      "step": 1580
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8098780061682148e-05,
+      "loss": 0.7998,
+      "step": 1581
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809609126951328e-05,
+      "loss": 0.7916,
+      "step": 1582
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8093400777404713e-05,
+      "loss": 0.7817,
+      "step": 1583
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8090708585921378e-05,
+      "loss": 0.6252,
+      "step": 1584
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8088014695628547e-05,
+      "loss": 0.6107,
+      "step": 1585
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8085319107091866e-05,
+      "loss": 0.6628,
+      "step": 1586
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808262182087732e-05,
+      "loss": 0.7633,
+      "step": 1587
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8079922837551267e-05,
+      "loss": 0.7879,
+      "step": 1588
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8077222157680408e-05,
+      "loss": 0.7735,
+      "step": 1589
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8074519781831813e-05,
+      "loss": 0.791,
+      "step": 1590
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.807181571057289e-05,
+      "loss": 0.7533,
+      "step": 1591
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8069109944471427e-05,
+      "loss": 0.7648,
+      "step": 1592
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8066402484095544e-05,
+      "loss": 0.7879,
+      "step": 1593
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8063693330013732e-05,
+      "loss": 0.7559,
+      "step": 1594
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8060982482794834e-05,
+      "loss": 0.7392,
+      "step": 1595
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8058269943008046e-05,
+      "loss": 0.7637,
+      "step": 1596
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8055555711222924e-05,
+      "loss": 0.7626,
+      "step": 1597
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805283978800937e-05,
+      "loss": 0.8014,
+      "step": 1598
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8050122173937655e-05,
+      "loss": 0.7423,
+      "step": 1599
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804740286957839e-05,
+      "loss": 0.5934,
+      "step": 1600
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804468187550256e-05,
+      "loss": 0.739,
+      "step": 1601
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8041959192281482e-05,
+      "loss": 0.8097,
+      "step": 1602
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8039234820486845e-05,
+      "loss": 0.7549,
+      "step": 1603
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8036508760690682e-05,
+      "loss": 0.8004,
+      "step": 1604
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803378101346539e-05,
+      "loss": 0.7626,
+      "step": 1605
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803105157938371e-05,
+      "loss": 0.7881,
+      "step": 1606
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8028320459018744e-05,
+      "loss": 0.637,
+      "step": 1607
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802558765294395e-05,
+      "loss": 0.7204,
+      "step": 1608
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8022853161733136e-05,
+      "loss": 0.7499,
+      "step": 1609
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802011698596046e-05,
+      "loss": 0.7537,
+      "step": 1610
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801737912620044e-05,
+      "loss": 0.8023,
+      "step": 1611
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8014639583027945e-05,
+      "loss": 0.776,
+      "step": 1612
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.80118983570182e-05,
+      "loss": 0.7973,
+      "step": 1613
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8009155448746784e-05,
+      "loss": 0.7619,
+      "step": 1614
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800641085878962e-05,
+      "loss": 0.7433,
+      "step": 1615
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8003664587722992e-05,
+      "loss": 0.7485,
+      "step": 1616
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800091663612354e-05,
+      "loss": 0.6297,
+      "step": 1617
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7998167004568248e-05,
+      "loss": 0.7612,
+      "step": 1618
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799541569363446e-05,
+      "loss": 0.7583,
+      "step": 1619
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7992662703899867e-05,
+      "loss": 0.78,
+      "step": 1620
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798990803594252e-05,
+      "loss": 0.7294,
+      "step": 1621
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7987151690340816e-05,
+      "loss": 0.7546,
+      "step": 1622
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7984393667673505e-05,
+      "loss": 0.7743,
+      "step": 1623
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798163396851969e-05,
+      "loss": 0.7617,
+      "step": 1624
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7978872593458823e-05,
+      "loss": 0.7823,
+      "step": 1625
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797610954307072e-05,
+      "loss": 0.7579,
+      "step": 1626
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7973344817935532e-05,
+      "loss": 0.7434,
+      "step": 1627
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7970578418633773e-05,
+      "loss": 0.6207,
+      "step": 1628
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7967810345746305e-05,
+      "loss": 0.5921,
+      "step": 1629
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7965040599854338e-05,
+      "loss": 0.764,
+      "step": 1630
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796226918153944e-05,
+      "loss": 0.81,
+      "step": 1631
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7959496091383527e-05,
+      "loss": 0.7502,
+      "step": 1632
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795672132996887e-05,
+      "loss": 0.6751,
+      "step": 1633
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7953944897878073e-05,
+      "loss": 0.7943,
+      "step": 1634
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7951166795694117e-05,
+      "loss": 0.6098,
+      "step": 1635
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794838702400032e-05,
+      "loss": 0.7999,
+      "step": 1636
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794560558338035e-05,
+      "loss": 0.6196,
+      "step": 1637
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794282247441822e-05,
+      "loss": 0.7807,
+      "step": 1638
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7940037697698317e-05,
+      "loss": 0.7914,
+      "step": 1639
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7937251253805347e-05,
+      "loss": 0.7478,
+      "step": 1640
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7934463143324386e-05,
+      "loss": 0.7854,
+      "step": 1641
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7931673366840855e-05,
+      "loss": 0.7855,
+      "step": 1642
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7928881924940526e-05,
+      "loss": 0.7625,
+      "step": 1643
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7926088818209515e-05,
+      "loss": 0.7284,
+      "step": 1644
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7923294047234296e-05,
+      "loss": 0.7852,
+      "step": 1645
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792049761260169e-05,
+      "loss": 0.7401,
+      "step": 1646
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7917699514898857e-05,
+      "loss": 0.7736,
+      "step": 1647
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7914899754713323e-05,
+      "loss": 0.7293,
+      "step": 1648
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7912098332632947e-05,
+      "loss": 0.6406,
+      "step": 1649
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790929524924595e-05,
+      "loss": 0.7758,
+      "step": 1650
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79064905051409e-05,
+      "loss": 0.6593,
+      "step": 1651
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7903684100906705e-05,
+      "loss": 0.7596,
+      "step": 1652
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790087603713263e-05,
+      "loss": 0.6343,
+      "step": 1653
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7898066314408284e-05,
+      "loss": 0.5853,
+      "step": 1654
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789525493332362e-05,
+      "loss": 0.7936,
+      "step": 1655
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7892441894468954e-05,
+      "loss": 0.6583,
+      "step": 1656
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7889627198434937e-05,
+      "loss": 0.7624,
+      "step": 1657
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7886810845812573e-05,
+      "loss": 0.7526,
+      "step": 1658
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7883992837193208e-05,
+      "loss": 0.7607,
+      "step": 1659
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7881173173168548e-05,
+      "loss": 0.7534,
+      "step": 1660
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7878351854330635e-05,
+      "loss": 0.7399,
+      "step": 1661
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787552888127186e-05,
+      "loss": 0.6144,
+      "step": 1662
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7872704254584968e-05,
+      "loss": 0.744,
+      "step": 1663
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7869877974863047e-05,
+      "loss": 0.7318,
+      "step": 1664
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7867050042699527e-05,
+      "loss": 0.7668,
+      "step": 1665
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7864220458688198e-05,
+      "loss": 0.7642,
+      "step": 1666
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7861389223423183e-05,
+      "loss": 0.7543,
+      "step": 1667
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785855633749896e-05,
+      "loss": 0.7293,
+      "step": 1668
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7855721801510347e-05,
+      "loss": 0.7307,
+      "step": 1669
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7852885616052523e-05,
+      "loss": 0.6296,
+      "step": 1670
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7850047781720993e-05,
+      "loss": 0.7325,
+      "step": 1671
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7847208299111623e-05,
+      "loss": 0.7469,
+      "step": 1672
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7844367168820617e-05,
+      "loss": 0.7684,
+      "step": 1673
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7841524391444534e-05,
+      "loss": 0.776,
+      "step": 1674
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783867996758027e-05,
+      "loss": 0.7877,
+      "step": 1675
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7835833897825064e-05,
+      "loss": 0.8096,
+      "step": 1676
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783298618277652e-05,
+      "loss": 0.7689,
+      "step": 1677
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783013682303256e-05,
+      "loss": 0.624,
+      "step": 1678
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782728581919147e-05,
+      "loss": 0.7866,
+      "step": 1679
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782443317185188e-05,
+      "loss": 0.7671,
+      "step": 1680
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782157888161276e-05,
+      "loss": 0.7583,
+      "step": 1681
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7818722949073427e-05,
+      "loss": 0.7305,
+      "step": 1682
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7815865374833537e-05,
+      "loss": 0.7144,
+      "step": 1683
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7813006159493102e-05,
+      "loss": 0.7256,
+      "step": 1684
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781014530365247e-05,
+      "loss": 0.7545,
+      "step": 1685
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780728280791234e-05,
+      "loss": 0.7207,
+      "step": 1686
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7804418672873745e-05,
+      "loss": 0.742,
+      "step": 1687
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7801552899138075e-05,
+      "loss": 0.5712,
+      "step": 1688
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7798685487307055e-05,
+      "loss": 0.7983,
+      "step": 1689
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779581643798275e-05,
+      "loss": 0.7685,
+      "step": 1690
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779294575176759e-05,
+      "loss": 0.7929,
+      "step": 1691
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7790073429264324e-05,
+      "loss": 0.7164,
+      "step": 1692
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7787199471076054e-05,
+      "loss": 0.7257,
+      "step": 1693
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778432387780623e-05,
+      "loss": 0.765,
+      "step": 1694
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7781446650058642e-05,
+      "loss": 0.7805,
+      "step": 1695
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7778567788437423e-05,
+      "loss": 0.6331,
+      "step": 1696
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7775687293547045e-05,
+      "loss": 0.752,
+      "step": 1697
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7772805165992326e-05,
+      "loss": 0.7064,
+      "step": 1698
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776992140637843e-05,
+      "loss": 0.8006,
+      "step": 1699
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776703601531087e-05,
+      "loss": 0.7514,
+      "step": 1700
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776414899339548e-05,
+      "loss": 0.7645,
+      "step": 1701
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7761260341238447e-05,
+      "loss": 0.7501,
+      "step": 1702
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7758370059446318e-05,
+      "loss": 0.7773,
+      "step": 1703
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7755478148625957e-05,
+      "loss": 0.7532,
+      "step": 1704
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775258460938458e-05,
+      "loss": 0.8121,
+      "step": 1705
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774968944232974e-05,
+      "loss": 0.7631,
+      "step": 1706
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7746792648069345e-05,
+      "loss": 0.5982,
+      "step": 1707
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7743894227211633e-05,
+      "loss": 0.7887,
+      "step": 1708
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7740994180365185e-05,
+      "loss": 0.7289,
+      "step": 1709
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7738092508138928e-05,
+      "loss": 0.7782,
+      "step": 1710
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7735189211142124e-05,
+      "loss": 0.749,
+      "step": 1711
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7732284289984385e-05,
+      "loss": 0.7902,
+      "step": 1712
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772937774527565e-05,
+      "loss": 0.7582,
+      "step": 1713
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7726469577626216e-05,
+      "loss": 0.8072,
+      "step": 1714
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772355978764671e-05,
+      "loss": 0.5588,
+      "step": 1715
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7720648375948095e-05,
+      "loss": 0.7327,
+      "step": 1716
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7717735343141687e-05,
+      "loss": 0.7743,
+      "step": 1717
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771482068983914e-05,
+      "loss": 0.7898,
+      "step": 1718
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771190441665244e-05,
+      "loss": 0.7997,
+      "step": 1719
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770898652419392e-05,
+      "loss": 0.7409,
+      "step": 1720
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770606701307625e-05,
+      "loss": 0.7404,
+      "step": 1721
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770314588391244e-05,
+      "loss": 0.7683,
+      "step": 1722
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7700223137315845e-05,
+      "loss": 0.7742,
+      "step": 1723
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769729877390015e-05,
+      "loss": 0.7524,
+      "step": 1724
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7694372794279388e-05,
+      "loss": 0.6155,
+      "step": 1725
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7691445199067926e-05,
+      "loss": 0.78,
+      "step": 1726
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7688515988880476e-05,
+      "loss": 0.7346,
+      "step": 1727
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7685585164332084e-05,
+      "loss": 0.7456,
+      "step": 1728
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7682652726038134e-05,
+      "loss": 0.6078,
+      "step": 1729
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7679718674614355e-05,
+      "loss": 0.7558,
+      "step": 1730
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7676783010676806e-05,
+      "loss": 0.7692,
+      "step": 1731
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7673845734841895e-05,
+      "loss": 0.7676,
+      "step": 1732
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7670906847726363e-05,
+      "loss": 0.7443,
+      "step": 1733
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7667966349947283e-05,
+      "loss": 0.6281,
+      "step": 1734
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766502424212208e-05,
+      "loss": 0.76,
+      "step": 1735
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7662080524868506e-05,
+      "loss": 0.7768,
+      "step": 1736
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7659135198804654e-05,
+      "loss": 0.7806,
+      "step": 1737
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7656188264548958e-05,
+      "loss": 0.672,
+      "step": 1738
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7653239722720182e-05,
+      "loss": 0.7181,
+      "step": 1739
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765028957393744e-05,
+      "loss": 0.7548,
+      "step": 1740
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7647337818820168e-05,
+      "loss": 0.7385,
+      "step": 1741
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7644384457988153e-05,
+      "loss": 0.7398,
+      "step": 1742
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7641429492061513e-05,
+      "loss": 0.7394,
+      "step": 1743
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7638472921660698e-05,
+      "loss": 0.7824,
+      "step": 1744
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7635514747406507e-05,
+      "loss": 0.7968,
+      "step": 1745
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763255496992007e-05,
+      "loss": 0.7745,
+      "step": 1746
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7629593589822843e-05,
+      "loss": 0.7495,
+      "step": 1747
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7626630607736636e-05,
+      "loss": 0.712,
+      "step": 1748
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7623666024283587e-05,
+      "loss": 0.7887,
+      "step": 1749
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7620699840086172e-05,
+      "loss": 0.7803,
+      "step": 1750
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7617732055767196e-05,
+      "loss": 0.7477,
+      "step": 1751
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7614762671949813e-05,
+      "loss": 0.5901,
+      "step": 1752
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7611791689257497e-05,
+      "loss": 0.6277,
+      "step": 1753
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7608819108314076e-05,
+      "loss": 0.7287,
+      "step": 1754
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76058449297437e-05,
+      "loss": 0.8421,
+      "step": 1755
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760286915417086e-05,
+      "loss": 0.7705,
+      "step": 1756
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7599891782220378e-05,
+      "loss": 0.7608,
+      "step": 1757
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7596912814517413e-05,
+      "loss": 0.7652,
+      "step": 1758
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7593932251687467e-05,
+      "loss": 0.7965,
+      "step": 1759
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759095009435636e-05,
+      "loss": 0.6389,
+      "step": 1760
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7587966343150267e-05,
+      "loss": 0.5756,
+      "step": 1761
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758498099869568e-05,
+      "loss": 0.6213,
+      "step": 1762
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758199406161944e-05,
+      "loss": 0.7514,
+      "step": 1763
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757900553254871e-05,
+      "loss": 0.7776,
+      "step": 1764
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7576015412110992e-05,
+      "loss": 0.5847,
+      "step": 1765
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7573023700934125e-05,
+      "loss": 0.622,
+      "step": 1766
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7570030399646284e-05,
+      "loss": 0.79,
+      "step": 1767
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7567035508875964e-05,
+      "loss": 0.7506,
+      "step": 1768
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756403902925201e-05,
+      "loss": 0.7633,
+      "step": 1769
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7561040961403598e-05,
+      "loss": 0.7823,
+      "step": 1770
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7558041305960226e-05,
+      "loss": 0.7616,
+      "step": 1771
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755504006355173e-05,
+      "loss": 0.7395,
+      "step": 1772
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7552037234808296e-05,
+      "loss": 0.6749,
+      "step": 1773
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7549032820360417e-05,
+      "loss": 0.6104,
+      "step": 1774
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7546026820838937e-05,
+      "loss": 0.7738,
+      "step": 1775
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7543019236875023e-05,
+      "loss": 0.7238,
+      "step": 1776
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754001006910018e-05,
+      "loss": 0.7792,
+      "step": 1777
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7536999318146248e-05,
+      "loss": 0.7562,
+      "step": 1778
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753398698464539e-05,
+      "loss": 0.7325,
+      "step": 1779
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7530973069230112e-05,
+      "loss": 0.748,
+      "step": 1780
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7527957572533243e-05,
+      "loss": 0.6233,
+      "step": 1781
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752494049518795e-05,
+      "loss": 0.7462,
+      "step": 1782
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7521921837827732e-05,
+      "loss": 0.7341,
+      "step": 1783
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751890160108641e-05,
+      "loss": 0.7437,
+      "step": 1784
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7515879785598154e-05,
+      "loss": 0.6038,
+      "step": 1785
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7512856391997452e-05,
+      "loss": 0.7438,
+      "step": 1786
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7509831420919126e-05,
+      "loss": 0.7779,
+      "step": 1787
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7506804872998333e-05,
+      "loss": 0.7428,
+      "step": 1788
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7503776748870554e-05,
+      "loss": 0.6265,
+      "step": 1789
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7500747049171613e-05,
+      "loss": 0.7271,
+      "step": 1790
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7497715774537648e-05,
+      "loss": 0.7407,
+      "step": 1791
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7494682925605144e-05,
+      "loss": 0.7733,
+      "step": 1792
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749164850301091e-05,
+      "loss": 0.6308,
+      "step": 1793
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748861250739208e-05,
+      "loss": 0.5977,
+      "step": 1794
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748557493938612e-05,
+      "loss": 0.7583,
+      "step": 1795
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7482535799630845e-05,
+      "loss": 0.747,
+      "step": 1796
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7479495088764368e-05,
+      "loss": 0.7533,
+      "step": 1797
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7476452807425158e-05,
+      "loss": 0.6057,
+      "step": 1798
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7473408956252e-05,
+      "loss": 0.7368,
+      "step": 1799
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7470363535884014e-05,
+      "loss": 0.6013,
+      "step": 1800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7467316546960648e-05,
+      "loss": 0.7143,
+      "step": 1801
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7464267990121682e-05,
+      "loss": 0.7068,
+      "step": 1802
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7461217866007215e-05,
+      "loss": 0.723,
+      "step": 1803
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7458166175257695e-05,
+      "loss": 0.7701,
+      "step": 1804
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7455112918513872e-05,
+      "loss": 0.5505,
+      "step": 1805
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7452058096416853e-05,
+      "loss": 0.761,
+      "step": 1806
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744900170960805e-05,
+      "loss": 0.7489,
+      "step": 1807
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7445943758729223e-05,
+      "loss": 0.7986,
+      "step": 1808
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7442884244422448e-05,
+      "loss": 0.7369,
+      "step": 1809
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7439823167330127e-05,
+      "loss": 0.7779,
+      "step": 1810
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7436760528095e-05,
+      "loss": 0.7628,
+      "step": 1811
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7433696327360133e-05,
+      "loss": 0.7197,
+      "step": 1812
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7430630565768916e-05,
+      "loss": 0.7139,
+      "step": 1813
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7427563243965063e-05,
+      "loss": 0.6203,
+      "step": 1814
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742449436259263e-05,
+      "loss": 0.7627,
+      "step": 1815
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7421423922295984e-05,
+      "loss": 0.7357,
+      "step": 1816
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741835192371983e-05,
+      "loss": 0.7552,
+      "step": 1817
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7415278367509194e-05,
+      "loss": 0.6914,
+      "step": 1818
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741220325430944e-05,
+      "loss": 0.7217,
+      "step": 1819
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7409126584766238e-05,
+      "loss": 0.7666,
+      "step": 1820
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7406048359525608e-05,
+      "loss": 0.7822,
+      "step": 1821
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740296857923388e-05,
+      "loss": 0.7909,
+      "step": 1822
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739988724453772e-05,
+      "loss": 0.762,
+      "step": 1823
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739680435608412e-05,
+      "loss": 0.7665,
+      "step": 1824
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739371991452039e-05,
+      "loss": 0.7701,
+      "step": 1825
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739063392049417e-05,
+      "loss": 0.7475,
+      "step": 1826
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7387546374653435e-05,
+      "loss": 0.6027,
+      "step": 1827
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7384457277646475e-05,
+      "loss": 0.782,
+      "step": 1828
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7381366630121904e-05,
+      "loss": 0.7648,
+      "step": 1829
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7378274432728673e-05,
+      "loss": 0.7431,
+      "step": 1830
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737518068611605e-05,
+      "loss": 0.7098,
+      "step": 1831
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737208539093363e-05,
+      "loss": 0.6131,
+      "step": 1832
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7368988547831335e-05,
+      "loss": 0.6392,
+      "step": 1833
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7365890157459405e-05,
+      "loss": 0.7828,
+      "step": 1834
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7362790220468414e-05,
+      "loss": 0.7362,
+      "step": 1835
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7359688737509263e-05,
+      "loss": 0.7673,
+      "step": 1836
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7356585709233163e-05,
+      "loss": 0.7221,
+      "step": 1837
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7353481136291657e-05,
+      "loss": 0.7598,
+      "step": 1838
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7350375019336624e-05,
+      "loss": 0.7526,
+      "step": 1839
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7347267359020246e-05,
+      "loss": 0.6231,
+      "step": 1840
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7344158155995047e-05,
+      "loss": 0.7466,
+      "step": 1841
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7341047410913864e-05,
+      "loss": 0.7502,
+      "step": 1842
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733793512442986e-05,
+      "loss": 0.7102,
+      "step": 1843
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733482129719653e-05,
+      "loss": 0.7425,
+      "step": 1844
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7331705929867676e-05,
+      "loss": 0.5431,
+      "step": 1845
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732858902309744e-05,
+      "loss": 0.7728,
+      "step": 1846
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7325470577540277e-05,
+      "loss": 0.7449,
+      "step": 1847
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7322350593850973e-05,
+      "loss": 0.7679,
+      "step": 1848
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7319229072684623e-05,
+      "loss": 0.7665,
+      "step": 1849
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7316106014696667e-05,
+      "loss": 0.759,
+      "step": 1850
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731298142054284e-05,
+      "loss": 0.7194,
+      "step": 1851
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7309855290879224e-05,
+      "loss": 0.7099,
+      "step": 1852
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7306727626362215e-05,
+      "loss": 0.7511,
+      "step": 1853
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7303598427648527e-05,
+      "loss": 0.7637,
+      "step": 1854
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7300467695395196e-05,
+      "loss": 0.6125,
+      "step": 1855
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7297335430259588e-05,
+      "loss": 0.7107,
+      "step": 1856
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7294201632899383e-05,
+      "loss": 0.7421,
+      "step": 1857
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7291066303972588e-05,
+      "loss": 0.6955,
+      "step": 1858
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7287929444137528e-05,
+      "loss": 0.7391,
+      "step": 1859
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7284791054052848e-05,
+      "loss": 0.7568,
+      "step": 1860
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7281651134377525e-05,
+      "loss": 0.7864,
+      "step": 1861
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7278509685770842e-05,
+      "loss": 0.7385,
+      "step": 1862
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7275366708892415e-05,
+      "loss": 0.7734,
+      "step": 1863
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7272222204402173e-05,
+      "loss": 0.7633,
+      "step": 1864
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726907617296037e-05,
+      "loss": 0.6421,
+      "step": 1865
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7265928615227575e-05,
+      "loss": 0.7624,
+      "step": 1866
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726277953186469e-05,
+      "loss": 0.7602,
+      "step": 1867
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7259628923532924e-05,
+      "loss": 0.7534,
+      "step": 1868
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7256476790893817e-05,
+      "loss": 0.7602,
+      "step": 1869
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7253323134609218e-05,
+      "loss": 0.7168,
+      "step": 1870
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.72501679553413e-05,
+      "loss": 0.7677,
+      "step": 1871
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7247011253752564e-05,
+      "loss": 0.7018,
+      "step": 1872
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724385303050582e-05,
+      "loss": 0.7461,
+      "step": 1873
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7240693286264203e-05,
+      "loss": 0.7699,
+      "step": 1874
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7237532021691165e-05,
+      "loss": 0.7325,
+      "step": 1875
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723436923745048e-05,
+      "loss": 0.7622,
+      "step": 1876
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7231204934206234e-05,
+      "loss": 0.7293,
+      "step": 1877
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7228039112622844e-05,
+      "loss": 0.7308,
+      "step": 1878
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722487177336503e-05,
+      "loss": 0.7686,
+      "step": 1879
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722170291709785e-05,
+      "loss": 0.7787,
+      "step": 1880
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7218532544486662e-05,
+      "loss": 0.7369,
+      "step": 1881
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7215360656197155e-05,
+      "loss": 0.7879,
+      "step": 1882
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7212187252895326e-05,
+      "loss": 0.754,
+      "step": 1883
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7209012335247502e-05,
+      "loss": 0.632,
+      "step": 1884
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7205835903920323e-05,
+      "loss": 0.7739,
+      "step": 1885
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720265795958074e-05,
+      "loss": 0.6117,
+      "step": 1886
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7199478502896027e-05,
+      "loss": 0.6052,
+      "step": 1887
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719629753453378e-05,
+      "loss": 0.6582,
+      "step": 1888
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719311505516191e-05,
+      "loss": 0.8073,
+      "step": 1889
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7189931065448634e-05,
+      "loss": 0.7047,
+      "step": 1890
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7186745566062506e-05,
+      "loss": 0.7809,
+      "step": 1891
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7183558557672385e-05,
+      "loss": 0.7562,
+      "step": 1892
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7180370040947443e-05,
+      "loss": 0.7201,
+      "step": 1893
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7177180016557174e-05,
+      "loss": 0.6701,
+      "step": 1894
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71739884851714e-05,
+      "loss": 0.7521,
+      "step": 1895
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7170795447460236e-05,
+      "loss": 0.7615,
+      "step": 1896
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716760090409413e-05,
+      "loss": 0.7677,
+      "step": 1897
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716440485574384e-05,
+      "loss": 0.627,
+      "step": 1898
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7161207303080444e-05,
+      "loss": 0.5849,
+      "step": 1899
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7158008246775334e-05,
+      "loss": 0.8211,
+      "step": 1900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715480768750021e-05,
+      "loss": 0.7765,
+      "step": 1901
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7151605625927106e-05,
+      "loss": 0.764,
+      "step": 1902
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7148402062728352e-05,
+      "loss": 0.742,
+      "step": 1903
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7145196998576607e-05,
+      "loss": 0.7751,
+      "step": 1904
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7141990434144835e-05,
+      "loss": 0.7307,
+      "step": 1905
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713878237010632e-05,
+      "loss": 0.786,
+      "step": 1906
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7135572807134662e-05,
+      "loss": 0.7319,
+      "step": 1907
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7132361745903775e-05,
+      "loss": 0.7772,
+      "step": 1908
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712914918708789e-05,
+      "loss": 0.7904,
+      "step": 1909
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7125935131361543e-05,
+      "loss": 0.6202,
+      "step": 1910
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7122719579399592e-05,
+      "loss": 0.7309,
+      "step": 1911
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711950253187721e-05,
+      "loss": 0.715,
+      "step": 1912
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7116283989469883e-05,
+      "loss": 0.7327,
+      "step": 1913
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7113063952853403e-05,
+      "loss": 0.7376,
+      "step": 1914
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710984242270389e-05,
+      "loss": 0.7512,
+      "step": 1915
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7106619399697763e-05,
+      "loss": 0.7213,
+      "step": 1916
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7103394884511773e-05,
+      "loss": 0.7289,
+      "step": 1917
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710016887782296e-05,
+      "loss": 0.7694,
+      "step": 1918
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7096941380308694e-05,
+      "loss": 0.7576,
+      "step": 1919
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709371239264666e-05,
+      "loss": 0.7513,
+      "step": 1920
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7090481915514843e-05,
+      "loss": 0.7748,
+      "step": 1921
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708724994959155e-05,
+      "loss": 0.7581,
+      "step": 1922
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7084016495555394e-05,
+      "loss": 0.7285,
+      "step": 1923
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7080781554085314e-05,
+      "loss": 0.6135,
+      "step": 1924
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7077545125860548e-05,
+      "loss": 0.7346,
+      "step": 1925
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7074307211560644e-05,
+      "loss": 0.7872,
+      "step": 1926
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 0.7934,
+      "step": 1927
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706782692745522e-05,
+      "loss": 0.7401,
+      "step": 1928
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7064584559010366e-05,
+      "loss": 0.7752,
+      "step": 1929
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7061340707211712e-05,
+      "loss": 0.7095,
+      "step": 1930
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7058095372740374e-05,
+      "loss": 0.6388,
+      "step": 1931
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7054848556277778e-05,
+      "loss": 0.7819,
+      "step": 1932
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7051600258505658e-05,
+      "loss": 0.7591,
+      "step": 1933
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7048350480106055e-05,
+      "loss": 0.7812,
+      "step": 1934
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7045099221761334e-05,
+      "loss": 0.737,
+      "step": 1935
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7041846484154162e-05,
+      "loss": 0.7821,
+      "step": 1936
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7038592267967514e-05,
+      "loss": 0.7849,
+      "step": 1937
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703533657388468e-05,
+      "loss": 0.7612,
+      "step": 1938
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703207940258926e-05,
+      "loss": 0.7157,
+      "step": 1939
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7028820754765167e-05,
+      "loss": 0.6004,
+      "step": 1940
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7025560631096614e-05,
+      "loss": 0.5788,
+      "step": 1941
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702229903226814e-05,
+      "loss": 0.587,
+      "step": 1942
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7019035958964573e-05,
+      "loss": 0.7726,
+      "step": 1943
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7015771411871073e-05,
+      "loss": 0.6368,
+      "step": 1944
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7012505391673087e-05,
+      "loss": 0.782,
+      "step": 1945
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7009237899056394e-05,
+      "loss": 0.5816,
+      "step": 1946
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700596893470706e-05,
+      "loss": 0.8027,
+      "step": 1947
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700269849931148e-05,
+      "loss": 0.6109,
+      "step": 1948
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6999426593556342e-05,
+      "loss": 0.7602,
+      "step": 1949
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6996153218128653e-05,
+      "loss": 0.707,
+      "step": 1950
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6992878373715725e-05,
+      "loss": 0.7604,
+      "step": 1951
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6989602061005176e-05,
+      "loss": 0.7262,
+      "step": 1952
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698632428068494e-05,
+      "loss": 0.7474,
+      "step": 1953
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698304503344325e-05,
+      "loss": 0.7462,
+      "step": 1954
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6979764319968645e-05,
+      "loss": 0.75,
+      "step": 1955
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697648214094999e-05,
+      "loss": 0.7105,
+      "step": 1956
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6973198497076437e-05,
+      "loss": 0.719,
+      "step": 1957
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696991338903746e-05,
+      "loss": 0.7568,
+      "step": 1958
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6966626817522827e-05,
+      "loss": 0.7687,
+      "step": 1959
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6963338783222626e-05,
+      "loss": 0.7494,
+      "step": 1960
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6960049286827248e-05,
+      "loss": 0.7652,
+      "step": 1961
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6956758329027385e-05,
+      "loss": 0.7353,
+      "step": 1962
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6953465910514045e-05,
+      "loss": 0.7415,
+      "step": 1963
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695017203197854e-05,
+      "loss": 0.7578,
+      "step": 1964
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6946876694112478e-05,
+      "loss": 0.7098,
+      "step": 1965
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6943579897607795e-05,
+      "loss": 0.6492,
+      "step": 1966
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6940281643156714e-05,
+      "loss": 0.7508,
+      "step": 1967
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6936981931451766e-05,
+      "loss": 0.7645,
+      "step": 1968
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6933680763185802e-05,
+      "loss": 0.754,
+      "step": 1969
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6930378139051967e-05,
+      "loss": 0.7424,
+      "step": 1970
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6927074059743712e-05,
+      "loss": 0.7419,
+      "step": 1971
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6923768525954796e-05,
+      "loss": 0.6024,
+      "step": 1972
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6920461538379286e-05,
+      "loss": 0.7802,
+      "step": 1973
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691715309771155e-05,
+      "loss": 0.59,
+      "step": 1974
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691384320464626e-05,
+      "loss": 0.6196,
+      "step": 1975
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69105318598784e-05,
+      "loss": 0.7454,
+      "step": 1976
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6907219064103255e-05,
+      "loss": 0.7494,
+      "step": 1977
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690390481801641e-05,
+      "loss": 0.7525,
+      "step": 1978
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6900589122313765e-05,
+      "loss": 0.603,
+      "step": 1979
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689727197769151e-05,
+      "loss": 0.7344,
+      "step": 1980
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689395338484615e-05,
+      "loss": 0.7306,
+      "step": 1981
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689063334447449e-05,
+      "loss": 0.7735,
+      "step": 1982
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688731185727365e-05,
+      "loss": 0.8187,
+      "step": 1983
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6883988923941032e-05,
+      "loss": 0.7493,
+      "step": 1984
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6880664545174356e-05,
+      "loss": 0.7623,
+      "step": 1985
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687733872167165e-05,
+      "loss": 0.7213,
+      "step": 1986
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687401145413123e-05,
+      "loss": 0.741,
+      "step": 1987
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687068274325173e-05,
+      "loss": 0.7667,
+      "step": 1988
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686735258973208e-05,
+      "loss": 0.7229,
+      "step": 1989
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6864020994271513e-05,
+      "loss": 0.7599,
+      "step": 1990
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686068795756956e-05,
+      "loss": 0.7049,
+      "step": 1991
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685735348032607e-05,
+      "loss": 0.77,
+      "step": 1992
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685401756324118e-05,
+      "loss": 0.6919,
+      "step": 1993
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685068020701533e-05,
+      "loss": 0.767,
+      "step": 1994
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6847341412349272e-05,
+      "loss": 0.7213,
+      "step": 1995
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684400117994405e-05,
+      "loss": 0.7345,
+      "step": 1996
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684065951050102e-05,
+      "loss": 0.6449,
+      "step": 1997
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6837316404721825e-05,
+      "loss": 0.73,
+      "step": 1998
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6833971863308424e-05,
+      "loss": 0.7561,
+      "step": 1999
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683062588696307e-05,
+      "loss": 0.7554,
+      "step": 2000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682727847638832e-05,
+      "loss": 0.7485,
+      "step": 2001
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682392963228703e-05,
+      "loss": 0.6994,
+      "step": 2002
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682057935536236e-05,
+      "loss": 0.7626,
+      "step": 2003
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6817227646317773e-05,
+      "loss": 0.7506,
+      "step": 2004
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6813874505857018e-05,
+      "loss": 0.7917,
+      "step": 2005
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6810519934684164e-05,
+      "loss": 0.758,
+      "step": 2006
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680716393350357e-05,
+      "loss": 0.784,
+      "step": 2007
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6803806503019894e-05,
+      "loss": 0.7874,
+      "step": 2008
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6800447643938097e-05,
+      "loss": 0.7437,
+      "step": 2009
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6797087356963445e-05,
+      "loss": 0.7449,
+      "step": 2010
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6793725642801496e-05,
+      "loss": 0.7535,
+      "step": 2011
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679036250215811e-05,
+      "loss": 0.7348,
+      "step": 2012
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6786997935739444e-05,
+      "loss": 0.7729,
+      "step": 2013
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6783631944251965e-05,
+      "loss": 0.7679,
+      "step": 2014
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6780264528402426e-05,
+      "loss": 0.7623,
+      "step": 2015
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6776895688897888e-05,
+      "loss": 0.7772,
+      "step": 2016
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6773525426445706e-05,
+      "loss": 0.7362,
+      "step": 2017
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6770153741753534e-05,
+      "loss": 0.7421,
+      "step": 2018
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6766780635529327e-05,
+      "loss": 0.7219,
+      "step": 2019
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6763406108481342e-05,
+      "loss": 0.7731,
+      "step": 2020
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676003016131812e-05,
+      "loss": 0.7165,
+      "step": 2021
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675665279474852e-05,
+      "loss": 0.7249,
+      "step": 2022
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6753274009481683e-05,
+      "loss": 0.7418,
+      "step": 2023
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6749893806227058e-05,
+      "loss": 0.6998,
+      "step": 2024
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6746512185694385e-05,
+      "loss": 0.6109,
+      "step": 2025
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6743129148593708e-05,
+      "loss": 0.7707,
+      "step": 2026
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6739744695635354e-05,
+      "loss": 0.7692,
+      "step": 2027
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6736358827529973e-05,
+      "loss": 0.7757,
+      "step": 2028
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6732971544988487e-05,
+      "loss": 0.7591,
+      "step": 2029
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672958284872213e-05,
+      "loss": 0.7192,
+      "step": 2030
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6726192739442423e-05,
+      "loss": 0.7618,
+      "step": 2031
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6722801217861192e-05,
+      "loss": 0.7527,
+      "step": 2032
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671940828469056e-05,
+      "loss": 0.6195,
+      "step": 2033
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6716013940642932e-05,
+      "loss": 0.7433,
+      "step": 2034
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6712618186431033e-05,
+      "loss": 0.7684,
+      "step": 2035
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670922102276786e-05,
+      "loss": 0.7859,
+      "step": 2036
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6705822450366726e-05,
+      "loss": 0.762,
+      "step": 2037
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6702422469941216e-05,
+      "loss": 0.7483,
+      "step": 2038
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6699021082205243e-05,
+      "loss": 0.7765,
+      "step": 2039
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669561828787299e-05,
+      "loss": 0.7337,
+      "step": 2040
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669221408765894e-05,
+      "loss": 0.7436,
+      "step": 2041
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668880848227788e-05,
+      "loss": 0.6961,
+      "step": 2042
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6685401472444885e-05,
+      "loss": 0.7317,
+      "step": 2043
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6681993058875324e-05,
+      "loss": 0.7154,
+      "step": 2044
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6678583242284864e-05,
+      "loss": 0.7298,
+      "step": 2045
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6675172023389462e-05,
+      "loss": 0.7519,
+      "step": 2046
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6671759402905383e-05,
+      "loss": 0.6299,
+      "step": 2047
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6668345381549163e-05,
+      "loss": 0.6229,
+      "step": 2048
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6664929960037662e-05,
+      "loss": 0.7312,
+      "step": 2049
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6661513139088002e-05,
+      "loss": 0.5996,
+      "step": 2050
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6658094919417624e-05,
+      "loss": 0.7187,
+      "step": 2051
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665467530174425e-05,
+      "loss": 0.7224,
+      "step": 2052
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6651254286785893e-05,
+      "loss": 0.7719,
+      "step": 2053
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6647831875260877e-05,
+      "loss": 0.733,
+      "step": 2054
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6644408067887795e-05,
+      "loss": 0.7913,
+      "step": 2055
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6640982865385552e-05,
+      "loss": 0.7216,
+      "step": 2056
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6637556268473338e-05,
+      "loss": 0.7431,
+      "step": 2057
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663412827787064e-05,
+      "loss": 0.5934,
+      "step": 2058
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6630698894297224e-05,
+      "loss": 0.7593,
+      "step": 2059
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6627268118473175e-05,
+      "loss": 0.7699,
+      "step": 2060
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6623835951118843e-05,
+      "loss": 0.5961,
+      "step": 2061
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6620402392954884e-05,
+      "loss": 0.7656,
+      "step": 2062
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6616967444702245e-05,
+      "loss": 0.7705,
+      "step": 2063
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661353110708216e-05,
+      "loss": 0.7492,
+      "step": 2064
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6610093380816167e-05,
+      "loss": 0.745,
+      "step": 2065
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6606654266626076e-05,
+      "loss": 0.7624,
+      "step": 2066
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6603213765234006e-05,
+      "loss": 0.7541,
+      "step": 2067
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6599771877362355e-05,
+      "loss": 0.6234,
+      "step": 2068
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6596328603733828e-05,
+      "loss": 0.6294,
+      "step": 2069
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6592883945071402e-05,
+      "loss": 0.7595,
+      "step": 2070
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6589437902098354e-05,
+      "loss": 0.7659,
+      "step": 2071
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6585990475538252e-05,
+      "loss": 0.7119,
+      "step": 2072
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6582541666114954e-05,
+      "loss": 0.7037,
+      "step": 2073
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657909147455261e-05,
+      "loss": 0.7222,
+      "step": 2074
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6575639901575656e-05,
+      "loss": 0.6187,
+      "step": 2075
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657218694790882e-05,
+      "loss": 0.7627,
+      "step": 2076
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.656873261427712e-05,
+      "loss": 0.7507,
+      "step": 2077
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6565276901405866e-05,
+      "loss": 0.7424,
+      "step": 2078
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6561819810020653e-05,
+      "loss": 0.7221,
+      "step": 2079
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655836134084737e-05,
+      "loss": 0.7597,
+      "step": 2080
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655490149461219e-05,
+      "loss": 0.717,
+      "step": 2081
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655144027204158e-05,
+      "loss": 0.7311,
+      "step": 2082
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6547977673862298e-05,
+      "loss": 0.7919,
+      "step": 2083
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6544513700801382e-05,
+      "loss": 0.6926,
+      "step": 2084
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654104835358617e-05,
+      "loss": 0.7237,
+      "step": 2085
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6537581632944272e-05,
+      "loss": 0.7764,
+      "step": 2086
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653411353960361e-05,
+      "loss": 0.7821,
+      "step": 2087
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6530644074292366e-05,
+      "loss": 0.6198,
+      "step": 2088
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652717323773904e-05,
+      "loss": 0.6998,
+      "step": 2089
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6523701030672396e-05,
+      "loss": 0.7388,
+      "step": 2090
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.65202274538215e-05,
+      "loss": 0.7367,
+      "step": 2091
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6516752507915693e-05,
+      "loss": 0.7506,
+      "step": 2092
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6513276193684618e-05,
+      "loss": 0.7233,
+      "step": 2093
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.65097985118582e-05,
+      "loss": 0.7485,
+      "step": 2094
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6506319463166635e-05,
+      "loss": 0.7757,
+      "step": 2095
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6502839048340435e-05,
+      "loss": 0.7558,
+      "step": 2096
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6499357268110382e-05,
+      "loss": 0.7503,
+      "step": 2097
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.649587412320754e-05,
+      "loss": 0.7551,
+      "step": 2098
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.649238961436327e-05,
+      "loss": 0.7215,
+      "step": 2099
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648890374230922e-05,
+      "loss": 0.6245,
+      "step": 2100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6485416507777316e-05,
+      "loss": 0.7356,
+      "step": 2101
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6481927911499774e-05,
+      "loss": 0.7408,
+      "step": 2102
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6478437954209096e-05,
+      "loss": 0.753,
+      "step": 2103
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647494663663807e-05,
+      "loss": 0.7619,
+      "step": 2104
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647145395951977e-05,
+      "loss": 0.7419,
+      "step": 2105
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6467959923587555e-05,
+      "loss": 0.7689,
+      "step": 2106
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6464464529575068e-05,
+      "loss": 0.7621,
+      "step": 2107
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646096777821624e-05,
+      "loss": 0.5987,
+      "step": 2108
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6457469670245284e-05,
+      "loss": 0.7407,
+      "step": 2109
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6453970206396705e-05,
+      "loss": 0.7437,
+      "step": 2110
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6450469387405277e-05,
+      "loss": 0.7313,
+      "step": 2111
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6446967214006072e-05,
+      "loss": 0.7606,
+      "step": 2112
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6443463686934445e-05,
+      "loss": 0.8049,
+      "step": 2113
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6439958806926032e-05,
+      "loss": 0.7247,
+      "step": 2114
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6436452574716754e-05,
+      "loss": 0.7222,
+      "step": 2115
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643294499104282e-05,
+      "loss": 0.7258,
+      "step": 2116
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6429436056640708e-05,
+      "loss": 0.7345,
+      "step": 2117
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6425925772247202e-05,
+      "loss": 0.7518,
+      "step": 2118
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6422414138599353e-05,
+      "loss": 0.7469,
+      "step": 2119
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6418901156434502e-05,
+      "loss": 0.7594,
+      "step": 2120
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6415386826490264e-05,
+      "loss": 0.7463,
+      "step": 2121
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641187114950456e-05,
+      "loss": 0.7508,
+      "step": 2122
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640835412621556e-05,
+      "loss": 0.7227,
+      "step": 2123
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6404835757361745e-05,
+      "loss": 0.7709,
+      "step": 2124
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6401316043681867e-05,
+      "loss": 0.7263,
+      "step": 2125
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6397794985914964e-05,
+      "loss": 0.7407,
+      "step": 2126
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639427258480035e-05,
+      "loss": 0.7818,
+      "step": 2127
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6390748841077625e-05,
+      "loss": 0.7397,
+      "step": 2128
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6387223755486674e-05,
+      "loss": 0.8163,
+      "step": 2129
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6383697328767655e-05,
+      "loss": 0.7828,
+      "step": 2130
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6380169561661023e-05,
+      "loss": 0.7726,
+      "step": 2131
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6376640454907495e-05,
+      "loss": 0.588,
+      "step": 2132
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6373110009248087e-05,
+      "loss": 0.7489,
+      "step": 2133
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6369578225424083e-05,
+      "loss": 0.7412,
+      "step": 2134
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6366045104177052e-05,
+      "loss": 0.7379,
+      "step": 2135
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636251064624885e-05,
+      "loss": 0.5802,
+      "step": 2136
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6358974852381607e-05,
+      "loss": 0.7216,
+      "step": 2137
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635543772331773e-05,
+      "loss": 0.7097,
+      "step": 2138
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635189925979992e-05,
+      "loss": 0.7712,
+      "step": 2139
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6348359462571146e-05,
+      "loss": 0.7282,
+      "step": 2140
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634481833237466e-05,
+      "loss": 0.7711,
+      "step": 2141
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634127586995399e-05,
+      "loss": 0.6875,
+      "step": 2142
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6337732076052956e-05,
+      "loss": 0.7525,
+      "step": 2143
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633418695141565e-05,
+      "loss": 0.7932,
+      "step": 2144
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633064049678643e-05,
+      "loss": 0.7354,
+      "step": 2145
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6327092712909966e-05,
+      "loss": 0.7492,
+      "step": 2146
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6323543600531175e-05,
+      "loss": 0.7228,
+      "step": 2147
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631999316039527e-05,
+      "loss": 0.7577,
+      "step": 2148
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631644139324774e-05,
+      "loss": 0.7501,
+      "step": 2149
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6312888299834347e-05,
+      "loss": 0.7344,
+      "step": 2150
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6309333880901136e-05,
+      "loss": 0.6215,
+      "step": 2151
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630577813719443e-05,
+      "loss": 0.7575,
+      "step": 2152
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630222106946083e-05,
+      "loss": 0.5966,
+      "step": 2153
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629866267844722e-05,
+      "loss": 0.5856,
+      "step": 2154
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629510296490075e-05,
+      "loss": 0.6017,
+      "step": 2155
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6291541929568858e-05,
+      "loss": 0.8025,
+      "step": 2156
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6287979573199257e-05,
+      "loss": 0.7759,
+      "step": 2157
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6284415896539933e-05,
+      "loss": 0.7502,
+      "step": 2158
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6280850900339155e-05,
+      "loss": 0.7267,
+      "step": 2159
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6277284585345466e-05,
+      "loss": 0.7292,
+      "step": 2160
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6273716952307687e-05,
+      "loss": 0.7638,
+      "step": 2161
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6270148001974918e-05,
+      "loss": 0.7353,
+      "step": 2162
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626657773509653e-05,
+      "loss": 0.7704,
+      "step": 2163
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626300615242217e-05,
+      "loss": 0.6708,
+      "step": 2164
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625943325470177e-05,
+      "loss": 0.7536,
+      "step": 2165
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6255859042685533e-05,
+      "loss": 0.7245,
+      "step": 2166
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6252283517123935e-05,
+      "loss": 0.7535,
+      "step": 2167
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624870667876773e-05,
+      "loss": 0.6451,
+      "step": 2168
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6245128528367956e-05,
+      "loss": 0.7051,
+      "step": 2169
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6241549066675907e-05,
+      "loss": 0.615,
+      "step": 2170
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6237968294443172e-05,
+      "loss": 0.6276,
+      "step": 2171
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6234386212421606e-05,
+      "loss": 0.7469,
+      "step": 2172
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623080282136334e-05,
+      "loss": 0.7421,
+      "step": 2173
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6227218122020777e-05,
+      "loss": 0.7058,
+      "step": 2174
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.62236321151466e-05,
+      "loss": 0.7665,
+      "step": 2175
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6220044801493767e-05,
+      "loss": 0.5908,
+      "step": 2176
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6216456181815506e-05,
+      "loss": 0.6022,
+      "step": 2177
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6212866256865322e-05,
+      "loss": 0.754,
+      "step": 2178
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620927502739699e-05,
+      "loss": 0.6923,
+      "step": 2179
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620568249416456e-05,
+      "loss": 0.6046,
+      "step": 2180
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6202088657922365e-05,
+      "loss": 0.7003,
+      "step": 2181
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6198493519425e-05,
+      "loss": 0.753,
+      "step": 2182
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6194897079427335e-05,
+      "loss": 0.6372,
+      "step": 2183
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6191299338684524e-05,
+      "loss": 0.6133,
+      "step": 2184
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6187700297951982e-05,
+      "loss": 0.7339,
+      "step": 2185
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6184099957985397e-05,
+      "loss": 0.7278,
+      "step": 2186
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618049831954074e-05,
+      "loss": 0.8108,
+      "step": 2187
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6176895383374245e-05,
+      "loss": 0.6218,
+      "step": 2188
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617329115024242e-05,
+      "loss": 0.7957,
+      "step": 2189
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6169685620902057e-05,
+      "loss": 0.7652,
+      "step": 2190
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6166078796110198e-05,
+      "loss": 0.7426,
+      "step": 2191
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6162470676624177e-05,
+      "loss": 0.7291,
+      "step": 2192
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6158861263201592e-05,
+      "loss": 0.7575,
+      "step": 2193
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6155250556600313e-05,
+      "loss": 0.7493,
+      "step": 2194
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6151638557578476e-05,
+      "loss": 0.5858,
+      "step": 2195
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6148025266894502e-05,
+      "loss": 0.6848,
+      "step": 2196
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614441068530707e-05,
+      "loss": 0.5986,
+      "step": 2197
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614079481357514e-05,
+      "loss": 0.6849,
+      "step": 2198
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613717765245793e-05,
+      "loss": 0.7198,
+      "step": 2199
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6133559202714946e-05,
+      "loss": 0.7202,
+      "step": 2200
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612993946510595e-05,
+      "loss": 0.7174,
+      "step": 2201
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6126318440390984e-05,
+      "loss": 0.7569,
+      "step": 2202
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6122696129330347e-05,
+      "loss": 0.7718,
+      "step": 2203
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6119072532684627e-05,
+      "loss": 0.7529,
+      "step": 2204
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6115447651214668e-05,
+      "loss": 0.7345,
+      "step": 2205
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6111821485681585e-05,
+      "loss": 0.7666,
+      "step": 2206
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6108194036846773e-05,
+      "loss": 0.7416,
+      "step": 2207
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610456530547188e-05,
+      "loss": 0.7238,
+      "step": 2208
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6100935292318842e-05,
+      "loss": 0.809,
+      "step": 2209
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6097303998149846e-05,
+      "loss": 0.7638,
+      "step": 2210
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6093671423727363e-05,
+      "loss": 0.6126,
+      "step": 2211
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6090037569814122e-05,
+      "loss": 0.7372,
+      "step": 2212
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6086402437173126e-05,
+      "loss": 0.7237,
+      "step": 2213
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.608276602656764e-05,
+      "loss": 0.7556,
+      "step": 2214
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6079128338761215e-05,
+      "loss": 0.7308,
+      "step": 2215
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607548937451765e-05,
+      "loss": 0.6875,
+      "step": 2216
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607184913460101e-05,
+      "loss": 0.7474,
+      "step": 2217
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.606820761977566e-05,
+      "loss": 0.7032,
+      "step": 2218
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6064564830806195e-05,
+      "loss": 0.7444,
+      "step": 2219
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6060920768457498e-05,
+      "loss": 0.6639,
+      "step": 2220
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605727543349471e-05,
+      "loss": 0.6069,
+      "step": 2221
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6053628826683247e-05,
+      "loss": 0.7614,
+      "step": 2222
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604998094878879e-05,
+      "loss": 0.7277,
+      "step": 2223
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604633180057728e-05,
+      "loss": 0.7543,
+      "step": 2224
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6042681382814934e-05,
+      "loss": 0.7284,
+      "step": 2225
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6039029696268234e-05,
+      "loss": 0.7348,
+      "step": 2226
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6035376741703923e-05,
+      "loss": 0.7501,
+      "step": 2227
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6031722519889014e-05,
+      "loss": 0.618,
+      "step": 2228
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6028067031590785e-05,
+      "loss": 0.705,
+      "step": 2229
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602441027757678e-05,
+      "loss": 0.724,
+      "step": 2230
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6020752258614808e-05,
+      "loss": 0.7109,
+      "step": 2231
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601709297547295e-05,
+      "loss": 0.7408,
+      "step": 2232
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6013432428919543e-05,
+      "loss": 0.7538,
+      "step": 2233
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6009770619723194e-05,
+      "loss": 0.7155,
+      "step": 2234
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6006107548652773e-05,
+      "loss": 0.7567,
+      "step": 2235
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6002443216477422e-05,
+      "loss": 0.745,
+      "step": 2236
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5998777623966536e-05,
+      "loss": 0.7324,
+      "step": 2237
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599511077188979e-05,
+      "loss": 0.7386,
+      "step": 2238
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59914426610171e-05,
+      "loss": 0.6213,
+      "step": 2239
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5987773292118678e-05,
+      "loss": 0.5934,
+      "step": 2240
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5984102665964968e-05,
+      "loss": 0.7363,
+      "step": 2241
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5980430783326704e-05,
+      "loss": 0.7491,
+      "step": 2242
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5976757644974866e-05,
+      "loss": 0.7492,
+      "step": 2243
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5973083251680707e-05,
+      "loss": 0.7496,
+      "step": 2244
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596940760421574e-05,
+      "loss": 0.7629,
+      "step": 2245
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596573070335174e-05,
+      "loss": 0.6181,
+      "step": 2246
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596205254986075e-05,
+      "loss": 0.7403,
+      "step": 2247
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5958373144515075e-05,
+      "loss": 0.7811,
+      "step": 2248
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.595469248808728e-05,
+      "loss": 0.6302,
+      "step": 2249
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5951010581350192e-05,
+      "loss": 0.6253,
+      "step": 2250
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5947327425076908e-05,
+      "loss": 0.7415,
+      "step": 2251
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5943643020040774e-05,
+      "loss": 0.78,
+      "step": 2252
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5939957367015414e-05,
+      "loss": 0.7239,
+      "step": 2253
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5936270466774697e-05,
+      "loss": 0.7743,
+      "step": 2254
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5932582320092773e-05,
+      "loss": 0.7388,
+      "step": 2255
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5928892927744034e-05,
+      "loss": 0.7594,
+      "step": 2256
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5925202290503153e-05,
+      "loss": 0.7062,
+      "step": 2257
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5921510409145046e-05,
+      "loss": 0.7391,
+      "step": 2258
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5917817284444907e-05,
+      "loss": 0.7795,
+      "step": 2259
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5914122917178176e-05,
+      "loss": 0.5715,
+      "step": 2260
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5910427308120564e-05,
+      "loss": 0.7086,
+      "step": 2261
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590673045804804e-05,
+      "loss": 0.7044,
+      "step": 2262
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590303236773683e-05,
+      "loss": 0.7358,
+      "step": 2263
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5899333037963426e-05,
+      "loss": 0.7527,
+      "step": 2264
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589563246950458e-05,
+      "loss": 0.7565,
+      "step": 2265
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5891930663137295e-05,
+      "loss": 0.7285,
+      "step": 2266
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.588822761963885e-05,
+      "loss": 0.7249,
+      "step": 2267
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5884523339786772e-05,
+      "loss": 0.7149,
+      "step": 2268
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5880817824358846e-05,
+      "loss": 0.7152,
+      "step": 2269
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5877111074133117e-05,
+      "loss": 0.7479,
+      "step": 2270
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5873403089887903e-05,
+      "loss": 0.7108,
+      "step": 2271
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5869693872401773e-05,
+      "loss": 0.7299,
+      "step": 2272
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586598342245354e-05,
+      "loss": 0.7769,
+      "step": 2273
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5862271740822296e-05,
+      "loss": 0.7761,
+      "step": 2274
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5858558828287386e-05,
+      "loss": 0.7691,
+      "step": 2275
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5854844685628407e-05,
+      "loss": 0.7245,
+      "step": 2276
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5851129313625225e-05,
+      "loss": 0.7302,
+      "step": 2277
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5847412713057952e-05,
+      "loss": 0.7113,
+      "step": 2278
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584369488470697e-05,
+      "loss": 0.7299,
+      "step": 2279
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.583997582935291e-05,
+      "loss": 0.7682,
+      "step": 2280
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.583625554777666e-05,
+      "loss": 0.753,
+      "step": 2281
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.583253404075938e-05,
+      "loss": 0.7494,
+      "step": 2282
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582881130908246e-05,
+      "loss": 0.7581,
+      "step": 2283
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582508735352758e-05,
+      "loss": 0.7701,
+      "step": 2284
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582136217487665e-05,
+      "loss": 0.7481,
+      "step": 2285
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5817635773911854e-05,
+      "loss": 0.6831,
+      "step": 2286
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581390815141562e-05,
+      "loss": 0.7221,
+      "step": 2287
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581017930817064e-05,
+      "loss": 0.748,
+      "step": 2288
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5806449244959864e-05,
+      "loss": 0.7209,
+      "step": 2289
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5802717962566493e-05,
+      "loss": 0.7316,
+      "step": 2290
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5798985461773985e-05,
+      "loss": 0.7154,
+      "step": 2291
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5795251743366056e-05,
+      "loss": 0.7555,
+      "step": 2292
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5791516808126674e-05,
+      "loss": 0.7093,
+      "step": 2293
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.578778065684007e-05,
+      "loss": 0.7239,
+      "step": 2294
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.578404329029072e-05,
+      "loss": 0.5918,
+      "step": 2295
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5780304709263363e-05,
+      "loss": 0.7536,
+      "step": 2296
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.577656491454299e-05,
+      "loss": 0.6147,
+      "step": 2297
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5772823906914848e-05,
+      "loss": 0.7359,
+      "step": 2298
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576908168716444e-05,
+      "loss": 0.6903,
+      "step": 2299
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576533825607751e-05,
+      "loss": 0.7411,
+      "step": 2300
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5761593614440088e-05,
+      "loss": 0.7591,
+      "step": 2301
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5757847763038418e-05,
+      "loss": 0.7216,
+      "step": 2302
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575410070265903e-05,
+      "loss": 0.5874,
+      "step": 2303
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5750352434088695e-05,
+      "loss": 0.7024,
+      "step": 2304
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5746602958114428e-05,
+      "loss": 0.7135,
+      "step": 2305
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5742852275523524e-05,
+      "loss": 0.6925,
+      "step": 2306
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.57391003871035e-05,
+      "loss": 0.7661,
+      "step": 2307
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5735347293642152e-05,
+      "loss": 0.7519,
+      "step": 2308
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573159299592752e-05,
+      "loss": 0.7377,
+      "step": 2309
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5727837494747885e-05,
+      "loss": 0.6173,
+      "step": 2310
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5724080790891803e-05,
+      "loss": 0.6838,
+      "step": 2311
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572032288514806e-05,
+      "loss": 0.7716,
+      "step": 2312
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571656377830571e-05,
+      "loss": 0.7193,
+      "step": 2313
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5712803471154055e-05,
+      "loss": 0.7484,
+      "step": 2314
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.570904196448265e-05,
+      "loss": 0.7353,
+      "step": 2315
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5705279259081296e-05,
+      "loss": 0.7771,
+      "step": 2316
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5701515355740048e-05,
+      "loss": 0.7061,
+      "step": 2317
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5697750255249222e-05,
+      "loss": 0.7155,
+      "step": 2318
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5693983958399375e-05,
+      "loss": 0.6111,
+      "step": 2319
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5690216465981313e-05,
+      "loss": 0.7429,
+      "step": 2320
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56864477787861e-05,
+      "loss": 0.7334,
+      "step": 2321
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5682677897605058e-05,
+      "loss": 0.7424,
+      "step": 2322
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5678906823229735e-05,
+      "loss": 0.5867,
+      "step": 2323
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5675134556451954e-05,
+      "loss": 0.7352,
+      "step": 2324
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567136109806378e-05,
+      "loss": 0.7921,
+      "step": 2325
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5667586448857525e-05,
+      "loss": 0.7363,
+      "step": 2326
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5663810609625758e-05,
+      "loss": 0.6986,
+      "step": 2327
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566003358116129e-05,
+      "loss": 0.7267,
+      "step": 2328
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565625536425718e-05,
+      "loss": 0.7467,
+      "step": 2329
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565247595970675e-05,
+      "loss": 0.6114,
+      "step": 2330
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564869536830356e-05,
+      "loss": 0.6989,
+      "step": 2331
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5644913590841424e-05,
+      "loss": 0.7343,
+      "step": 2332
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5641130628114403e-05,
+      "loss": 0.6211,
+      "step": 2333
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5637346480916806e-05,
+      "loss": 0.7616,
+      "step": 2334
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563356115004319e-05,
+      "loss": 0.7887,
+      "step": 2335
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5629774636288368e-05,
+      "loss": 0.7379,
+      "step": 2336
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5625986940447394e-05,
+      "loss": 0.728,
+      "step": 2337
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562219806331557e-05,
+      "loss": 0.6188,
+      "step": 2338
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5618408005688444e-05,
+      "loss": 0.742,
+      "step": 2339
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5614616768361826e-05,
+      "loss": 0.7453,
+      "step": 2340
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5610824352131758e-05,
+      "loss": 0.7548,
+      "step": 2341
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5607030757794537e-05,
+      "loss": 0.7334,
+      "step": 2342
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5603235986146703e-05,
+      "loss": 0.6113,
+      "step": 2343
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5599440037985048e-05,
+      "loss": 0.7468,
+      "step": 2344
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559564291410661e-05,
+      "loss": 0.7439,
+      "step": 2345
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559184461530867e-05,
+      "loss": 0.6148,
+      "step": 2346
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.558804514238876e-05,
+      "loss": 0.7388,
+      "step": 2347
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5584244496144652e-05,
+      "loss": 0.7384,
+      "step": 2348
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5580442677374378e-05,
+      "loss": 0.7356,
+      "step": 2349
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5576639686876204e-05,
+      "loss": 0.6629,
+      "step": 2350
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5572835525448642e-05,
+      "loss": 0.7156,
+      "step": 2351
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5569030193890458e-05,
+      "loss": 0.7339,
+      "step": 2352
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5565223693000654e-05,
+      "loss": 0.7022,
+      "step": 2353
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5561416023578487e-05,
+      "loss": 0.6842,
+      "step": 2354
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5557607186423455e-05,
+      "loss": 0.7207,
+      "step": 2355
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55537971823353e-05,
+      "loss": 0.7686,
+      "step": 2356
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5549986012114008e-05,
+      "loss": 0.7046,
+      "step": 2357
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5546173676559814e-05,
+      "loss": 0.6932,
+      "step": 2358
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55423601764732e-05,
+      "loss": 0.714,
+      "step": 2359
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5538545512654877e-05,
+      "loss": 0.7386,
+      "step": 2360
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5534729685905822e-05,
+      "loss": 0.7225,
+      "step": 2361
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.553091269702724e-05,
+      "loss": 0.7101,
+      "step": 2362
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552709454682059e-05,
+      "loss": 0.7492,
+      "step": 2363
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552327523608757e-05,
+      "loss": 0.7275,
+      "step": 2364
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5519454765630123e-05,
+      "loss": 0.7368,
+      "step": 2365
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5515633136250425e-05,
+      "loss": 0.6385,
+      "step": 2366
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.551181034875092e-05,
+      "loss": 0.782,
+      "step": 2367
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5507986403934268e-05,
+      "loss": 0.6442,
+      "step": 2368
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.550416130260339e-05,
+      "loss": 0.7087,
+      "step": 2369
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5500335045561445e-05,
+      "loss": 0.7516,
+      "step": 2370
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549650763361183e-05,
+      "loss": 0.6979,
+      "step": 2371
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5492679067558197e-05,
+      "loss": 0.7322,
+      "step": 2372
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5488849348204417e-05,
+      "loss": 0.7942,
+      "step": 2373
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548501847635463e-05,
+      "loss": 0.7279,
+      "step": 2374
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.54811864528132e-05,
+      "loss": 0.7165,
+      "step": 2375
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5477353278384747e-05,
+      "loss": 0.7581,
+      "step": 2376
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5473518953874115e-05,
+      "loss": 0.6802,
+      "step": 2377
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5469683480086398e-05,
+      "loss": 0.7342,
+      "step": 2378
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5465846857826935e-05,
+      "loss": 0.7665,
+      "step": 2379
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5462009087901305e-05,
+      "loss": 0.7372,
+      "step": 2380
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5458170171115328e-05,
+      "loss": 0.7314,
+      "step": 2381
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5454330108275054e-05,
+      "loss": 0.7291,
+      "step": 2382
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5450488900186793e-05,
+      "loss": 0.7198,
+      "step": 2383
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5446646547657078e-05,
+      "loss": 0.6756,
+      "step": 2384
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544280305149269e-05,
+      "loss": 0.7214,
+      "step": 2385
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.543895841250066e-05,
+      "loss": 0.7602,
+      "step": 2386
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5435112631488233e-05,
+      "loss": 0.7294,
+      "step": 2387
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5431265709262916e-05,
+      "loss": 0.7241,
+      "step": 2388
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5427417646632447e-05,
+      "loss": 0.6703,
+      "step": 2389
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5423568444404816e-05,
+      "loss": 0.6733,
+      "step": 2390
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5419718103388225e-05,
+      "loss": 0.5694,
+      "step": 2391
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541586662439114e-05,
+      "loss": 0.7614,
+      "step": 2392
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5412014008222263e-05,
+      "loss": 0.7567,
+      "step": 2393
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540816025569052e-05,
+      "loss": 0.6342,
+      "step": 2394
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5404305367605087e-05,
+      "loss": 0.7518,
+      "step": 2395
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540044934477538e-05,
+      "loss": 0.6235,
+      "step": 2396
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5396592188011047e-05,
+      "loss": 0.5867,
+      "step": 2397
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.539273389812198e-05,
+      "loss": 0.7501,
+      "step": 2398
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5388874475918302e-05,
+      "loss": 0.7349,
+      "step": 2399
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538501392221038e-05,
+      "loss": 0.6155,
+      "step": 2400
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5381152237808816e-05,
+      "loss": 0.7438,
+      "step": 2401
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5377289423524445e-05,
+      "loss": 0.6955,
+      "step": 2402
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.537342548016835e-05,
+      "loss": 0.7018,
+      "step": 2403
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5369560408551844e-05,
+      "loss": 0.7458,
+      "step": 2404
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5365694209486475e-05,
+      "loss": 0.6842,
+      "step": 2405
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.536182688378403e-05,
+      "loss": 0.7093,
+      "step": 2406
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5357958432256535e-05,
+      "loss": 0.7667,
+      "step": 2407
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.535408885571625e-05,
+      "loss": 0.6219,
+      "step": 2408
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5350218154975672e-05,
+      "loss": 0.7754,
+      "step": 2409
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5346346330847534e-05,
+      "loss": 0.7466,
+      "step": 2410
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5342473384144804e-05,
+      "loss": 0.7558,
+      "step": 2411
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5338599315680682e-05,
+      "loss": 0.7463,
+      "step": 2412
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533472412626861e-05,
+      "loss": 0.6334,
+      "step": 2413
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533084781672227e-05,
+      "loss": 0.7277,
+      "step": 2414
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532697038785557e-05,
+      "loss": 0.7338,
+      "step": 2415
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5323091840482646e-05,
+      "loss": 0.729,
+      "step": 2416
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5319212175417884e-05,
+      "loss": 0.7131,
+      "step": 2417
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5315331393475905e-05,
+      "loss": 0.6024,
+      "step": 2418
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531144949547155e-05,
+      "loss": 0.7519,
+      "step": 2419
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530756648221991e-05,
+      "loss": 0.748,
+      "step": 2420
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5303682354536294e-05,
+      "loss": 0.6207,
+      "step": 2421
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5299797113236264e-05,
+      "loss": 0.7523,
+      "step": 2422
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5295910759135594e-05,
+      "loss": 0.7336,
+      "step": 2423
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529202329305031e-05,
+      "loss": 0.728,
+      "step": 2424
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528813471579667e-05,
+      "loss": 0.7602,
+      "step": 2425
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5284245028191152e-05,
+      "loss": 0.7104,
+      "step": 2426
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528035423105048e-05,
+      "loss": 0.7364,
+      "step": 2427
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5276462325191603e-05,
+      "loss": 0.7355,
+      "step": 2428
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527256931143171e-05,
+      "loss": 0.7246,
+      "step": 2429
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5268675190588217e-05,
+      "loss": 0.726,
+      "step": 2430
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5264779963478767e-05,
+      "loss": 0.7295,
+      "step": 2431
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5260883630921254e-05,
+      "loss": 0.6983,
+      "step": 2432
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5256986193733792e-05,
+      "loss": 0.6467,
+      "step": 2433
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5253087652734717e-05,
+      "loss": 0.7385,
+      "step": 2434
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5249188008742615e-05,
+      "loss": 0.7371,
+      "step": 2435
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5245287262576298e-05,
+      "loss": 0.7081,
+      "step": 2436
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.52413854150548e-05,
+      "loss": 0.714,
+      "step": 2437
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5237482466997397e-05,
+      "loss": 0.6248,
+      "step": 2438
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5233578419223593e-05,
+      "loss": 0.7404,
+      "step": 2439
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5229673272553127e-05,
+      "loss": 0.7391,
+      "step": 2440
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5225767027805956e-05,
+      "loss": 0.73,
+      "step": 2441
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5221859685802278e-05,
+      "loss": 0.7426,
+      "step": 2442
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5217951247362526e-05,
+      "loss": 0.7123,
+      "step": 2443
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.521404171330735e-05,
+      "loss": 0.6129,
+      "step": 2444
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5210131084457635e-05,
+      "loss": 0.7198,
+      "step": 2445
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5206219361634502e-05,
+      "loss": 0.7697,
+      "step": 2446
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5202306545659297e-05,
+      "loss": 0.7246,
+      "step": 2447
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5198392637353592e-05,
+      "loss": 0.7393,
+      "step": 2448
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5194477637539194e-05,
+      "loss": 0.7161,
+      "step": 2449
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5190561547038138e-05,
+      "loss": 0.7514,
+      "step": 2450
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5186644366672689e-05,
+      "loss": 0.7484,
+      "step": 2451
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5182726097265334e-05,
+      "loss": 0.7549,
+      "step": 2452
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5178806739638794e-05,
+      "loss": 0.7258,
+      "step": 2453
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5174886294616022e-05,
+      "loss": 0.6072,
+      "step": 2454
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5170964763020195e-05,
+      "loss": 0.7056,
+      "step": 2455
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5167042145674717e-05,
+      "loss": 0.7292,
+      "step": 2456
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5163118443403218e-05,
+      "loss": 0.7607,
+      "step": 2457
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5159193657029567e-05,
+      "loss": 0.7138,
+      "step": 2458
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.515526778737785e-05,
+      "loss": 0.7368,
+      "step": 2459
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.515134083527238e-05,
+      "loss": 0.6823,
+      "step": 2460
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.514741280153771e-05,
+      "loss": 0.7355,
+      "step": 2461
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.51434836869986e-05,
+      "loss": 0.7328,
+      "step": 2462
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5139553492480051e-05,
+      "loss": 0.7654,
+      "step": 2463
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5135622218807289e-05,
+      "loss": 0.7355,
+      "step": 2464
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.513168986680577e-05,
+      "loss": 0.7425,
+      "step": 2465
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5127756437301165e-05,
+      "loss": 0.7258,
+      "step": 2466
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5123821931119375e-05,
+      "loss": 0.7631,
+      "step": 2467
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.511988634908654e-05,
+      "loss": 0.746,
+      "step": 2468
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.511594969202901e-05,
+      "loss": 0.7335,
+      "step": 2469
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5112011960773365e-05,
+      "loss": 0.7016,
+      "step": 2470
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5108073156146414e-05,
+      "loss": 0.7502,
+      "step": 2471
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5104133278975192e-05,
+      "loss": 0.7555,
+      "step": 2472
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5100192330086954e-05,
+      "loss": 0.7452,
+      "step": 2473
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.509625031030918e-05,
+      "loss": 0.7763,
+      "step": 2474
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5092307220469581e-05,
+      "loss": 0.6771,
+      "step": 2475
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5088363061396094e-05,
+      "loss": 0.7458,
+      "step": 2476
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5084417833916866e-05,
+      "loss": 0.7313,
+      "step": 2477
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.508047153886028e-05,
+      "loss": 0.747,
+      "step": 2478
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5076524177054949e-05,
+      "loss": 0.7382,
+      "step": 2479
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5072575749329694e-05,
+      "loss": 0.6785,
+      "step": 2480
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5068626256513575e-05,
+      "loss": 0.7149,
+      "step": 2481
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5064675699435859e-05,
+      "loss": 0.7114,
+      "step": 2482
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5060724078926056e-05,
+      "loss": 0.729,
+      "step": 2483
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5056771395813884e-05,
+      "loss": 0.7205,
+      "step": 2484
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5052817650929288e-05,
+      "loss": 0.7845,
+      "step": 2485
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5048862845102443e-05,
+      "loss": 0.735,
+      "step": 2486
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5044906979163738e-05,
+      "loss": 0.5822,
+      "step": 2487
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.504095005394379e-05,
+      "loss": 0.6108,
+      "step": 2488
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5036992070273431e-05,
+      "loss": 0.6447,
+      "step": 2489
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.503303302898373e-05,
+      "loss": 0.7127,
+      "step": 2490
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.502907293090596e-05,
+      "loss": 0.7159,
+      "step": 2491
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5025111776871622e-05,
+      "loss": 0.7035,
+      "step": 2492
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5021149567712452e-05,
+      "loss": 0.7338,
+      "step": 2493
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5017186304260392e-05,
+      "loss": 0.7569,
+      "step": 2494
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5013221987347605e-05,
+      "loss": 0.6968,
+      "step": 2495
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5009256617806487e-05,
+      "loss": 0.7085,
+      "step": 2496
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5005290196469644e-05,
+      "loss": 0.5885,
+      "step": 2497
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.500132272416991e-05,
+      "loss": 0.7277,
+      "step": 2498
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4997354201740338e-05,
+      "loss": 0.6746,
+      "step": 2499
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4993384630014194e-05,
+      "loss": 0.6966,
+      "step": 2500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4989414009824978e-05,
+      "loss": 0.6575,
+      "step": 2501
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4985442342006397e-05,
+      "loss": 0.7332,
+      "step": 2502
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4981469627392387e-05,
+      "loss": 0.7042,
+      "step": 2503
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4977495866817102e-05,
+      "loss": 0.7614,
+      "step": 2504
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.497352106111491e-05,
+      "loss": 0.7682,
+      "step": 2505
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4969545211120407e-05,
+      "loss": 0.7092,
+      "step": 2506
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4965568317668396e-05,
+      "loss": 0.7302,
+      "step": 2507
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4961590381593918e-05,
+      "loss": 0.6939,
+      "step": 2508
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4957611403732216e-05,
+      "loss": 0.7805,
+      "step": 2509
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4953631384918754e-05,
+      "loss": 0.5706,
+      "step": 2510
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4949650325989227e-05,
+      "loss": 0.7371,
+      "step": 2511
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4945668227779534e-05,
+      "loss": 0.7363,
+      "step": 2512
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4941685091125797e-05,
+      "loss": 0.7577,
+      "step": 2513
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4937700916864359e-05,
+      "loss": 0.7261,
+      "step": 2514
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4933715705831779e-05,
+      "loss": 0.7499,
+      "step": 2515
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4929729458864834e-05,
+      "loss": 0.6236,
+      "step": 2516
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4925742176800518e-05,
+      "loss": 0.6651,
+      "step": 2517
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4921753860476038e-05,
+      "loss": 0.7062,
+      "step": 2518
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4917764510728827e-05,
+      "loss": 0.6918,
+      "step": 2519
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4913774128396527e-05,
+      "loss": 0.6047,
+      "step": 2520
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4909782714317003e-05,
+      "loss": 0.6425,
+      "step": 2521
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4905790269328337e-05,
+      "loss": 0.585,
+      "step": 2522
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4901796794268816e-05,
+      "loss": 0.6973,
+      "step": 2523
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.489780228997696e-05,
+      "loss": 0.7422,
+      "step": 2524
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.489380675729149e-05,
+      "loss": 0.7584,
+      "step": 2525
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4889810197051353e-05,
+      "loss": 0.7972,
+      "step": 2526
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.488581261009571e-05,
+      "loss": 0.7313,
+      "step": 2527
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.488181399726393e-05,
+      "loss": 0.7622,
+      "step": 2528
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4877814359395612e-05,
+      "loss": 0.7329,
+      "step": 2529
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4873813697330557e-05,
+      "loss": 0.7764,
+      "step": 2530
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4869812011908784e-05,
+      "loss": 0.6072,
+      "step": 2531
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4865809303970528e-05,
+      "loss": 0.7069,
+      "step": 2532
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4861805574356247e-05,
+      "loss": 0.6969,
+      "step": 2533
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4857800823906598e-05,
+      "loss": 0.6247,
+      "step": 2534
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4853795053462461e-05,
+      "loss": 0.5994,
+      "step": 2535
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4849788263864934e-05,
+      "loss": 0.7643,
+      "step": 2536
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4845780455955321e-05,
+      "loss": 0.7528,
+      "step": 2537
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4841771630575141e-05,
+      "loss": 0.721,
+      "step": 2538
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4837761788566133e-05,
+      "loss": 0.7214,
+      "step": 2539
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4833750930770243e-05,
+      "loss": 0.7437,
+      "step": 2540
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4829739058029632e-05,
+      "loss": 0.6914,
+      "step": 2541
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4825726171186675e-05,
+      "loss": 0.755,
+      "step": 2542
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4821712271083964e-05,
+      "loss": 0.7692,
+      "step": 2543
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4817697358564292e-05,
+      "loss": 0.7241,
+      "step": 2544
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4813681434470675e-05,
+      "loss": 0.7459,
+      "step": 2545
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4809664499646336e-05,
+      "loss": 0.7479,
+      "step": 2546
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.480564655493472e-05,
+      "loss": 0.6915,
+      "step": 2547
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.480162760117947e-05,
+      "loss": 0.7575,
+      "step": 2548
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.479760763922445e-05,
+      "loss": 0.637,
+      "step": 2549
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4793586669913727e-05,
+      "loss": 0.5941,
+      "step": 2550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4789564694091595e-05,
+      "loss": 0.6089,
+      "step": 2551
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4785541712602545e-05,
+      "loss": 0.7086,
+      "step": 2552
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4781517726291283e-05,
+      "loss": 0.7359,
+      "step": 2553
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4777492736002732e-05,
+      "loss": 0.6225,
+      "step": 2554
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4773466742582017e-05,
+      "loss": 0.7221,
+      "step": 2555
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4769439746874477e-05,
+      "loss": 0.7112,
+      "step": 2556
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.476541174972566e-05,
+      "loss": 0.7599,
+      "step": 2557
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4761382751981334e-05,
+      "loss": 0.7011,
+      "step": 2558
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4757352754487465e-05,
+      "loss": 0.7093,
+      "step": 2559
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4753321758090231e-05,
+      "loss": 0.7798,
+      "step": 2560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4749289763636026e-05,
+      "loss": 0.7766,
+      "step": 2561
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4745256771971446e-05,
+      "loss": 0.7248,
+      "step": 2562
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4741222783943303e-05,
+      "loss": 0.668,
+      "step": 2563
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4737187800398611e-05,
+      "loss": 0.7416,
+      "step": 2564
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4733151822184603e-05,
+      "loss": 0.7606,
+      "step": 2565
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4729114850148711e-05,
+      "loss": 0.7038,
+      "step": 2566
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4725076885138581e-05,
+      "loss": 0.7178,
+      "step": 2567
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4721037928002066e-05,
+      "loss": 0.7061,
+      "step": 2568
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4716997979587229e-05,
+      "loss": 0.7318,
+      "step": 2569
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4712957040742334e-05,
+      "loss": 0.6185,
+      "step": 2570
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4708915112315864e-05,
+      "loss": 0.6203,
+      "step": 2571
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4704872195156508e-05,
+      "loss": 0.7581,
+      "step": 2572
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4700828290113152e-05,
+      "loss": 0.7545,
+      "step": 2573
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4696783398034897e-05,
+      "loss": 0.6698,
+      "step": 2574
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4692737519771052e-05,
+      "loss": 0.7313,
+      "step": 2575
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4688690656171133e-05,
+      "loss": 0.7002,
+      "step": 2576
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4684642808084863e-05,
+      "loss": 0.7282,
+      "step": 2577
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4680593976362168e-05,
+      "loss": 0.7128,
+      "step": 2578
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.467654416185318e-05,
+      "loss": 0.7237,
+      "step": 2579
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4672493365408245e-05,
+      "loss": 0.7352,
+      "step": 2580
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4668441587877908e-05,
+      "loss": 0.78,
+      "step": 2581
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4664388830112924e-05,
+      "loss": 0.782,
+      "step": 2582
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4660335092964253e-05,
+      "loss": 0.7344,
+      "step": 2583
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4656280377283056e-05,
+      "loss": 0.7342,
+      "step": 2584
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4652224683920708e-05,
+      "loss": 0.5887,
+      "step": 2585
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.464816801372878e-05,
+      "loss": 0.5985,
+      "step": 2586
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4644110367559056e-05,
+      "loss": 0.7063,
+      "step": 2587
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4640051746263521e-05,
+      "loss": 0.7289,
+      "step": 2588
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4635992150694364e-05,
+      "loss": 0.7527,
+      "step": 2589
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4631931581703984e-05,
+      "loss": 0.6821,
+      "step": 2590
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4627870040144978e-05,
+      "loss": 0.7238,
+      "step": 2591
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4623807526870146e-05,
+      "loss": 0.7127,
+      "step": 2592
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4619744042732503e-05,
+      "loss": 0.7512,
+      "step": 2593
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4615679588585252e-05,
+      "loss": 0.5851,
+      "step": 2594
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4611614165281813e-05,
+      "loss": 0.7663,
+      "step": 2595
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4607547773675805e-05,
+      "loss": 0.689,
+      "step": 2596
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4603480414621049e-05,
+      "loss": 0.7478,
+      "step": 2597
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4599412088971572e-05,
+      "loss": 0.7554,
+      "step": 2598
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4595342797581597e-05,
+      "loss": 0.7405,
+      "step": 2599
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4591272541305558e-05,
+      "loss": 0.603,
+      "step": 2600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4587201320998095e-05,
+      "loss": 0.7404,
+      "step": 2601
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4583129137514032e-05,
+      "loss": 0.6699,
+      "step": 2602
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4579055991708413e-05,
+      "loss": 0.7799,
+      "step": 2603
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4574981884436477e-05,
+      "loss": 0.7628,
+      "step": 2604
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4570906816553667e-05,
+      "loss": 0.6136,
+      "step": 2605
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4566830788915628e-05,
+      "loss": 0.7441,
+      "step": 2606
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4562753802378197e-05,
+      "loss": 0.7166,
+      "step": 2607
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4558675857797435e-05,
+      "loss": 0.7348,
+      "step": 2608
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.455459695602958e-05,
+      "loss": 0.7285,
+      "step": 2609
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.455051709793108e-05,
+      "loss": 0.7258,
+      "step": 2610
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.454643628435859e-05,
+      "loss": 0.7233,
+      "step": 2611
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4542354516168955e-05,
+      "loss": 0.6876,
+      "step": 2612
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4538271794219227e-05,
+      "loss": 0.7517,
+      "step": 2613
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4534188119366659e-05,
+      "loss": 0.7514,
+      "step": 2614
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4530103492468701e-05,
+      "loss": 0.7274,
+      "step": 2615
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4526017914383004e-05,
+      "loss": 0.7274,
+      "step": 2616
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4521931385967419e-05,
+      "loss": 0.7211,
+      "step": 2617
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4517843908079992e-05,
+      "loss": 0.6311,
+      "step": 2618
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4513755481578978e-05,
+      "loss": 0.7574,
+      "step": 2619
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4509666107322821e-05,
+      "loss": 0.5845,
+      "step": 2620
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4505575786170171e-05,
+      "loss": 0.694,
+      "step": 2621
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4501484518979876e-05,
+      "loss": 0.6333,
+      "step": 2622
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.449739230661098e-05,
+      "loss": 0.7745,
+      "step": 2623
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4493299149922724e-05,
+      "loss": 0.7241,
+      "step": 2624
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.448920504977455e-05,
+      "loss": 0.7196,
+      "step": 2625
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4485110007026105e-05,
+      "loss": 0.7124,
+      "step": 2626
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4481014022537218e-05,
+      "loss": 0.7179,
+      "step": 2627
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4476917097167928e-05,
+      "loss": 0.7219,
+      "step": 2628
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.447281923177847e-05,
+      "loss": 0.7674,
+      "step": 2629
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.446872042722927e-05,
+      "loss": 0.6664,
+      "step": 2630
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4464620684380961e-05,
+      "loss": 0.7496,
+      "step": 2631
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4460520004094362e-05,
+      "loss": 0.6285,
+      "step": 2632
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4456418387230504e-05,
+      "loss": 0.7623,
+      "step": 2633
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4452315834650594e-05,
+      "loss": 0.7236,
+      "step": 2634
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4448212347216054e-05,
+      "loss": 0.733,
+      "step": 2635
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.444410792578849e-05,
+      "loss": 0.7229,
+      "step": 2636
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4440002571229714e-05,
+      "loss": 0.7239,
+      "step": 2637
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4435896284401726e-05,
+      "loss": 0.7477,
+      "step": 2638
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4431789066166729e-05,
+      "loss": 0.7164,
+      "step": 2639
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.442768091738711e-05,
+      "loss": 0.7004,
+      "step": 2640
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4423571838925462e-05,
+      "loss": 0.7592,
+      "step": 2641
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4419461831644574e-05,
+      "loss": 0.7025,
+      "step": 2642
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.441535089640742e-05,
+      "loss": 0.7382,
+      "step": 2643
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4411239034077178e-05,
+      "loss": 0.5889,
+      "step": 2644
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4407126245517219e-05,
+      "loss": 0.7644,
+      "step": 2645
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4403012531591103e-05,
+      "loss": 0.724,
+      "step": 2646
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.439889789316259e-05,
+      "loss": 0.59,
+      "step": 2647
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4394782331095632e-05,
+      "loss": 0.5977,
+      "step": 2648
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4390665846254377e-05,
+      "loss": 0.737,
+      "step": 2649
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.438654843950316e-05,
+      "loss": 0.6978,
+      "step": 2650
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4382430111706524e-05,
+      "loss": 0.7098,
+      "step": 2651
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4378310863729188e-05,
+      "loss": 0.748,
+      "step": 2652
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4374190696436074e-05,
+      "loss": 0.6198,
+      "step": 2653
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4370069610692298e-05,
+      "loss": 0.7528,
+      "step": 2654
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4365947607363162e-05,
+      "loss": 0.7333,
+      "step": 2655
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4361824687314168e-05,
+      "loss": 0.7003,
+      "step": 2656
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4357700851411005e-05,
+      "loss": 0.7809,
+      "step": 2657
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.435357610051956e-05,
+      "loss": 0.7591,
+      "step": 2658
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4349450435505903e-05,
+      "loss": 0.745,
+      "step": 2659
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4345323857236307e-05,
+      "loss": 0.7542,
+      "step": 2660
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4341196366577228e-05,
+      "loss": 0.7161,
+      "step": 2661
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4337067964395318e-05,
+      "loss": 0.6119,
+      "step": 2662
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4332938651557422e-05,
+      "loss": 0.7447,
+      "step": 2663
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.432880842893057e-05,
+      "loss": 0.7498,
+      "step": 2664
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4324677297381988e-05,
+      "loss": 0.7013,
+      "step": 2665
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4320545257779091e-05,
+      "loss": 0.7574,
+      "step": 2666
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4316412310989485e-05,
+      "loss": 0.5895,
+      "step": 2667
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4312278457880966e-05,
+      "loss": 0.7261,
+      "step": 2668
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4308143699321523e-05,
+      "loss": 0.7288,
+      "step": 2669
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4304008036179333e-05,
+      "loss": 0.7064,
+      "step": 2670
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.429987146932276e-05,
+      "loss": 0.7601,
+      "step": 2671
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4295733999620362e-05,
+      "loss": 0.7418,
+      "step": 2672
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4291595627940886e-05,
+      "loss": 0.6849,
+      "step": 2673
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4287456355153269e-05,
+      "loss": 0.7336,
+      "step": 2674
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.428331618212663e-05,
+      "loss": 0.7118,
+      "step": 2675
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4279175109730293e-05,
+      "loss": 0.7453,
+      "step": 2676
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.427503313883375e-05,
+      "loss": 0.7591,
+      "step": 2677
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.42708902703067e-05,
+      "loss": 0.713,
+      "step": 2678
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4266746505019016e-05,
+      "loss": 0.612,
+      "step": 2679
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4262601843840774e-05,
+      "loss": 0.6816,
+      "step": 2680
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4258456287642223e-05,
+      "loss": 0.7239,
+      "step": 2681
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4254309837293811e-05,
+      "loss": 0.734,
+      "step": 2682
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4250162493666173e-05,
+      "loss": 0.7229,
+      "step": 2683
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4246014257630123e-05,
+      "loss": 0.6732,
+      "step": 2684
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4241865130056669e-05,
+      "loss": 0.6887,
+      "step": 2685
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4237715111817006e-05,
+      "loss": 0.7467,
+      "step": 2686
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4233564203782516e-05,
+      "loss": 0.6336,
+      "step": 2687
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4229412406824765e-05,
+      "loss": 0.7577,
+      "step": 2688
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.422525972181551e-05,
+      "loss": 0.7089,
+      "step": 2689
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.422110614962669e-05,
+      "loss": 0.7123,
+      "step": 2690
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.421695169113043e-05,
+      "loss": 0.5758,
+      "step": 2691
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4212796347199049e-05,
+      "loss": 0.6831,
+      "step": 2692
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4208640118705041e-05,
+      "loss": 0.7458,
+      "step": 2693
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4204483006521094e-05,
+      "loss": 0.714,
+      "step": 2694
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.420032501152008e-05,
+      "loss": 0.7196,
+      "step": 2695
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4196166134575053e-05,
+      "loss": 0.7272,
+      "step": 2696
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4192006376559253e-05,
+      "loss": 0.6403,
+      "step": 2697
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4187845738346109e-05,
+      "loss": 0.7428,
+      "step": 2698
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4183684220809228e-05,
+      "loss": 0.6983,
+      "step": 2699
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4179521824822411e-05,
+      "loss": 0.7284,
+      "step": 2700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4175358551259633e-05,
+      "loss": 0.6956,
+      "step": 2701
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4171194400995058e-05,
+      "loss": 0.7214,
+      "step": 2702
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4167029374903041e-05,
+      "loss": 0.6945,
+      "step": 2703
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4162863473858107e-05,
+      "loss": 0.5747,
+      "step": 2704
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4158696698734976e-05,
+      "loss": 0.7382,
+      "step": 2705
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4154529050408546e-05,
+      "loss": 0.7295,
+      "step": 2706
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4150360529753902e-05,
+      "loss": 0.765,
+      "step": 2707
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4146191137646305e-05,
+      "loss": 0.6206,
+      "step": 2708
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4142020874961212e-05,
+      "loss": 0.7313,
+      "step": 2709
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4137849742574245e-05,
+      "loss": 0.759,
+      "step": 2710
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4133677741361227e-05,
+      "loss": 0.6902,
+      "step": 2711
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.412950487219815e-05,
+      "loss": 0.7641,
+      "step": 2712
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4125331135961195e-05,
+      "loss": 0.7224,
+      "step": 2713
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4121156533526723e-05,
+      "loss": 0.7198,
+      "step": 2714
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4116981065771275e-05,
+      "loss": 0.6473,
+      "step": 2715
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4112804733571582e-05,
+      "loss": 0.746,
+      "step": 2716
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4108627537804542e-05,
+      "loss": 0.6853,
+      "step": 2717
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4104449479347247e-05,
+      "loss": 0.6655,
+      "step": 2718
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4100270559076966e-05,
+      "loss": 0.5945,
+      "step": 2719
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4096090777871145e-05,
+      "loss": 0.7433,
+      "step": 2720
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.409191013660742e-05,
+      "loss": 0.7126,
+      "step": 2721
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4087728636163598e-05,
+      "loss": 0.6932,
+      "step": 2722
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.408354627741767e-05,
+      "loss": 0.7356,
+      "step": 2723
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.407936306124781e-05,
+      "loss": 0.7146,
+      "step": 2724
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.407517898853237e-05,
+      "loss": 0.7092,
+      "step": 2725
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4070994060149881e-05,
+      "loss": 0.7308,
+      "step": 2726
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4066808276979051e-05,
+      "loss": 0.5773,
+      "step": 2727
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4062621639898772e-05,
+      "loss": 0.6694,
+      "step": 2728
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4058434149788118e-05,
+      "loss": 0.7656,
+      "step": 2729
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4054245807526334e-05,
+      "loss": 0.7205,
+      "step": 2730
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4050056613992851e-05,
+      "loss": 0.6842,
+      "step": 2731
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.404586657006727e-05,
+      "loss": 0.5922,
+      "step": 2732
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4041675676629383e-05,
+      "loss": 0.8084,
+      "step": 2733
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4037483934559151e-05,
+      "loss": 0.7096,
+      "step": 2734
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4033291344736714e-05,
+      "loss": 0.6958,
+      "step": 2735
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4029097908042392e-05,
+      "loss": 0.653,
+      "step": 2736
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.402490362535669e-05,
+      "loss": 0.7698,
+      "step": 2737
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4020708497560272e-05,
+      "loss": 0.7225,
+      "step": 2738
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4016512525534e-05,
+      "loss": 0.7229,
+      "step": 2739
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4012315710158897e-05,
+      "loss": 0.7462,
+      "step": 2740
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4008118052316173e-05,
+      "loss": 0.734,
+      "step": 2741
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4003919552887216e-05,
+      "loss": 0.7352,
+      "step": 2742
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.399972021275358e-05,
+      "loss": 0.725,
+      "step": 2743
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3995520032797005e-05,
+      "loss": 0.7113,
+      "step": 2744
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3991319013899407e-05,
+      "loss": 0.7304,
+      "step": 2745
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.398711715694287e-05,
+      "loss": 0.5436,
+      "step": 2746
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3982914462809668e-05,
+      "loss": 0.7381,
+      "step": 2747
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3978710932382233e-05,
+      "loss": 0.551,
+      "step": 2748
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3974506566543189e-05,
+      "loss": 0.7684,
+      "step": 2749
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3970301366175325e-05,
+      "loss": 0.7753,
+      "step": 2750
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3966095332161613e-05,
+      "loss": 0.723,
+      "step": 2751
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.396188846538519e-05,
+      "loss": 0.7562,
+      "step": 2752
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3957680766729376e-05,
+      "loss": 0.7278,
+      "step": 2753
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3953472237077664e-05,
+      "loss": 0.7045,
+      "step": 2754
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3949262877313721e-05,
+      "loss": 0.7408,
+      "step": 2755
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3945052688321385e-05,
+      "loss": 0.7581,
+      "step": 2756
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3940841670984675e-05,
+      "loss": 0.588,
+      "step": 2757
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3936629826187777e-05,
+      "loss": 0.7356,
+      "step": 2758
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3932417154815057e-05,
+      "loss": 0.7433,
+      "step": 2759
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.392820365775105e-05,
+      "loss": 0.6564,
+      "step": 2760
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3923989335880464e-05,
+      "loss": 0.7538,
+      "step": 2761
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3919774190088182e-05,
+      "loss": 0.7514,
+      "step": 2762
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3915558221259264e-05,
+      "loss": 0.7314,
+      "step": 2763
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3911341430278935e-05,
+      "loss": 0.7088,
+      "step": 2764
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3907123818032598e-05,
+      "loss": 0.753,
+      "step": 2765
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3902905385405824e-05,
+      "loss": 0.6255,
+      "step": 2766
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3898686133284364e-05,
+      "loss": 0.7017,
+      "step": 2767
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3894466062554133e-05,
+      "loss": 0.7368,
+      "step": 2768
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.389024517410122e-05,
+      "loss": 0.7246,
+      "step": 2769
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.388602346881189e-05,
+      "loss": 0.7071,
+      "step": 2770
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3881800947572573e-05,
+      "loss": 0.7412,
+      "step": 2771
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3877577611269874e-05,
+      "loss": 0.7288,
+      "step": 2772
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3873353460790573e-05,
+      "loss": 0.7024,
+      "step": 2773
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3869128497021614e-05,
+      "loss": 0.6366,
+      "step": 2774
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3864902720850109e-05,
+      "loss": 0.7583,
+      "step": 2775
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3860676133163353e-05,
+      "loss": 0.7265,
+      "step": 2776
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3856448734848805e-05,
+      "loss": 0.7315,
+      "step": 2777
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3852220526794087e-05,
+      "loss": 0.6626,
+      "step": 2778
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3847991509887004e-05,
+      "loss": 0.747,
+      "step": 2779
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.384376168501552e-05,
+      "loss": 0.7238,
+      "step": 2780
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3839531053067776e-05,
+      "loss": 0.5759,
+      "step": 2781
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3835299614932078e-05,
+      "loss": 0.618,
+      "step": 2782
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3831067371496904e-05,
+      "loss": 0.718,
+      "step": 2783
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3826834323650899e-05,
+      "loss": 0.7064,
+      "step": 2784
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3822600472282878e-05,
+      "loss": 0.5646,
+      "step": 2785
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3818365818281824e-05,
+      "loss": 0.7366,
+      "step": 2786
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3814130362536892e-05,
+      "loss": 0.7282,
+      "step": 2787
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3809894105937401e-05,
+      "loss": 0.7552,
+      "step": 2788
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3805657049372835e-05,
+      "loss": 0.7054,
+      "step": 2789
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3801419193732859e-05,
+      "loss": 0.753,
+      "step": 2790
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3797180539907289e-05,
+      "loss": 0.7393,
+      "step": 2791
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3792941088786122e-05,
+      "loss": 0.7121,
+      "step": 2792
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3788700841259517e-05,
+      "loss": 0.7581,
+      "step": 2793
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.37844597982178e-05,
+      "loss": 0.5777,
+      "step": 2794
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3780217960551461e-05,
+      "loss": 0.7573,
+      "step": 2795
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3775975329151165e-05,
+      "loss": 0.7495,
+      "step": 2796
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.377173190490774e-05,
+      "loss": 0.7027,
+      "step": 2797
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3767487688712175e-05,
+      "loss": 0.7485,
+      "step": 2798
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3763242681455633e-05,
+      "loss": 0.7167,
+      "step": 2799
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3758996884029438e-05,
+      "loss": 0.7153,
+      "step": 2800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3754750297325084e-05,
+      "loss": 0.7066,
+      "step": 2801
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3750502922234228e-05,
+      "loss": 0.7292,
+      "step": 2802
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3746254759648692e-05,
+      "loss": 0.5869,
+      "step": 2803
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3742005810460465e-05,
+      "loss": 0.5593,
+      "step": 2804
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.37377560755617e-05,
+      "loss": 0.6935,
+      "step": 2805
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.373350555584472e-05,
+      "loss": 0.7135,
+      "step": 2806
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3729254252202002e-05,
+      "loss": 0.7508,
+      "step": 2807
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3725002165526198e-05,
+      "loss": 0.7396,
+      "step": 2808
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3720749296710118e-05,
+      "loss": 0.7491,
+      "step": 2809
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.371649564664674e-05,
+      "loss": 0.6472,
+      "step": 2810
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3712241216229208e-05,
+      "loss": 0.7256,
+      "step": 2811
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.370798600635082e-05,
+      "loss": 0.7562,
+      "step": 2812
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3703730017905051e-05,
+      "loss": 0.6252,
+      "step": 2813
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3699473251785525e-05,
+      "loss": 0.7335,
+      "step": 2814
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3695215708886043e-05,
+      "loss": 0.7338,
+      "step": 2815
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.369095739010056e-05,
+      "loss": 0.6827,
+      "step": 2816
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.36866982963232e-05,
+      "loss": 0.7316,
+      "step": 2817
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3682438428448245e-05,
+      "loss": 0.6976,
+      "step": 2818
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3678177787370141e-05,
+      "loss": 0.6085,
+      "step": 2819
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3673916373983496e-05,
+      "loss": 0.6163,
+      "step": 2820
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.366965418918308e-05,
+      "loss": 0.6264,
+      "step": 2821
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3665391233863832e-05,
+      "loss": 0.7089,
+      "step": 2822
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3661127508920838e-05,
+      "loss": 0.7209,
+      "step": 2823
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.365686301524936e-05,
+      "loss": 0.6855,
+      "step": 2824
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3652597753744811e-05,
+      "loss": 0.6427,
+      "step": 2825
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3648331725302775e-05,
+      "loss": 0.7493,
+      "step": 2826
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3644064930818987e-05,
+      "loss": 0.6263,
+      "step": 2827
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.363979737118935e-05,
+      "loss": 0.7223,
+      "step": 2828
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3635529047309925e-05,
+      "loss": 0.5956,
+      "step": 2829
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3631259960076935e-05,
+      "loss": 0.6029,
+      "step": 2830
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.362699011038676e-05,
+      "loss": 0.7406,
+      "step": 2831
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3622719499135941e-05,
+      "loss": 0.7452,
+      "step": 2832
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3618448127221184e-05,
+      "loss": 0.7113,
+      "step": 2833
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3614175995539349e-05,
+      "loss": 0.7183,
+      "step": 2834
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3609903104987457e-05,
+      "loss": 0.7573,
+      "step": 2835
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3605629456462687e-05,
+      "loss": 0.698,
+      "step": 2836
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3601355050862382e-05,
+      "loss": 0.5639,
+      "step": 2837
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3597079889084037e-05,
+      "loss": 0.7131,
+      "step": 2838
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3592803972025315e-05,
+      "loss": 0.686,
+      "step": 2839
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3588527300584025e-05,
+      "loss": 0.5368,
+      "step": 2840
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3584249875658148e-05,
+      "loss": 0.7766,
+      "step": 2841
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3579971698145813e-05,
+      "loss": 0.7593,
+      "step": 2842
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3575692768945313e-05,
+      "loss": 0.6391,
+      "step": 2843
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3571413088955093e-05,
+      "loss": 0.6465,
+      "step": 2844
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3567132659073763e-05,
+      "loss": 0.696,
+      "step": 2845
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3562851480200086e-05,
+      "loss": 0.7552,
+      "step": 2846
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.355856955323298e-05,
+      "loss": 0.7417,
+      "step": 2847
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3554286879071525e-05,
+      "loss": 0.6141,
+      "step": 2848
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.355000345861496e-05,
+      "loss": 0.7446,
+      "step": 2849
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.354571929276267e-05,
+      "loss": 0.7061,
+      "step": 2850
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3541434382414205e-05,
+      "loss": 0.704,
+      "step": 2851
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3537148728469269e-05,
+      "loss": 0.7284,
+      "step": 2852
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3532862331827725e-05,
+      "loss": 0.7346,
+      "step": 2853
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.352857519338959e-05,
+      "loss": 0.6002,
+      "step": 2854
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3524287314055034e-05,
+      "loss": 0.7345,
+      "step": 2855
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3519998694724384e-05,
+      "loss": 0.7262,
+      "step": 2856
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3515709336298124e-05,
+      "loss": 0.6285,
+      "step": 2857
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3511419239676894e-05,
+      "loss": 0.7316,
+      "step": 2858
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3507128405761487e-05,
+      "loss": 0.7076,
+      "step": 2859
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.350283683545285e-05,
+      "loss": 0.65,
+      "step": 2860
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3498544529652087e-05,
+      "loss": 0.5897,
+      "step": 2861
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3494251489260452e-05,
+      "loss": 0.7051,
+      "step": 2862
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3489957715179362e-05,
+      "loss": 0.7293,
+      "step": 2863
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3485663208310378e-05,
+      "loss": 0.7179,
+      "step": 2864
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3481367969555224e-05,
+      "loss": 0.7081,
+      "step": 2865
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3477071999815767e-05,
+      "loss": 0.7008,
+      "step": 2866
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3472775299994036e-05,
+      "loss": 0.7379,
+      "step": 2867
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3468477870992213e-05,
+      "loss": 0.725,
+      "step": 2868
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3464179713712633e-05,
+      "loss": 0.6035,
+      "step": 2869
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3459880829057775e-05,
+      "loss": 0.6874,
+      "step": 2870
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.345558121793028e-05,
+      "loss": 0.7277,
+      "step": 2871
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3451280881232942e-05,
+      "loss": 0.5925,
+      "step": 2872
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3446979819868703e-05,
+      "loss": 0.7122,
+      "step": 2873
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3442678034740659e-05,
+      "loss": 0.7673,
+      "step": 2874
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3438375526752055e-05,
+      "loss": 0.7695,
+      "step": 2875
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3434072296806295e-05,
+      "loss": 0.5822,
+      "step": 2876
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3429768345806926e-05,
+      "loss": 0.7406,
+      "step": 2877
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3425463674657652e-05,
+      "loss": 0.5753,
+      "step": 2878
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3421158284262327e-05,
+      "loss": 0.6879,
+      "step": 2879
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3416852175524955e-05,
+      "loss": 0.7241,
+      "step": 2880
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.341254534934969e-05,
+      "loss": 0.7101,
+      "step": 2881
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3408237806640842e-05,
+      "loss": 0.5831,
+      "step": 2882
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3403929548302865e-05,
+      "loss": 0.682,
+      "step": 2883
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3399620575240367e-05,
+      "loss": 0.7653,
+      "step": 2884
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3395310888358102e-05,
+      "loss": 0.7404,
+      "step": 2885
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3391000488560982e-05,
+      "loss": 0.7209,
+      "step": 2886
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3386689376754056e-05,
+      "loss": 0.729,
+      "step": 2887
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3382377553842536e-05,
+      "loss": 0.7108,
+      "step": 2888
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3378065020731774e-05,
+      "loss": 0.6999,
+      "step": 2889
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3373751778327279e-05,
+      "loss": 0.7617,
+      "step": 2890
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3369437827534698e-05,
+      "loss": 0.6108,
+      "step": 2891
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3365123169259839e-05,
+      "loss": 0.7127,
+      "step": 2892
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.336080780440865e-05,
+      "loss": 0.7185,
+      "step": 2893
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3356491733887228e-05,
+      "loss": 0.7428,
+      "step": 2894
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3352174958601823e-05,
+      "loss": 0.5714,
+      "step": 2895
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3347857479458828e-05,
+      "loss": 0.5753,
+      "step": 2896
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3343539297364789e-05,
+      "loss": 0.7104,
+      "step": 2897
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3339220413226394e-05,
+      "loss": 0.7075,
+      "step": 2898
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.333490082795048e-05,
+      "loss": 0.771,
+      "step": 2899
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3330580542444036e-05,
+      "loss": 0.7239,
+      "step": 2900
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3326259557614189e-05,
+      "loss": 0.582,
+      "step": 2901
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3321937874368223e-05,
+      "loss": 0.7375,
+      "step": 2902
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.331761549361356e-05,
+      "loss": 0.5984,
+      "step": 2903
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3313292416257776e-05,
+      "loss": 0.6996,
+      "step": 2904
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3308968643208583e-05,
+      "loss": 0.6049,
+      "step": 2905
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3304644175373854e-05,
+      "loss": 0.7319,
+      "step": 2906
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3300319013661592e-05,
+      "loss": 0.6306,
+      "step": 2907
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3295993158979956e-05,
+      "loss": 0.7039,
+      "step": 2908
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3291666612237246e-05,
+      "loss": 0.7328,
+      "step": 2909
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3287339374341914e-05,
+      "loss": 0.6002,
+      "step": 2910
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3283011446202545e-05,
+      "loss": 0.7526,
+      "step": 2911
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.327868282872788e-05,
+      "loss": 0.7197,
+      "step": 2912
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.32743535228268e-05,
+      "loss": 0.7232,
+      "step": 2913
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.327002352940833e-05,
+      "loss": 0.7766,
+      "step": 2914
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.326569284938164e-05,
+      "loss": 0.6444,
+      "step": 2915
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3261361483656048e-05,
+      "loss": 0.7193,
+      "step": 2916
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.325702943314101e-05,
+      "loss": 0.7183,
+      "step": 2917
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3252696698746132e-05,
+      "loss": 0.7656,
+      "step": 2918
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3248363281381154e-05,
+      "loss": 0.7077,
+      "step": 2919
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3244029181955969e-05,
+      "loss": 0.7067,
+      "step": 2920
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3239694401380609e-05,
+      "loss": 0.5875,
+      "step": 2921
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.323535894056525e-05,
+      "loss": 0.6017,
+      "step": 2922
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3231022800420212e-05,
+      "loss": 0.5716,
+      "step": 2923
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3226685981855956e-05,
+      "loss": 0.7432,
+      "step": 2924
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3222348485783084e-05,
+      "loss": 0.74,
+      "step": 2925
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3218010313112341e-05,
+      "loss": 0.7446,
+      "step": 2926
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3213671464754616e-05,
+      "loss": 0.737,
+      "step": 2927
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3209331941620943e-05,
+      "loss": 0.7159,
+      "step": 2928
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3204991744622485e-05,
+      "loss": 0.5806,
+      "step": 2929
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.320065087467056e-05,
+      "loss": 0.7063,
+      "step": 2930
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3196309332676624e-05,
+      "loss": 0.7567,
+      "step": 2931
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.319196711955227e-05,
+      "loss": 0.6671,
+      "step": 2932
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3187624236209233e-05,
+      "loss": 0.5998,
+      "step": 2933
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3183280683559393e-05,
+      "loss": 0.7209,
+      "step": 2934
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3178936462514767e-05,
+      "loss": 0.7388,
+      "step": 2935
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3174591573987512e-05,
+      "loss": 0.7398,
+      "step": 2936
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3170246018889925e-05,
+      "loss": 0.729,
+      "step": 2937
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3165899798134444e-05,
+      "loss": 0.7353,
+      "step": 2938
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.316155291263365e-05,
+      "loss": 0.745,
+      "step": 2939
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3157205363300258e-05,
+      "loss": 0.7254,
+      "step": 2940
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3152857151047126e-05,
+      "loss": 0.7217,
+      "step": 2941
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3148508276787248e-05,
+      "loss": 0.7019,
+      "step": 2942
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3144158741433763e-05,
+      "loss": 0.7113,
+      "step": 2943
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3139808545899939e-05,
+      "loss": 0.7527,
+      "step": 2944
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3135457691099192e-05,
+      "loss": 0.6977,
+      "step": 2945
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3131106177945071e-05,
+      "loss": 0.5947,
+      "step": 2946
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.312675400735127e-05,
+      "loss": 0.7169,
+      "step": 2947
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3122401180231607e-05,
+      "loss": 0.6876,
+      "step": 2948
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3118047697500055e-05,
+      "loss": 0.7127,
+      "step": 2949
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3113693560070714e-05,
+      "loss": 0.7226,
+      "step": 2950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3109338768857824e-05,
+      "loss": 0.7644,
+      "step": 2951
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3104983324775762e-05,
+      "loss": 0.7427,
+      "step": 2952
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3100627228739045e-05,
+      "loss": 0.7063,
+      "step": 2953
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.309627048166232e-05,
+      "loss": 0.7269,
+      "step": 2954
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3091913084460376e-05,
+      "loss": 0.5686,
+      "step": 2955
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3087555038048139e-05,
+      "loss": 0.7143,
+      "step": 2956
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.308319634334067e-05,
+      "loss": 0.7243,
+      "step": 2957
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3078837001253164e-05,
+      "loss": 0.7248,
+      "step": 2958
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3074477012700956e-05,
+      "loss": 0.7069,
+      "step": 2959
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3070116378599513e-05,
+      "loss": 0.6098,
+      "step": 2960
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3065755099864446e-05,
+      "loss": 0.7544,
+      "step": 2961
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.306139317741148e-05,
+      "loss": 0.6809,
+      "step": 2962
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3057030612156502e-05,
+      "loss": 0.7174,
+      "step": 2963
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3052667405015518e-05,
+      "loss": 0.7147,
+      "step": 2964
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3048303556904674e-05,
+      "loss": 0.7582,
+      "step": 2965
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3043939068740246e-05,
+      "loss": 0.5903,
+      "step": 2966
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3039573941438648e-05,
+      "loss": 0.7259,
+      "step": 2967
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3035208175916431e-05,
+      "loss": 0.6333,
+      "step": 2968
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.303084177309027e-05,
+      "loss": 0.7214,
+      "step": 2969
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3026474733876988e-05,
+      "loss": 0.7477,
+      "step": 2970
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.302210705919353e-05,
+      "loss": 0.6341,
+      "step": 2971
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3017738749956976e-05,
+      "loss": 0.7122,
+      "step": 2972
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3013369807084547e-05,
+      "loss": 0.6047,
+      "step": 2973
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3009000231493591e-05,
+      "loss": 0.7176,
+      "step": 2974
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3004630024101589e-05,
+      "loss": 0.778,
+      "step": 2975
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3000259185826151e-05,
+      "loss": 0.5767,
+      "step": 2976
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.299588771758503e-05,
+      "loss": 0.6961,
+      "step": 2977
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.29915156202961e-05,
+      "loss": 0.7547,
+      "step": 2978
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2987142894877378e-05,
+      "loss": 0.7429,
+      "step": 2979
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2982769542246998e-05,
+      "loss": 0.6953,
+      "step": 2980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2978395563323245e-05,
+      "loss": 0.6699,
+      "step": 2981
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2974020959024517e-05,
+      "loss": 0.7859,
+      "step": 2982
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2969645730269357e-05,
+      "loss": 0.5687,
+      "step": 2983
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.296526987797643e-05,
+      "loss": 0.7065,
+      "step": 2984
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2960893403064539e-05,
+      "loss": 0.6169,
+      "step": 2985
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2956516306452613e-05,
+      "loss": 0.7042,
+      "step": 2986
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2952138589059708e-05,
+      "loss": 0.7334,
+      "step": 2987
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.294776025180502e-05,
+      "loss": 0.6912,
+      "step": 2988
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2943381295607872e-05,
+      "loss": 0.6665,
+      "step": 2989
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2939001721387712e-05,
+      "loss": 0.5789,
+      "step": 2990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.293462153006412e-05,
+      "loss": 0.7242,
+      "step": 2991
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.293024072255681e-05,
+      "loss": 0.7571,
+      "step": 2992
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2925859299785623e-05,
+      "loss": 0.7226,
+      "step": 2993
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2921477262670523e-05,
+      "loss": 0.7247,
+      "step": 2994
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2917094612131611e-05,
+      "loss": 0.7358,
+      "step": 2995
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2912711349089113e-05,
+      "loss": 0.7368,
+      "step": 2996
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.290832747446339e-05,
+      "loss": 0.7359,
+      "step": 2997
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2903942989174915e-05,
+      "loss": 0.7336,
+      "step": 2998
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.289955789414431e-05,
+      "loss": 0.7234,
+      "step": 2999
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2895172190292311e-05,
+      "loss": 0.6952,
+      "step": 3000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2890785878539789e-05,
+      "loss": 0.707,
+      "step": 3001
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2886398959807736e-05,
+      "loss": 0.6021,
+      "step": 3002
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2882011435017277e-05,
+      "loss": 0.7313,
+      "step": 3003
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2877623305089663e-05,
+      "loss": 0.7279,
+      "step": 3004
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2873234570946268e-05,
+      "loss": 0.7187,
+      "step": 3005
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2868845233508604e-05,
+      "loss": 0.7023,
+      "step": 3006
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2864455293698295e-05,
+      "loss": 0.7104,
+      "step": 3007
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.28600647524371e-05,
+      "loss": 0.7235,
+      "step": 3008
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2855673610646903e-05,
+      "loss": 0.7014,
+      "step": 3009
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2851281869249717e-05,
+      "loss": 0.7071,
+      "step": 3010
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2846889529167676e-05,
+      "loss": 0.7037,
+      "step": 3011
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2842496591323037e-05,
+      "loss": 0.7304,
+      "step": 3012
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2838103056638191e-05,
+      "loss": 0.7143,
+      "step": 3013
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.283370892603565e-05,
+      "loss": 0.6833,
+      "step": 3014
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2829314200438057e-05,
+      "loss": 0.7389,
+      "step": 3015
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2824918880768163e-05,
+      "loss": 0.5906,
+      "step": 3016
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2820522967948863e-05,
+      "loss": 0.7071,
+      "step": 3017
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2816126462903169e-05,
+      "loss": 0.744,
+      "step": 3018
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2811729366554208e-05,
+      "loss": 0.5841,
+      "step": 3019
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2807331679825252e-05,
+      "loss": 0.7175,
+      "step": 3020
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.280293340363968e-05,
+      "loss": 0.5888,
+      "step": 3021
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2798534538921002e-05,
+      "loss": 0.5674,
+      "step": 3022
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.279413508659284e-05,
+      "loss": 0.6226,
+      "step": 3023
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2789735047578962e-05,
+      "loss": 0.7257,
+      "step": 3024
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2785334422803239e-05,
+      "loss": 0.7347,
+      "step": 3025
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2780933213189672e-05,
+      "loss": 0.7444,
+      "step": 3026
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2776531419662388e-05,
+      "loss": 0.702,
+      "step": 3027
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2772129043145631e-05,
+      "loss": 0.7103,
+      "step": 3028
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2767726084563772e-05,
+      "loss": 0.7225,
+      "step": 3029
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.27633225448413e-05,
+      "loss": 0.5801,
+      "step": 3030
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2758918424902826e-05,
+      "loss": 0.7486,
+      "step": 3031
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.275451372567309e-05,
+      "loss": 0.6839,
+      "step": 3032
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2750108448076945e-05,
+      "loss": 0.7367,
+      "step": 3033
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2745702593039369e-05,
+      "loss": 0.7456,
+      "step": 3034
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2741296161485463e-05,
+      "loss": 0.759,
+      "step": 3035
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.273688915434045e-05,
+      "loss": 0.5993,
+      "step": 3036
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2732481572529664e-05,
+      "loss": 0.7311,
+      "step": 3037
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2728073416978572e-05,
+      "loss": 0.727,
+      "step": 3038
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2723664688612756e-05,
+      "loss": 0.7098,
+      "step": 3039
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2719255388357915e-05,
+      "loss": 0.5793,
+      "step": 3040
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2714845517139876e-05,
+      "loss": 0.7169,
+      "step": 3041
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2710435075884577e-05,
+      "loss": 0.6859,
+      "step": 3042
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2706024065518087e-05,
+      "loss": 0.7069,
+      "step": 3043
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2701612486966578e-05,
+      "loss": 0.7285,
+      "step": 3044
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2697200341156361e-05,
+      "loss": 0.7526,
+      "step": 3045
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2692787629013852e-05,
+      "loss": 0.7183,
+      "step": 3046
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2688374351465589e-05,
+      "loss": 0.7146,
+      "step": 3047
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2683960509438227e-05,
+      "loss": 0.7336,
+      "step": 3048
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2679546103858549e-05,
+      "loss": 0.7373,
+      "step": 3049
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2675131135653444e-05,
+      "loss": 0.6969,
+      "step": 3050
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2670715605749928e-05,
+      "loss": 0.7097,
+      "step": 3051
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2666299515075131e-05,
+      "loss": 0.7179,
+      "step": 3052
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2661882864556301e-05,
+      "loss": 0.7243,
+      "step": 3053
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2657465655120807e-05,
+      "loss": 0.7191,
+      "step": 3054
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2653047887696123e-05,
+      "loss": 0.7088,
+      "step": 3055
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.264862956320986e-05,
+      "loss": 0.5389,
+      "step": 3056
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2644210682589731e-05,
+      "loss": 0.7234,
+      "step": 3057
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.263979124676357e-05,
+      "loss": 0.5851,
+      "step": 3058
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2635371256659328e-05,
+      "loss": 0.7043,
+      "step": 3059
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2630950713205072e-05,
+      "loss": 0.5585,
+      "step": 3060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2626529617328989e-05,
+      "loss": 0.7576,
+      "step": 3061
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2622107969959373e-05,
+      "loss": 0.7076,
+      "step": 3062
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2617685772024641e-05,
+      "loss": 0.682,
+      "step": 3063
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2613263024453323e-05,
+      "loss": 0.6987,
+      "step": 3064
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.260883972817407e-05,
+      "loss": 0.696,
+      "step": 3065
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2604415884115637e-05,
+      "loss": 0.7393,
+      "step": 3066
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2599991493206904e-05,
+      "loss": 0.5597,
+      "step": 3067
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2595566556376863e-05,
+      "loss": 0.6926,
+      "step": 3068
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2591141074554616e-05,
+      "loss": 0.7077,
+      "step": 3069
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2586715048669387e-05,
+      "loss": 0.7104,
+      "step": 3070
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2582288479650508e-05,
+      "loss": 0.7168,
+      "step": 3071
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2577861368427431e-05,
+      "loss": 0.7144,
+      "step": 3072
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2573433715929712e-05,
+      "loss": 0.6999,
+      "step": 3073
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2569005523087033e-05,
+      "loss": 0.7287,
+      "step": 3074
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2564576790829182e-05,
+      "loss": 0.6906,
+      "step": 3075
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2560147520086063e-05,
+      "loss": 0.7602,
+      "step": 3076
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2555717711787687e-05,
+      "loss": 0.7473,
+      "step": 3077
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2551287366864187e-05,
+      "loss": 0.7727,
+      "step": 3078
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2546856486245804e-05,
+      "loss": 0.7401,
+      "step": 3079
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2542425070862889e-05,
+      "loss": 0.7129,
+      "step": 3080
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2537993121645913e-05,
+      "loss": 0.5995,
+      "step": 3081
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.253356063952545e-05,
+      "loss": 0.7156,
+      "step": 3082
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2529127625432193e-05,
+      "loss": 0.7233,
+      "step": 3083
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2524694080296942e-05,
+      "loss": 0.5795,
+      "step": 3084
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2520260005050608e-05,
+      "loss": 0.7043,
+      "step": 3085
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2515825400624224e-05,
+      "loss": 0.6827,
+      "step": 3086
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2511390267948918e-05,
+      "loss": 0.6999,
+      "step": 3087
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2506954607955936e-05,
+      "loss": 0.6165,
+      "step": 3088
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2502518421576641e-05,
+      "loss": 0.7005,
+      "step": 3089
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2498081709742503e-05,
+      "loss": 0.6521,
+      "step": 3090
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.249364447338509e-05,
+      "loss": 0.6036,
+      "step": 3091
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2489206713436096e-05,
+      "loss": 0.7048,
+      "step": 3092
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.248476843082732e-05,
+      "loss": 0.7571,
+      "step": 3093
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.248032962649067e-05,
+      "loss": 0.7199,
+      "step": 3094
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2475890301358163e-05,
+      "loss": 0.7431,
+      "step": 3095
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2471450456361925e-05,
+      "loss": 0.7133,
+      "step": 3096
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2467010092434195e-05,
+      "loss": 0.6382,
+      "step": 3097
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2462569210507312e-05,
+      "loss": 0.5995,
+      "step": 3098
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2458127811513736e-05,
+      "loss": 0.7404,
+      "step": 3099
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.245368589638603e-05,
+      "loss": 0.6072,
+      "step": 3100
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2449243466056862e-05,
+      "loss": 0.5976,
+      "step": 3101
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2444800521459007e-05,
+      "loss": 0.7263,
+      "step": 3102
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.244035706352536e-05,
+      "loss": 0.7064,
+      "step": 3103
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2435913093188911e-05,
+      "loss": 0.6682,
+      "step": 3104
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2431468611382759e-05,
+      "loss": 0.7184,
+      "step": 3105
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2427023619040124e-05,
+      "loss": 0.7322,
+      "step": 3106
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2422578117094312e-05,
+      "loss": 0.6939,
+      "step": 3107
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2418132106478758e-05,
+      "loss": 0.7787,
+      "step": 3108
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.241368558812698e-05,
+      "loss": 0.7335,
+      "step": 3109
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2409238562972626e-05,
+      "loss": 0.7233,
+      "step": 3110
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2404791031949435e-05,
+      "loss": 0.7149,
+      "step": 3111
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2400342995991255e-05,
+      "loss": 0.5633,
+      "step": 3112
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2395894456032045e-05,
+      "loss": 0.7268,
+      "step": 3113
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2391445413005866e-05,
+      "loss": 0.7175,
+      "step": 3114
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2386995867846891e-05,
+      "loss": 0.6831,
+      "step": 3115
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2382545821489383e-05,
+      "loss": 0.747,
+      "step": 3116
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2378095274867725e-05,
+      "loss": 0.7054,
+      "step": 3117
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2373644228916402e-05,
+      "loss": 0.5553,
+      "step": 3118
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2369192684569997e-05,
+      "loss": 0.7139,
+      "step": 3119
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2364740642763205e-05,
+      "loss": 0.5961,
+      "step": 3120
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2360288104430824e-05,
+      "loss": 0.7217,
+      "step": 3121
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.235583507050776e-05,
+      "loss": 0.7421,
+      "step": 3122
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2351381541929006e-05,
+      "loss": 0.7469,
+      "step": 3123
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2346927519629684e-05,
+      "loss": 0.5779,
+      "step": 3124
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2342473004545e-05,
+      "loss": 0.585,
+      "step": 3125
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2338017997610271e-05,
+      "loss": 0.6762,
+      "step": 3126
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2333562499760919e-05,
+      "loss": 0.7752,
+      "step": 3127
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2329106511932465e-05,
+      "loss": 0.6781,
+      "step": 3128
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2324650035060537e-05,
+      "loss": 0.7061,
+      "step": 3129
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2320193070080855e-05,
+      "loss": 0.7117,
+      "step": 3130
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2315735617929262e-05,
+      "loss": 0.7412,
+      "step": 3131
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2311277679541683e-05,
+      "loss": 0.6876,
+      "step": 3132
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.230681925585416e-05,
+      "loss": 0.6884,
+      "step": 3133
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2302360347802819e-05,
+      "loss": 0.71,
+      "step": 3134
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2297900956323907e-05,
+      "loss": 0.7067,
+      "step": 3135
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2293441082353768e-05,
+      "loss": 0.6637,
+      "step": 3136
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2288980726828834e-05,
+      "loss": 0.6125,
+      "step": 3137
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2284519890685652e-05,
+      "loss": 0.7094,
+      "step": 3138
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2280058574860864e-05,
+      "loss": 0.7329,
+      "step": 3139
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.227559678029122e-05,
+      "loss": 0.7126,
+      "step": 3140
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2271134507913557e-05,
+      "loss": 0.7079,
+      "step": 3141
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2266671758664827e-05,
+      "loss": 0.6936,
+      "step": 3142
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2262208533482073e-05,
+      "loss": 0.7567,
+      "step": 3143
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2257744833302441e-05,
+      "loss": 0.662,
+      "step": 3144
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2253280659063172e-05,
+      "loss": 0.7134,
+      "step": 3145
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2248816011701612e-05,
+      "loss": 0.7204,
+      "step": 3146
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.224435089215521e-05,
+      "loss": 0.6689,
+      "step": 3147
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2239885301361502e-05,
+      "loss": 0.6211,
+      "step": 3148
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2235419240258136e-05,
+      "loss": 0.7299,
+      "step": 3149
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2230952709782848e-05,
+      "loss": 0.7379,
+      "step": 3150
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2226485710873481e-05,
+      "loss": 0.5827,
+      "step": 3151
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2222018244467968e-05,
+      "loss": 0.7107,
+      "step": 3152
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2217550311504352e-05,
+      "loss": 0.7375,
+      "step": 3153
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2213081912920762e-05,
+      "loss": 0.7361,
+      "step": 3154
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2208613049655427e-05,
+      "loss": 0.705,
+      "step": 3155
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2204143722646677e-05,
+      "loss": 0.7028,
+      "step": 3156
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2199673932832945e-05,
+      "loss": 0.7085,
+      "step": 3157
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2195203681152752e-05,
+      "loss": 0.7129,
+      "step": 3158
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2190732968544712e-05,
+      "loss": 0.717,
+      "step": 3159
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2186261795947549e-05,
+      "loss": 0.7604,
+      "step": 3160
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2181790164300079e-05,
+      "loss": 0.7157,
+      "step": 3161
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2177318074541205e-05,
+      "loss": 0.6874,
+      "step": 3162
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2172845527609936e-05,
+      "loss": 0.7452,
+      "step": 3163
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2168372524445376e-05,
+      "loss": 0.6066,
+      "step": 3164
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2163899065986726e-05,
+      "loss": 0.7098,
+      "step": 3165
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2159425153173275e-05,
+      "loss": 0.7172,
+      "step": 3166
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2154950786944414e-05,
+      "loss": 0.7305,
+      "step": 3167
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2150475968239632e-05,
+      "loss": 0.6264,
+      "step": 3168
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2146000697998502e-05,
+      "loss": 0.6616,
+      "step": 3169
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2141524977160699e-05,
+      "loss": 0.7103,
+      "step": 3170
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2137048806665997e-05,
+      "loss": 0.6875,
+      "step": 3171
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2132572187454254e-05,
+      "loss": 0.7333,
+      "step": 3172
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2128095120465434e-05,
+      "loss": 0.7612,
+      "step": 3173
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.212361760663958e-05,
+      "loss": 0.7268,
+      "step": 3174
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2119139646916842e-05,
+      "loss": 0.6965,
+      "step": 3175
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2114661242237465e-05,
+      "loss": 0.7197,
+      "step": 3176
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2110182393541769e-05,
+      "loss": 0.7273,
+      "step": 3177
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.210570310177019e-05,
+      "loss": 0.6547,
+      "step": 3178
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2101223367863244e-05,
+      "loss": 0.6792,
+      "step": 3179
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2096743192761542e-05,
+      "loss": 0.6011,
+      "step": 3180
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2092262577405784e-05,
+      "loss": 0.7569,
+      "step": 3181
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2087781522736772e-05,
+      "loss": 0.7216,
+      "step": 3182
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2083300029695398e-05,
+      "loss": 0.7323,
+      "step": 3183
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2078818099222633e-05,
+      "loss": 0.6021,
+      "step": 3184
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2074335732259559e-05,
+      "loss": 0.7018,
+      "step": 3185
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2069852929747339e-05,
+      "loss": 0.6205,
+      "step": 3186
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2065369692627225e-05,
+      "loss": 0.5834,
+      "step": 3187
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2060886021840563e-05,
+      "loss": 0.5939,
+      "step": 3188
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2056401918328798e-05,
+      "loss": 0.7304,
+      "step": 3189
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2051917383033458e-05,
+      "loss": 0.5856,
+      "step": 3190
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2047432416896154e-05,
+      "loss": 0.7226,
+      "step": 3191
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2042947020858606e-05,
+      "loss": 0.6992,
+      "step": 3192
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2038461195862612e-05,
+      "loss": 0.6766,
+      "step": 3193
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2033974942850059e-05,
+      "loss": 0.7387,
+      "step": 3194
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.202948826276293e-05,
+      "loss": 0.7087,
+      "step": 3195
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2025001156543297e-05,
+      "loss": 0.7313,
+      "step": 3196
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2020513625133319e-05,
+      "loss": 0.6881,
+      "step": 3197
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2016025669475243e-05,
+      "loss": 0.6188,
+      "step": 3198
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2011537290511401e-05,
+      "loss": 0.712,
+      "step": 3199
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2007048489184233e-05,
+      "loss": 0.728,
+      "step": 3200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2002559266436249e-05,
+      "loss": 0.7041,
+      "step": 3201
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1998069623210047e-05,
+      "loss": 0.6715,
+      "step": 3202
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1993579560448325e-05,
+      "loss": 0.5991,
+      "step": 3203
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1989089079093865e-05,
+      "loss": 0.6406,
+      "step": 3204
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1984598180089533e-05,
+      "loss": 0.7442,
+      "step": 3205
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.198010686437828e-05,
+      "loss": 0.7179,
+      "step": 3206
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1975615132903158e-05,
+      "loss": 0.6987,
+      "step": 3207
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1971122986607295e-05,
+      "loss": 0.7384,
+      "step": 3208
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1966630426433903e-05,
+      "loss": 0.7378,
+      "step": 3209
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1962137453326295e-05,
+      "loss": 0.6811,
+      "step": 3210
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1957644068227861e-05,
+      "loss": 0.7363,
+      "step": 3211
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1953150272082074e-05,
+      "loss": 0.7526,
+      "step": 3212
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.19486560658325e-05,
+      "loss": 0.7196,
+      "step": 3213
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.194416145042279e-05,
+      "loss": 0.6992,
+      "step": 3214
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1939666426796687e-05,
+      "loss": 0.739,
+      "step": 3215
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1935170995898002e-05,
+      "loss": 0.7427,
+      "step": 3216
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1930675158670648e-05,
+      "loss": 0.5622,
+      "step": 3217
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.192617891605862e-05,
+      "loss": 0.6945,
+      "step": 3218
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1921682269005989e-05,
+      "loss": 0.7373,
+      "step": 3219
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1917185218456921e-05,
+      "loss": 0.6214,
+      "step": 3220
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1912687765355669e-05,
+      "loss": 0.5764,
+      "step": 3221
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1908189910646562e-05,
+      "loss": 0.7346,
+      "step": 3222
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1903691655274012e-05,
+      "loss": 0.5933,
+      "step": 3223
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.189919300018252e-05,
+      "loss": 0.6883,
+      "step": 3224
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.189469394631668e-05,
+      "loss": 0.7166,
+      "step": 3225
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.189019449462115e-05,
+      "loss": 0.6796,
+      "step": 3226
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1885694646040685e-05,
+      "loss": 0.7073,
+      "step": 3227
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.188119440152012e-05,
+      "loss": 0.7137,
+      "step": 3228
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1876693762004377e-05,
+      "loss": 0.7119,
+      "step": 3229
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1872192728438453e-05,
+      "loss": 0.5554,
+      "step": 3230
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1867691301767428e-05,
+      "loss": 0.7001,
+      "step": 3231
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1863189482936475e-05,
+      "loss": 0.702,
+      "step": 3232
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1858687272890846e-05,
+      "loss": 0.5891,
+      "step": 3233
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1854184672575863e-05,
+      "loss": 0.6897,
+      "step": 3234
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.184968168293694e-05,
+      "loss": 0.6117,
+      "step": 3235
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.184517830491958e-05,
+      "loss": 0.7166,
+      "step": 3236
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1840674539469352e-05,
+      "loss": 0.5664,
+      "step": 3237
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1836170387531914e-05,
+      "loss": 0.7232,
+      "step": 3238
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1831665850053004e-05,
+      "loss": 0.7055,
+      "step": 3239
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.182716092797845e-05,
+      "loss": 0.6899,
+      "step": 3240
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.182265562225414e-05,
+      "loss": 0.7184,
+      "step": 3241
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1818149933826062e-05,
+      "loss": 0.6883,
+      "step": 3242
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1813643863640278e-05,
+      "loss": 0.698,
+      "step": 3243
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1809137412642927e-05,
+      "loss": 0.6211,
+      "step": 3244
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1804630581780227e-05,
+      "loss": 0.6853,
+      "step": 3245
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1800123371998487e-05,
+      "loss": 0.7367,
+      "step": 3246
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1795615784244086e-05,
+      "loss": 0.725,
+      "step": 3247
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1791107819463479e-05,
+      "loss": 0.5799,
+      "step": 3248
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1786599478603206e-05,
+      "loss": 0.7156,
+      "step": 3249
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1782090762609893e-05,
+      "loss": 0.7524,
+      "step": 3250
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.177758167243023e-05,
+      "loss": 0.7021,
+      "step": 3251
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.177307220901099e-05,
+      "loss": 0.7239,
+      "step": 3252
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1768562373299033e-05,
+      "loss": 0.6975,
+      "step": 3253
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1764052166241293e-05,
+      "loss": 0.7327,
+      "step": 3254
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1759541588784774e-05,
+      "loss": 0.6272,
+      "step": 3255
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1755030641876562e-05,
+      "loss": 0.7286,
+      "step": 3256
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1750519326463829e-05,
+      "loss": 0.6756,
+      "step": 3257
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1746007643493818e-05,
+      "loss": 0.6321,
+      "step": 3258
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1741495593913843e-05,
+      "loss": 0.7314,
+      "step": 3259
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1736983178671303e-05,
+      "loss": 0.7522,
+      "step": 3260
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1732470398713678e-05,
+      "loss": 0.724,
+      "step": 3261
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.172795725498851e-05,
+      "loss": 0.7278,
+      "step": 3262
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1723443748443428e-05,
+      "loss": 0.6974,
+      "step": 3263
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1718929880026137e-05,
+      "loss": 0.712,
+      "step": 3264
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1714415650684417e-05,
+      "loss": 0.6842,
+      "step": 3265
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1709901061366118e-05,
+      "loss": 0.7476,
+      "step": 3266
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.170538611301917e-05,
+      "loss": 0.6996,
+      "step": 3267
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1700870806591586e-05,
+      "loss": 0.6473,
+      "step": 3268
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.169635514303144e-05,
+      "loss": 0.7251,
+      "step": 3269
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1691839123286886e-05,
+      "loss": 0.6874,
+      "step": 3270
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.168732274830616e-05,
+      "loss": 0.6664,
+      "step": 3271
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.168280601903757e-05,
+      "loss": 0.742,
+      "step": 3272
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1678288936429487e-05,
+      "loss": 0.7714,
+      "step": 3273
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1673771501430368e-05,
+      "loss": 0.5348,
+      "step": 3274
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1669253714988744e-05,
+      "loss": 0.6918,
+      "step": 3275
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1664735578053214e-05,
+      "loss": 0.7206,
+      "step": 3276
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1660217091572453e-05,
+      "loss": 0.7311,
+      "step": 3277
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1655698256495206e-05,
+      "loss": 0.7116,
+      "step": 3278
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1651179073770302e-05,
+      "loss": 0.6842,
+      "step": 3279
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1646659544346631e-05,
+      "loss": 0.5899,
+      "step": 3280
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1642139669173159e-05,
+      "loss": 0.7069,
+      "step": 3281
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1637619449198932e-05,
+      "loss": 0.5561,
+      "step": 3282
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1633098885373057e-05,
+      "loss": 0.7,
+      "step": 3283
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1628577978644721e-05,
+      "loss": 0.7098,
+      "step": 3284
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1624056729963181e-05,
+      "loss": 0.71,
+      "step": 3285
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1619535140277767e-05,
+      "loss": 0.5837,
+      "step": 3286
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1615013210537876e-05,
+      "loss": 0.6966,
+      "step": 3287
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1610490941692976e-05,
+      "loss": 0.6481,
+      "step": 3288
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1605968334692617e-05,
+      "loss": 0.7185,
+      "step": 3289
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1601445390486414e-05,
+      "loss": 0.581,
+      "step": 3290
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1596922110024043e-05,
+      "loss": 0.6732,
+      "step": 3291
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1592398494255264e-05,
+      "loss": 0.7069,
+      "step": 3292
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1587874544129905e-05,
+      "loss": 0.7206,
+      "step": 3293
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1583350260597855e-05,
+      "loss": 0.5887,
+      "step": 3294
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1578825644609085e-05,
+      "loss": 0.7302,
+      "step": 3295
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1574300697113627e-05,
+      "loss": 0.6894,
+      "step": 3296
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.156977541906159e-05,
+      "loss": 0.7183,
+      "step": 3297
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1565249811403148e-05,
+      "loss": 0.701,
+      "step": 3298
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1560723875088539e-05,
+      "loss": 0.7357,
+      "step": 3299
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1556197611068087e-05,
+      "loss": 0.7515,
+      "step": 3300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1551671020292164e-05,
+      "loss": 0.695,
+      "step": 3301
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1547144103711221e-05,
+      "loss": 0.7528,
+      "step": 3302
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1542616862275777e-05,
+      "loss": 0.7223,
+      "step": 3303
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1538089296936423e-05,
+      "loss": 0.6979,
+      "step": 3304
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1533561408643813e-05,
+      "loss": 0.723,
+      "step": 3305
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1529033198348663e-05,
+      "loss": 0.7215,
+      "step": 3306
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.152450466700177e-05,
+      "loss": 0.6242,
+      "step": 3307
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1519975815553992e-05,
+      "loss": 0.7477,
+      "step": 3308
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1515446644956247e-05,
+      "loss": 0.6603,
+      "step": 3309
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1510917156159533e-05,
+      "loss": 0.6826,
+      "step": 3310
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1506387350114909e-05,
+      "loss": 0.7019,
+      "step": 3311
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1501857227773497e-05,
+      "loss": 0.6751,
+      "step": 3312
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1497326790086489e-05,
+      "loss": 0.6744,
+      "step": 3313
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1492796038005142e-05,
+      "loss": 0.6863,
+      "step": 3314
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1488264972480788e-05,
+      "loss": 0.6821,
+      "step": 3315
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1483733594464805e-05,
+      "loss": 0.6755,
+      "step": 3316
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1479201904908655e-05,
+      "loss": 0.7486,
+      "step": 3317
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1474669904763859e-05,
+      "loss": 0.7116,
+      "step": 3318
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1470137594982002e-05,
+      "loss": 0.7349,
+      "step": 3319
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1465604976514732e-05,
+      "loss": 0.7145,
+      "step": 3320
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1461072050313767e-05,
+      "loss": 0.7044,
+      "step": 3321
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1456538817330892e-05,
+      "loss": 0.7204,
+      "step": 3322
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1452005278517945e-05,
+      "loss": 0.721,
+      "step": 3323
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1447471434826836e-05,
+      "loss": 0.6953,
+      "step": 3324
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1442937287209543e-05,
+      "loss": 0.7598,
+      "step": 3325
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1438402836618099e-05,
+      "loss": 0.6911,
+      "step": 3326
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1433868084004604e-05,
+      "loss": 0.6929,
+      "step": 3327
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1429333030321221e-05,
+      "loss": 0.7156,
+      "step": 3328
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1424797676520185e-05,
+      "loss": 0.6504,
+      "step": 3329
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1420262023553776e-05,
+      "loss": 0.7415,
+      "step": 3330
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1415726072374352e-05,
+      "loss": 0.7153,
+      "step": 3331
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1411189823934328e-05,
+      "loss": 0.7081,
+      "step": 3332
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1406653279186185e-05,
+      "loss": 0.7014,
+      "step": 3333
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1402116439082455e-05,
+      "loss": 0.5535,
+      "step": 3334
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1397579304575748e-05,
+      "loss": 0.6803,
+      "step": 3335
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1393041876618727e-05,
+      "loss": 0.6979,
+      "step": 3336
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1388504156164112e-05,
+      "loss": 0.6725,
+      "step": 3337
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1383966144164698e-05,
+      "loss": 0.709,
+      "step": 3338
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1379427841573326e-05,
+      "loss": 0.7096,
+      "step": 3339
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1374889249342913e-05,
+      "loss": 0.7026,
+      "step": 3340
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1370350368426422e-05,
+      "loss": 0.7126,
+      "step": 3341
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1365811199776884e-05,
+      "loss": 0.6999,
+      "step": 3342
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1361271744347397e-05,
+      "loss": 0.6035,
+      "step": 3343
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.135673200309111e-05,
+      "loss": 0.724,
+      "step": 3344
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1352191976961228e-05,
+      "loss": 0.7122,
+      "step": 3345
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1347651666911026e-05,
+      "loss": 0.6636,
+      "step": 3346
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.134311107389384e-05,
+      "loss": 0.5383,
+      "step": 3347
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1338570198863053e-05,
+      "loss": 0.7104,
+      "step": 3348
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1334029042772114e-05,
+      "loss": 0.6983,
+      "step": 3349
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1329487606574542e-05,
+      "loss": 0.7407,
+      "step": 3350
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1324945891223893e-05,
+      "loss": 0.6864,
+      "step": 3351
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1320403897673797e-05,
+      "loss": 0.725,
+      "step": 3352
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1315861626877935e-05,
+      "loss": 0.696,
+      "step": 3353
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.131131907979006e-05,
+      "loss": 0.6929,
+      "step": 3354
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.130677625736396e-05,
+      "loss": 0.6862,
+      "step": 3355
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.13022331605535e-05,
+      "loss": 0.7246,
+      "step": 3356
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1297689790312592e-05,
+      "loss": 0.6992,
+      "step": 3357
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1293146147595216e-05,
+      "loss": 0.7062,
+      "step": 3358
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1288602233355394e-05,
+      "loss": 0.7302,
+      "step": 3359
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1284058048547217e-05,
+      "loss": 0.6737,
+      "step": 3360
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1279513594124835e-05,
+      "loss": 0.7258,
+      "step": 3361
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1274968871042444e-05,
+      "loss": 0.7193,
+      "step": 3362
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.12704238802543e-05,
+      "loss": 0.6964,
+      "step": 3363
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1265878622714716e-05,
+      "loss": 0.6732,
+      "step": 3364
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1261333099378071e-05,
+      "loss": 0.697,
+      "step": 3365
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1256787311198778e-05,
+      "loss": 0.708,
+      "step": 3366
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1252241259131325e-05,
+      "loss": 0.7122,
+      "step": 3367
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1247694944130248e-05,
+      "loss": 0.7153,
+      "step": 3368
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1243148367150138e-05,
+      "loss": 0.6948,
+      "step": 3369
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1238601529145638e-05,
+      "loss": 0.7246,
+      "step": 3370
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1234054431071456e-05,
+      "loss": 0.6969,
+      "step": 3371
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1229507073882344e-05,
+      "loss": 0.7067,
+      "step": 3372
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1224959458533115e-05,
+      "loss": 0.7132,
+      "step": 3373
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.122041158597863e-05,
+      "loss": 0.7339,
+      "step": 3374
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1215863457173812e-05,
+      "loss": 0.6838,
+      "step": 3375
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.121131507307363e-05,
+      "loss": 0.6962,
+      "step": 3376
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1206766434633111e-05,
+      "loss": 0.6714,
+      "step": 3377
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1202217542807335e-05,
+      "loss": 0.6768,
+      "step": 3378
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1197668398551438e-05,
+      "loss": 0.6053,
+      "step": 3379
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1193119002820601e-05,
+      "loss": 0.6715,
+      "step": 3380
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1188569356570063e-05,
+      "loss": 0.7196,
+      "step": 3381
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1184019460755115e-05,
+      "loss": 0.6838,
+      "step": 3382
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1179469316331106e-05,
+      "loss": 0.725,
+      "step": 3383
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1174918924253425e-05,
+      "loss": 0.6773,
+      "step": 3384
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.117036828547752e-05,
+      "loss": 0.696,
+      "step": 3385
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1165817400958898e-05,
+      "loss": 0.7189,
+      "step": 3386
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1161266271653102e-05,
+      "loss": 0.7062,
+      "step": 3387
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.115671489851574e-05,
+      "loss": 0.7235,
+      "step": 3388
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1152163282502458e-05,
+      "loss": 0.7351,
+      "step": 3389
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1147611424568971e-05,
+      "loss": 0.7218,
+      "step": 3390
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1143059325671029e-05,
+      "loss": 0.6436,
+      "step": 3391
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1138506986764435e-05,
+      "loss": 0.7148,
+      "step": 3392
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1133954408805053e-05,
+      "loss": 0.667,
+      "step": 3393
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1129401592748783e-05,
+      "loss": 0.6971,
+      "step": 3394
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1124848539551587e-05,
+      "loss": 0.7365,
+      "step": 3395
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1120295250169465e-05,
+      "loss": 0.6944,
+      "step": 3396
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1115741725558482e-05,
+      "loss": 0.5712,
+      "step": 3397
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1111187966674737e-05,
+      "loss": 0.7049,
+      "step": 3398
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1106633974474387e-05,
+      "loss": 0.699,
+      "step": 3399
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1102079749913634e-05,
+      "loss": 0.6974,
+      "step": 3400
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1097525293948732e-05,
+      "loss": 0.6659,
+      "step": 3401
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.109297060753598e-05,
+      "loss": 0.6024,
+      "step": 3402
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1088415691631729e-05,
+      "loss": 0.7093,
+      "step": 3403
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.108386054719238e-05,
+      "loss": 0.7145,
+      "step": 3404
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.107930517517437e-05,
+      "loss": 0.7289,
+      "step": 3405
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.10747495765342e-05,
+      "loss": 0.6991,
+      "step": 3406
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1070193752228407e-05,
+      "loss": 0.7178,
+      "step": 3407
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1065637703213586e-05,
+      "loss": 0.6947,
+      "step": 3408
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1061081430446363e-05,
+      "loss": 0.7081,
+      "step": 3409
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1056524934883424e-05,
+      "loss": 0.6758,
+      "step": 3410
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.10519682174815e-05,
+      "loss": 0.6119,
+      "step": 3411
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1047411279197369e-05,
+      "loss": 0.6138,
+      "step": 3412
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1042854120987846e-05,
+      "loss": 0.7049,
+      "step": 3413
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1038296743809804e-05,
+      "loss": 0.6856,
+      "step": 3414
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1033739148620161e-05,
+      "loss": 0.6953,
+      "step": 3415
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1029181336375871e-05,
+      "loss": 0.7439,
+      "step": 3416
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1024623308033942e-05,
+      "loss": 0.7468,
+      "step": 3417
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1020065064551426e-05,
+      "loss": 0.7015,
+      "step": 3418
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1015506606885418e-05,
+      "loss": 0.7623,
+      "step": 3419
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1010947935993059e-05,
+      "loss": 0.7026,
+      "step": 3420
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1006389052831534e-05,
+      "loss": 0.7092,
+      "step": 3421
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1001829958358079e-05,
+      "loss": 0.723,
+      "step": 3422
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0997270653529966e-05,
+      "loss": 0.715,
+      "step": 3423
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.099271113930451e-05,
+      "loss": 0.7091,
+      "step": 3424
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.098815141663908e-05,
+      "loss": 0.6225,
+      "step": 3425
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.098359148649108e-05,
+      "loss": 0.7288,
+      "step": 3426
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.097903134981796e-05,
+      "loss": 0.613,
+      "step": 3427
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0974471007577215e-05,
+      "loss": 0.7299,
+      "step": 3428
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0969910460726386e-05,
+      "loss": 0.6873,
+      "step": 3429
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0965349710223044e-05,
+      "loss": 0.7316,
+      "step": 3430
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0960788757024818e-05,
+      "loss": 0.6131,
+      "step": 3431
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0956227602089368e-05,
+      "loss": 0.7197,
+      "step": 3432
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0951666246374413e-05,
+      "loss": 0.7137,
+      "step": 3433
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0947104690837691e-05,
+      "loss": 0.5803,
+      "step": 3434
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0942542936436998e-05,
+      "loss": 0.6952,
+      "step": 3435
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0937980984130168e-05,
+      "loss": 0.7033,
+      "step": 3436
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0933418834875074e-05,
+      "loss": 0.7069,
+      "step": 3437
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0928856489629636e-05,
+      "loss": 0.6758,
+      "step": 3438
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0924293949351806e-05,
+      "loss": 0.5593,
+      "step": 3439
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.091973121499959e-05,
+      "loss": 0.6904,
+      "step": 3440
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0915168287531022e-05,
+      "loss": 0.7039,
+      "step": 3441
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.091060516790418e-05,
+      "loss": 0.7113,
+      "step": 3442
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.090604185707719e-05,
+      "loss": 0.7142,
+      "step": 3443
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0901478356008209e-05,
+      "loss": 0.713,
+      "step": 3444
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0896914665655438e-05,
+      "loss": 0.729,
+      "step": 3445
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0892350786977113e-05,
+      "loss": 0.6467,
+      "step": 3446
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0887786720931523e-05,
+      "loss": 0.6554,
+      "step": 3447
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0883222468476977e-05,
+      "loss": 0.6754,
+      "step": 3448
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.087865803057184e-05,
+      "loss": 0.7115,
+      "step": 3449
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0874093408174504e-05,
+      "loss": 0.7488,
+      "step": 3450
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0869528602243407e-05,
+      "loss": 0.6849,
+      "step": 3451
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0864963613737023e-05,
+      "loss": 0.7397,
+      "step": 3452
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0860398443613865e-05,
+      "loss": 0.7414,
+      "step": 3453
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0855833092832482e-05,
+      "loss": 0.7278,
+      "step": 3454
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0851267562351464e-05,
+      "loss": 0.7289,
+      "step": 3455
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0846701853129438e-05,
+      "loss": 0.6702,
+      "step": 3456
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0842135966125065e-05,
+      "loss": 0.702,
+      "step": 3457
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.083756990229705e-05,
+      "loss": 0.6848,
+      "step": 3458
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0833003662604131e-05,
+      "loss": 0.7133,
+      "step": 3459
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0828437248005081e-05,
+      "loss": 0.6912,
+      "step": 3460
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0823870659458713e-05,
+      "loss": 0.6915,
+      "step": 3461
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0819303897923874e-05,
+      "loss": 0.6027,
+      "step": 3462
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0814736964359452e-05,
+      "loss": 0.6558,
+      "step": 3463
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0810169859724363e-05,
+      "loss": 0.6956,
+      "step": 3464
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.080560258497757e-05,
+      "loss": 0.6898,
+      "step": 3465
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0801035141078062e-05,
+      "loss": 0.6062,
+      "step": 3466
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0796467528984868e-05,
+      "loss": 0.7135,
+      "step": 3467
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0791899749657052e-05,
+      "loss": 0.6041,
+      "step": 3468
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0787331804053708e-05,
+      "loss": 0.707,
+      "step": 3469
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0782763693133978e-05,
+      "loss": 0.6896,
+      "step": 3470
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.077819541785702e-05,
+      "loss": 0.6939,
+      "step": 3471
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0773626979182048e-05,
+      "loss": 0.586,
+      "step": 3472
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.076905837806829e-05,
+      "loss": 0.6438,
+      "step": 3473
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0764489615475022e-05,
+      "loss": 0.6646,
+      "step": 3474
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0759920692361548e-05,
+      "loss": 0.6335,
+      "step": 3475
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0755351609687203e-05,
+      "loss": 0.7182,
+      "step": 3476
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0750782368411362e-05,
+      "loss": 0.5894,
+      "step": 3477
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0746212969493435e-05,
+      "loss": 0.7279,
+      "step": 3478
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0741643413892853e-05,
+      "loss": 0.6001,
+      "step": 3479
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0737073702569092e-05,
+      "loss": 0.6714,
+      "step": 3480
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0732503836481654e-05,
+      "loss": 0.5762,
+      "step": 3481
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0727933816590075e-05,
+      "loss": 0.6975,
+      "step": 3482
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0723363643853928e-05,
+      "loss": 0.7107,
+      "step": 3483
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0718793319232812e-05,
+      "loss": 0.7432,
+      "step": 3484
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0714222843686359e-05,
+      "loss": 0.6845,
+      "step": 3485
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0709652218174236e-05,
+      "loss": 0.72,
+      "step": 3486
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0705081443656135e-05,
+      "loss": 0.6686,
+      "step": 3487
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0700510521091789e-05,
+      "loss": 0.6963,
+      "step": 3488
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0695939451440951e-05,
+      "loss": 0.6569,
+      "step": 3489
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0691368235663416e-05,
+      "loss": 0.6961,
+      "step": 3490
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0686796874718998e-05,
+      "loss": 0.6215,
+      "step": 3491
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0682225369567554e-05,
+      "loss": 0.6789,
+      "step": 3492
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.067765372116896e-05,
+      "loss": 0.7249,
+      "step": 3493
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0673081930483129e-05,
+      "loss": 0.7115,
+      "step": 3494
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.066850999847e-05,
+      "loss": 0.7628,
+      "step": 3495
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0663937926089547e-05,
+      "loss": 0.6831,
+      "step": 3496
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0659365714301766e-05,
+      "loss": 0.6784,
+      "step": 3497
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.065479336406669e-05,
+      "loss": 0.6643,
+      "step": 3498
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0650220876344373e-05,
+      "loss": 0.6109,
+      "step": 3499
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0645648252094904e-05,
+      "loss": 0.721,
+      "step": 3500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.06410754922784e-05,
+      "loss": 0.5947,
+      "step": 3501
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0636502597855007e-05,
+      "loss": 0.724,
+      "step": 3502
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0631929569784893e-05,
+      "loss": 0.7253,
+      "step": 3503
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0627356409028264e-05,
+      "loss": 0.6975,
+      "step": 3504
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0622783116545343e-05,
+      "loss": 0.6903,
+      "step": 3505
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0618209693296388e-05,
+      "loss": 0.7463,
+      "step": 3506
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0613636140241683e-05,
+      "loss": 0.687,
+      "step": 3507
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0609062458341541e-05,
+      "loss": 0.7175,
+      "step": 3508
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0604488648556298e-05,
+      "loss": 0.5986,
+      "step": 3509
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0599914711846318e-05,
+      "loss": 0.6789,
+      "step": 3510
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0595340649171993e-05,
+      "loss": 0.6708,
+      "step": 3511
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.059076646149374e-05,
+      "loss": 0.7319,
+      "step": 3512
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0586192149772003e-05,
+      "loss": 0.6952,
+      "step": 3513
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0581617714967257e-05,
+      "loss": 0.6701,
+      "step": 3514
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0577043158039992e-05,
+      "loss": 0.7187,
+      "step": 3515
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0572468479950733e-05,
+      "loss": 0.6829,
+      "step": 3516
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0567893681660024e-05,
+      "loss": 0.7183,
+      "step": 3517
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.056331876412844e-05,
+      "loss": 0.5867,
+      "step": 3518
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0558743728316579e-05,
+      "loss": 0.6893,
+      "step": 3519
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0554168575185062e-05,
+      "loss": 0.7044,
+      "step": 3520
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0549593305694533e-05,
+      "loss": 0.7102,
+      "step": 3521
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.054501792080567e-05,
+      "loss": 0.7507,
+      "step": 3522
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.054044242147916e-05,
+      "loss": 0.7453,
+      "step": 3523
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.053586680867573e-05,
+      "loss": 0.5641,
+      "step": 3524
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0531291083356118e-05,
+      "loss": 0.7006,
+      "step": 3525
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0526715246481097e-05,
+      "loss": 0.7059,
+      "step": 3526
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0522139299011452e-05,
+      "loss": 0.6144,
+      "step": 3527
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0517563241907998e-05,
+      "loss": 0.678,
+      "step": 3528
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.051298707613157e-05,
+      "loss": 0.7627,
+      "step": 3529
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0508410802643032e-05,
+      "loss": 0.6718,
+      "step": 3530
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0503834422403263e-05,
+      "loss": 0.6724,
+      "step": 3531
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0499257936373167e-05,
+      "loss": 0.6964,
+      "step": 3532
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0494681345513673e-05,
+      "loss": 0.7174,
+      "step": 3533
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.049010465078573e-05,
+      "loss": 0.7077,
+      "step": 3534
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0485527853150304e-05,
+      "loss": 0.6116,
+      "step": 3535
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0480950953568389e-05,
+      "loss": 0.6759,
+      "step": 3536
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0476373953001002e-05,
+      "loss": 0.7074,
+      "step": 3537
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0471796852409175e-05,
+      "loss": 0.739,
+      "step": 3538
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0467219652753964e-05,
+      "loss": 0.7348,
+      "step": 3539
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0462642354996447e-05,
+      "loss": 0.6732,
+      "step": 3540
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0458064960097716e-05,
+      "loss": 0.6148,
+      "step": 3541
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0453487469018894e-05,
+      "loss": 0.7092,
+      "step": 3542
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0448909882721119e-05,
+      "loss": 0.6991,
+      "step": 3543
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0444332202165543e-05,
+      "loss": 0.6074,
+      "step": 3544
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.043975442831335e-05,
+      "loss": 0.6996,
+      "step": 3545
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0435176562125735e-05,
+      "loss": 0.7401,
+      "step": 3546
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0430598604563912e-05,
+      "loss": 0.7145,
+      "step": 3547
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0426020556589118e-05,
+      "loss": 0.676,
+      "step": 3548
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.042144241916261e-05,
+      "loss": 0.7221,
+      "step": 3549
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.041686419324566e-05,
+      "loss": 0.6902,
+      "step": 3550
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0412285879799557e-05,
+      "loss": 0.7017,
+      "step": 3551
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0407707479785615e-05,
+      "loss": 0.7109,
+      "step": 3552
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0403128994165164e-05,
+      "loss": 0.6946,
+      "step": 3553
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0398550423899547e-05,
+      "loss": 0.5931,
+      "step": 3554
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0393971769950128e-05,
+      "loss": 0.7517,
+      "step": 3555
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0389393033278291e-05,
+      "loss": 0.6932,
+      "step": 3556
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0384814214845436e-05,
+      "loss": 0.6721,
+      "step": 3557
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0380235315612977e-05,
+      "loss": 0.7021,
+      "step": 3558
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0375656336542347e-05,
+      "loss": 0.7122,
+      "step": 3559
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0371077278595001e-05,
+      "loss": 0.6896,
+      "step": 3560
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0366498142732401e-05,
+      "loss": 0.7102,
+      "step": 3561
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.036191892991603e-05,
+      "loss": 0.7204,
+      "step": 3562
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0357339641107391e-05,
+      "loss": 0.6898,
+      "step": 3563
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0352760277267996e-05,
+      "loss": 0.7188,
+      "step": 3564
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0348180839359377e-05,
+      "loss": 0.6938,
+      "step": 3565
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.034360132834308e-05,
+      "loss": 0.7054,
+      "step": 3566
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0339021745180669e-05,
+      "loss": 0.6202,
+      "step": 3567
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0334442090833718e-05,
+      "loss": 0.688,
+      "step": 3568
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.032986236626382e-05,
+      "loss": 0.7038,
+      "step": 3569
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0325282572432582e-05,
+      "loss": 0.7062,
+      "step": 3570
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0320702710301624e-05,
+      "loss": 0.6594,
+      "step": 3571
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0316122780832585e-05,
+      "loss": 0.7248,
+      "step": 3572
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0311542784987112e-05,
+      "loss": 0.6088,
+      "step": 3573
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0306962723726867e-05,
+      "loss": 0.7008,
+      "step": 3574
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0302382598013532e-05,
+      "loss": 0.6968,
+      "step": 3575
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0297802408808796e-05,
+      "loss": 0.5297,
+      "step": 3576
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.029322215707436e-05,
+      "loss": 0.6791,
+      "step": 3577
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0288641843771945e-05,
+      "loss": 0.7135,
+      "step": 3578
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0284061469863283e-05,
+      "loss": 0.6007,
+      "step": 3579
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.027948103631011e-05,
+      "loss": 0.7349,
+      "step": 3580
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0274900544074189e-05,
+      "loss": 0.7097,
+      "step": 3581
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0270319994117284e-05,
+      "loss": 0.7032,
+      "step": 3582
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0265739387401176e-05,
+      "loss": 0.6895,
+      "step": 3583
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0261158724887657e-05,
+      "loss": 0.6946,
+      "step": 3584
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0256578007538531e-05,
+      "loss": 0.7034,
+      "step": 3585
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0251997236315612e-05,
+      "loss": 0.6966,
+      "step": 3586
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0247416412180727e-05,
+      "loss": 0.7208,
+      "step": 3587
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0242835536095714e-05,
+      "loss": 0.594,
+      "step": 3588
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0238254609022419e-05,
+      "loss": 0.6971,
+      "step": 3589
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0233673631922707e-05,
+      "loss": 0.696,
+      "step": 3590
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0229092605758441e-05,
+      "loss": 0.6924,
+      "step": 3591
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0224511531491507e-05,
+      "loss": 0.7041,
+      "step": 3592
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0219930410083789e-05,
+      "loss": 0.7142,
+      "step": 3593
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0215349242497192e-05,
+      "loss": 0.7374,
+      "step": 3594
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0210768029693622e-05,
+      "loss": 0.7163,
+      "step": 3595
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0206186772635003e-05,
+      "loss": 0.6863,
+      "step": 3596
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0201605472283258e-05,
+      "loss": 0.7057,
+      "step": 3597
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0197024129600328e-05,
+      "loss": 0.6726,
+      "step": 3598
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0192442745548159e-05,
+      "loss": 0.7,
+      "step": 3599
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0187861321088706e-05,
+      "loss": 0.666,
+      "step": 3600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.018327985718393e-05,
+      "loss": 0.6888,
+      "step": 3601
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0178698354795805e-05,
+      "loss": 0.6839,
+      "step": 3602
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0174116814886312e-05,
+      "loss": 0.6859,
+      "step": 3603
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.016953523841744e-05,
+      "loss": 0.7111,
+      "step": 3604
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0164953626351181e-05,
+      "loss": 0.7243,
+      "step": 3605
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0160371979649539e-05,
+      "loss": 0.6604,
+      "step": 3606
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0155790299274526e-05,
+      "loss": 0.7059,
+      "step": 3607
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0151208586188157e-05,
+      "loss": 0.7138,
+      "step": 3608
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0146626841352458e-05,
+      "loss": 0.5973,
+      "step": 3609
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.014204506572946e-05,
+      "loss": 0.7184,
+      "step": 3610
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0137463260281198e-05,
+      "loss": 0.678,
+      "step": 3611
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0132881425969719e-05,
+      "loss": 0.7225,
+      "step": 3612
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0128299563757069e-05,
+      "loss": 0.6684,
+      "step": 3613
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0123717674605307e-05,
+      "loss": 0.6996,
+      "step": 3614
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0119135759476493e-05,
+      "loss": 0.6398,
+      "step": 3615
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0114553819332693e-05,
+      "loss": 0.6433,
+      "step": 3616
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0109971855135978e-05,
+      "loss": 0.659,
+      "step": 3617
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0105389867848428e-05,
+      "loss": 0.5429,
+      "step": 3618
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0100807858432123e-05,
+      "loss": 0.7126,
+      "step": 3619
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0096225827849149e-05,
+      "loss": 0.7516,
+      "step": 3620
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0091643777061598e-05,
+      "loss": 0.7283,
+      "step": 3621
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0087061707031563e-05,
+      "loss": 0.5586,
+      "step": 3622
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0082479618721144e-05,
+      "loss": 0.7537,
+      "step": 3623
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0077897513092448e-05,
+      "loss": 0.7178,
+      "step": 3624
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0073315391107575e-05,
+      "loss": 0.683,
+      "step": 3625
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0068733253728639e-05,
+      "loss": 0.7066,
+      "step": 3626
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0064151101917754e-05,
+      "loss": 0.7255,
+      "step": 3627
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0059568936637034e-05,
+      "loss": 0.7131,
+      "step": 3628
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0054986758848602e-05,
+      "loss": 0.7001,
+      "step": 3629
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0050404569514573e-05,
+      "loss": 0.6791,
+      "step": 3630
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0045822369597078e-05,
+      "loss": 0.5552,
+      "step": 3631
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0041240160058242e-05,
+      "loss": 0.68,
+      "step": 3632
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0036657941860189e-05,
+      "loss": 0.6713,
+      "step": 3633
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0032075715965053e-05,
+      "loss": 0.6451,
+      "step": 3634
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0027493483334966e-05,
+      "loss": 0.6552,
+      "step": 3635
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.002291124493206e-05,
+      "loss": 0.7056,
+      "step": 3636
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0018329001718472e-05,
+      "loss": 0.7174,
+      "step": 3637
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0013746754656335e-05,
+      "loss": 0.7535,
+      "step": 3638
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0009164504707787e-05,
+      "loss": 0.5838,
+      "step": 3639
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0004582252834962e-05,
+      "loss": 0.6844,
+      "step": 3640
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1e-05,
+      "loss": 0.674,
+      "step": 3641
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.995417747165041e-06,
+      "loss": 0.645,
+      "step": 3642
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.990835495292216e-06,
+      "loss": 0.5952,
+      "step": 3643
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.986253245343665e-06,
+      "loss": 0.6885,
+      "step": 3644
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.981670998281533e-06,
+      "loss": 0.7361,
+      "step": 3645
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.977088755067941e-06,
+      "loss": 0.6852,
+      "step": 3646
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.972506516665035e-06,
+      "loss": 0.6612,
+      "step": 3647
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.96792428403495e-06,
+      "loss": 0.5815,
+      "step": 3648
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.963342058139815e-06,
+      "loss": 0.6624,
+      "step": 3649
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.958759839941763e-06,
+      "loss": 0.7171,
+      "step": 3650
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.954177630402923e-06,
+      "loss": 0.6057,
+      "step": 3651
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.949595430485432e-06,
+      "loss": 0.6596,
+      "step": 3652
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.945013241151404e-06,
+      "loss": 0.5529,
+      "step": 3653
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.940431063362967e-06,
+      "loss": 0.6425,
+      "step": 3654
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.93584889808225e-06,
+      "loss": 0.7515,
+      "step": 3655
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.931266746271363e-06,
+      "loss": 0.7505,
+      "step": 3656
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.926684608892428e-06,
+      "loss": 0.5793,
+      "step": 3657
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.922102486907554e-06,
+      "loss": 0.5814,
+      "step": 3658
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.917520381278862e-06,
+      "loss": 0.6648,
+      "step": 3659
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.912938292968444e-06,
+      "loss": 0.6813,
+      "step": 3660
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.908356222938407e-06,
+      "loss": 0.5807,
+      "step": 3661
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.903774172150855e-06,
+      "loss": 0.7257,
+      "step": 3662
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.89919214156788e-06,
+      "loss": 0.5422,
+      "step": 3663
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.894610132151576e-06,
+      "loss": 0.5865,
+      "step": 3664
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.890028144864022e-06,
+      "loss": 0.6109,
+      "step": 3665
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.885446180667313e-06,
+      "loss": 0.6622,
+      "step": 3666
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.880864240523512e-06,
+      "loss": 0.6821,
+      "step": 3667
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.876282325394696e-06,
+      "loss": 0.7383,
+      "step": 3668
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.871700436242934e-06,
+      "loss": 0.7279,
+      "step": 3669
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.867118574030285e-06,
+      "loss": 0.7149,
+      "step": 3670
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.862536739718805e-06,
+      "loss": 0.7254,
+      "step": 3671
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.857954934270545e-06,
+      "loss": 0.6977,
+      "step": 3672
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.853373158647544e-06,
+      "loss": 0.6258,
+      "step": 3673
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.848791413811848e-06,
+      "loss": 0.6629,
+      "step": 3674
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.844209700725478e-06,
+      "loss": 0.7038,
+      "step": 3675
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.839628020350464e-06,
+      "loss": 0.7234,
+      "step": 3676
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.835046373648822e-06,
+      "loss": 0.7264,
+      "step": 3677
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.830464761582563e-06,
+      "loss": 0.7046,
+      "step": 3678
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.82588318511369e-06,
+      "loss": 0.6776,
+      "step": 3679
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.821301645204194e-06,
+      "loss": 0.7101,
+      "step": 3680
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.816720142816076e-06,
+      "loss": 0.5974,
+      "step": 3681
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.812138678911298e-06,
+      "loss": 0.6956,
+      "step": 3682
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.807557254451843e-06,
+      "loss": 0.732,
+      "step": 3683
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.802975870399675e-06,
+      "loss": 0.7047,
+      "step": 3684
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.798394527716743e-06,
+      "loss": 0.7018,
+      "step": 3685
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.793813227365e-06,
+      "loss": 0.6816,
+      "step": 3686
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.789231970306376e-06,
+      "loss": 0.7142,
+      "step": 3687
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.784650757502813e-06,
+      "loss": 0.6961,
+      "step": 3688
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.780069589916213e-06,
+      "loss": 0.7,
+      "step": 3689
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.775488468508496e-06,
+      "loss": 0.6967,
+      "step": 3690
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.77090739424156e-06,
+      "loss": 0.6539,
+      "step": 3691
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.766326368077296e-06,
+      "loss": 0.6847,
+      "step": 3692
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.761745390977583e-06,
+      "loss": 0.7205,
+      "step": 3693
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.757164463904287e-06,
+      "loss": 0.6894,
+      "step": 3694
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.752583587819277e-06,
+      "loss": 0.6876,
+      "step": 3695
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.748002763684393e-06,
+      "loss": 0.7186,
+      "step": 3696
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.743421992461472e-06,
+      "loss": 0.6985,
+      "step": 3697
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.738841275112345e-06,
+      "loss": 0.7079,
+      "step": 3698
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.734260612598825e-06,
+      "loss": 0.7063,
+      "step": 3699
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.72968000588272e-06,
+      "loss": 0.7005,
+      "step": 3700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.725099455925811e-06,
+      "loss": 0.7083,
+      "step": 3701
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.720518963689895e-06,
+      "loss": 0.6026,
+      "step": 3702
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.715938530136724e-06,
+      "loss": 0.7348,
+      "step": 3703
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.711358156228056e-06,
+      "loss": 0.7141,
+      "step": 3704
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.706777842925643e-06,
+      "loss": 0.7146,
+      "step": 3705
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.702197591191208e-06,
+      "loss": 0.7407,
+      "step": 3706
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.697617401986472e-06,
+      "loss": 0.541,
+      "step": 3707
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.693037276273133e-06,
+      "loss": 0.7032,
+      "step": 3708
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.688457215012894e-06,
+      "loss": 0.7019,
+      "step": 3709
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.68387721916742e-06,
+      "loss": 0.6952,
+      "step": 3710
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.679297289698379e-06,
+      "loss": 0.6803,
+      "step": 3711
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.674717427567421e-06,
+      "loss": 0.6398,
+      "step": 3712
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.670137633736182e-06,
+      "loss": 0.6906,
+      "step": 3713
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.665557909166285e-06,
+      "loss": 0.6304,
+      "step": 3714
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.660978254819335e-06,
+      "loss": 0.7162,
+      "step": 3715
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.656398671656924e-06,
+      "loss": 0.5774,
+      "step": 3716
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.651819160640627e-06,
+      "loss": 0.6791,
+      "step": 3717
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.647239722732007e-06,
+      "loss": 0.679,
+      "step": 3718
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.642660358892612e-06,
+      "loss": 0.7162,
+      "step": 3719
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.638081070083972e-06,
+      "loss": 0.6027,
+      "step": 3720
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.633501857267602e-06,
+      "loss": 0.6783,
+      "step": 3721
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.628922721405002e-06,
+      "loss": 0.7185,
+      "step": 3722
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.624343663457653e-06,
+      "loss": 0.7088,
+      "step": 3723
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.619764684387028e-06,
+      "loss": 0.6948,
+      "step": 3724
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.615185785154568e-06,
+      "loss": 0.6602,
+      "step": 3725
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.610606966721712e-06,
+      "loss": 0.7245,
+      "step": 3726
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.606028230049875e-06,
+      "loss": 0.7222,
+      "step": 3727
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.601449576100457e-06,
+      "loss": 0.5711,
+      "step": 3728
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.59687100583484e-06,
+      "loss": 0.7326,
+      "step": 3729
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.592292520214384e-06,
+      "loss": 0.5825,
+      "step": 3730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.587714120200448e-06,
+      "loss": 0.6982,
+      "step": 3731
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.583135806754346e-06,
+      "loss": 0.7137,
+      "step": 3732
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.578557580837393e-06,
+      "loss": 0.7344,
+      "step": 3733
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.573979443410883e-06,
+      "loss": 0.7034,
+      "step": 3734
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.569401395436091e-06,
+      "loss": 0.6637,
+      "step": 3735
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.564823437874268e-06,
+      "loss": 0.766,
+      "step": 3736
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.560245571686651e-06,
+      "loss": 0.692,
+      "step": 3737
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.55566779783446e-06,
+      "loss": 0.6813,
+      "step": 3738
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.551090117278888e-06,
+      "loss": 0.7184,
+      "step": 3739
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.546512530981108e-06,
+      "loss": 0.6997,
+      "step": 3740
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.541935039902287e-06,
+      "loss": 0.5986,
+      "step": 3741
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.537357645003558e-06,
+      "loss": 0.6883,
+      "step": 3742
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.532780347246039e-06,
+      "loss": 0.6992,
+      "step": 3743
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.528203147590825e-06,
+      "loss": 0.7162,
+      "step": 3744
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.523626046999003e-06,
+      "loss": 0.6142,
+      "step": 3745
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.519049046431616e-06,
+      "loss": 0.7103,
+      "step": 3746
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.5144721468497e-06,
+      "loss": 0.6246,
+      "step": 3747
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.509895349214274e-06,
+      "loss": 0.6401,
+      "step": 3748
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.50531865448633e-06,
+      "loss": 0.603,
+      "step": 3749
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.500742063626835e-06,
+      "loss": 0.6976,
+      "step": 3750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.496165577596737e-06,
+      "loss": 0.6745,
+      "step": 3751
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.491589197356973e-06,
+      "loss": 0.6863,
+      "step": 3752
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.487012923868434e-06,
+      "loss": 0.6688,
+      "step": 3753
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.482436758092007e-06,
+      "loss": 0.6732,
+      "step": 3754
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.477860700988551e-06,
+      "loss": 0.7574,
+      "step": 3755
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.473284753518906e-06,
+      "loss": 0.6785,
+      "step": 3756
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.468708916643883e-06,
+      "loss": 0.6962,
+      "step": 3757
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.464133191324273e-06,
+      "loss": 0.7377,
+      "step": 3758
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.459557578520844e-06,
+      "loss": 0.6975,
+      "step": 3759
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.454982079194337e-06,
+      "loss": 0.7134,
+      "step": 3760
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.45040669430547e-06,
+      "loss": 0.6769,
+      "step": 3761
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.445831424814941e-06,
+      "loss": 0.6548,
+      "step": 3762
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.441256271683424e-06,
+      "loss": 0.5983,
+      "step": 3763
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.436681235871563e-06,
+      "loss": 0.6567,
+      "step": 3764
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.432106318339978e-06,
+      "loss": 0.6887,
+      "step": 3765
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.427531520049274e-06,
+      "loss": 0.6752,
+      "step": 3766
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.422956841960013e-06,
+      "loss": 0.6971,
+      "step": 3767
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.418382285032747e-06,
+      "loss": 0.7318,
+      "step": 3768
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.413807850227998e-06,
+      "loss": 0.6743,
+      "step": 3769
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.409233538506263e-06,
+      "loss": 0.6778,
+      "step": 3770
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.404659350828012e-06,
+      "loss": 0.6949,
+      "step": 3771
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.400085288153686e-06,
+      "loss": 0.6892,
+      "step": 3772
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.395511351443704e-06,
+      "loss": 0.6932,
+      "step": 3773
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.390937541658464e-06,
+      "loss": 0.5568,
+      "step": 3774
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.38636385975832e-06,
+      "loss": 0.6914,
+      "step": 3775
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.381790306703615e-06,
+      "loss": 0.6974,
+      "step": 3776
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.37721688345466e-06,
+      "loss": 0.6975,
+      "step": 3777
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.37264359097174e-06,
+      "loss": 0.6916,
+      "step": 3778
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.368070430215109e-06,
+      "loss": 0.7306,
+      "step": 3779
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.363497402144993e-06,
+      "loss": 0.6976,
+      "step": 3780
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.358924507721603e-06,
+      "loss": 0.683,
+      "step": 3781
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.3543517479051e-06,
+      "loss": 0.7185,
+      "step": 3782
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.349779123655632e-06,
+      "loss": 0.6658,
+      "step": 3783
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.345206635933315e-06,
+      "loss": 0.6783,
+      "step": 3784
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.340634285698238e-06,
+      "loss": 0.7194,
+      "step": 3785
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.336062073910456e-06,
+      "loss": 0.7042,
+      "step": 3786
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.33149000153e-06,
+      "loss": 0.7003,
+      "step": 3787
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.326918069516876e-06,
+      "loss": 0.5773,
+      "step": 3788
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.322346278831045e-06,
+      "loss": 0.5612,
+      "step": 3789
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.31777463043245e-06,
+      "loss": 0.6852,
+      "step": 3790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.313203125281003e-06,
+      "loss": 0.7369,
+      "step": 3791
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.308631764336586e-06,
+      "loss": 0.6941,
+      "step": 3792
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.30406054855905e-06,
+      "loss": 0.6553,
+      "step": 3793
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.299489478908213e-06,
+      "loss": 0.5708,
+      "step": 3794
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.294918556343868e-06,
+      "loss": 0.5915,
+      "step": 3795
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.29034778182577e-06,
+      "loss": 0.7107,
+      "step": 3796
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.285777156313643e-06,
+      "loss": 0.5836,
+      "step": 3797
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.281206680767191e-06,
+      "loss": 0.6669,
+      "step": 3798
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.276636356146074e-06,
+      "loss": 0.604,
+      "step": 3799
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.272066183409927e-06,
+      "loss": 0.7181,
+      "step": 3800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.26749616351835e-06,
+      "loss": 0.6798,
+      "step": 3801
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.262926297430913e-06,
+      "loss": 0.6969,
+      "step": 3802
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.258356586107152e-06,
+      "loss": 0.7024,
+      "step": 3803
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.25378703050657e-06,
+      "loss": 0.7033,
+      "step": 3804
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.24921763158864e-06,
+      "loss": 0.6246,
+      "step": 3805
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.2446483903128e-06,
+      "loss": 0.7214,
+      "step": 3806
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.240079307638456e-06,
+      "loss": 0.6901,
+      "step": 3807
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.23551038452498e-06,
+      "loss": 0.7084,
+      "step": 3808
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.230941621931713e-06,
+      "loss": 0.5951,
+      "step": 3809
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.226373020817955e-06,
+      "loss": 0.6863,
+      "step": 3810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.221804582142981e-06,
+      "loss": 0.651,
+      "step": 3811
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.217236306866026e-06,
+      "loss": 0.6852,
+      "step": 3812
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.212668195946293e-06,
+      "loss": 0.6996,
+      "step": 3813
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.208100250342952e-06,
+      "loss": 0.5961,
+      "step": 3814
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.203532471015135e-06,
+      "loss": 0.5714,
+      "step": 3815
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.198964858921943e-06,
+      "loss": 0.7072,
+      "step": 3816
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.194397415022435e-06,
+      "loss": 0.6693,
+      "step": 3817
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.189830140275642e-06,
+      "loss": 0.6407,
+      "step": 3818
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.185263035640553e-06,
+      "loss": 0.6902,
+      "step": 3819
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.18069610207613e-06,
+      "loss": 0.6891,
+      "step": 3820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.176129340541292e-06,
+      "loss": 0.6552,
+      "step": 3821
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.171562751994922e-06,
+      "loss": 0.5727,
+      "step": 3822
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.166996337395869e-06,
+      "loss": 0.7236,
+      "step": 3823
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.162430097702954e-06,
+      "loss": 0.6996,
+      "step": 3824
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.157864033874939e-06,
+      "loss": 0.7171,
+      "step": 3825
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.153298146870565e-06,
+      "loss": 0.7182,
+      "step": 3826
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.14873243764854e-06,
+      "loss": 0.6869,
+      "step": 3827
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.144166907167521e-06,
+      "loss": 0.694,
+      "step": 3828
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.139601556386138e-06,
+      "loss": 0.6967,
+      "step": 3829
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.135036386262977e-06,
+      "loss": 0.7184,
+      "step": 3830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.130471397756598e-06,
+      "loss": 0.69,
+      "step": 3831
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.125906591825501e-06,
+      "loss": 0.6498,
+      "step": 3832
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.121341969428164e-06,
+      "loss": 0.64,
+      "step": 3833
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.116777531523025e-06,
+      "loss": 0.7025,
+      "step": 3834
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.11221327906848e-06,
+      "loss": 0.6135,
+      "step": 3835
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.107649213022887e-06,
+      "loss": 0.7069,
+      "step": 3836
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.103085334344565e-06,
+      "loss": 0.7619,
+      "step": 3837
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.098521643991796e-06,
+      "loss": 0.7174,
+      "step": 3838
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.093958142922813e-06,
+      "loss": 0.7217,
+      "step": 3839
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.089394832095821e-06,
+      "loss": 0.689,
+      "step": 3840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.084831712468981e-06,
+      "loss": 0.6885,
+      "step": 3841
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.080268785000414e-06,
+      "loss": 0.6907,
+      "step": 3842
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.075706050648196e-06,
+      "loss": 0.7217,
+      "step": 3843
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.071143510370368e-06,
+      "loss": 0.7156,
+      "step": 3844
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.06658116512493e-06,
+      "loss": 0.5886,
+      "step": 3845
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.062019015869837e-06,
+      "loss": 0.7032,
+      "step": 3846
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.057457063563006e-06,
+      "loss": 0.6533,
+      "step": 3847
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.052895309162314e-06,
+      "loss": 0.7302,
+      "step": 3848
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.04833375362559e-06,
+      "loss": 0.7316,
+      "step": 3849
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.043772397910632e-06,
+      "loss": 0.7035,
+      "step": 3850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.039211242975185e-06,
+      "loss": 0.584,
+      "step": 3851
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.034650289776961e-06,
+      "loss": 0.6918,
+      "step": 3852
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.030089539273621e-06,
+      "loss": 0.6908,
+      "step": 3853
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.025528992422788e-06,
+      "loss": 0.7121,
+      "step": 3854
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.020968650182043e-06,
+      "loss": 0.7571,
+      "step": 3855
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.016408513508922e-06,
+      "loss": 0.7101,
+      "step": 3856
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.011848583360922e-06,
+      "loss": 0.7131,
+      "step": 3857
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.007288860695492e-06,
+      "loss": 0.6968,
+      "step": 3858
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.00272934647004e-06,
+      "loss": 0.7255,
+      "step": 3859
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.998170041641924e-06,
+      "loss": 0.7174,
+      "step": 3860
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.99361094716847e-06,
+      "loss": 0.6987,
+      "step": 3861
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.989052064006945e-06,
+      "loss": 0.6205,
+      "step": 3862
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.984493393114587e-06,
+      "loss": 0.6598,
+      "step": 3863
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.979934935448577e-06,
+      "loss": 0.5911,
+      "step": 3864
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.97537669196606e-06,
+      "loss": 0.5885,
+      "step": 3865
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.970818663624134e-06,
+      "loss": 0.6842,
+      "step": 3866
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.966260851379844e-06,
+      "loss": 0.6898,
+      "step": 3867
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.961703256190198e-06,
+      "loss": 0.6852,
+      "step": 3868
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.957145879012155e-06,
+      "loss": 0.6783,
+      "step": 3869
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.952588720802636e-06,
+      "loss": 0.6682,
+      "step": 3870
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.948031782518501e-06,
+      "loss": 0.7027,
+      "step": 3871
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.943475065116578e-06,
+      "loss": 0.6843,
+      "step": 3872
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.93891856955364e-06,
+      "loss": 0.6999,
+      "step": 3873
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.93436229678642e-06,
+      "loss": 0.6363,
+      "step": 3874
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.929806247771594e-06,
+      "loss": 0.6869,
+      "step": 3875
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.925250423465801e-06,
+      "loss": 0.6999,
+      "step": 3876
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.920694824825632e-06,
+      "loss": 0.7133,
+      "step": 3877
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.916139452807624e-06,
+      "loss": 0.6443,
+      "step": 3878
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.911584308368273e-06,
+      "loss": 0.6959,
+      "step": 3879
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.907029392464023e-06,
+      "loss": 0.6638,
+      "step": 3880
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.902474706051275e-06,
+      "loss": 0.6903,
+      "step": 3881
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.897920250086372e-06,
+      "loss": 0.7397,
+      "step": 3882
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.893366025525616e-06,
+      "loss": 0.6968,
+      "step": 3883
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.888812033325264e-06,
+      "loss": 0.6727,
+      "step": 3884
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.88425827444152e-06,
+      "loss": 0.7251,
+      "step": 3885
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.879704749830536e-06,
+      "loss": 0.6933,
+      "step": 3886
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.875151460448417e-06,
+      "loss": 0.7043,
+      "step": 3887
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.87059840725122e-06,
+      "loss": 0.7327,
+      "step": 3888
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.866045591194952e-06,
+      "loss": 0.7151,
+      "step": 3889
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.86149301323557e-06,
+      "loss": 0.659,
+      "step": 3890
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.856940674328975e-06,
+      "loss": 0.6939,
+      "step": 3891
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.852388575431032e-06,
+      "loss": 0.6584,
+      "step": 3892
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.847836717497544e-06,
+      "loss": 0.6965,
+      "step": 3893
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.843285101484265e-06,
+      "loss": 0.699,
+      "step": 3894
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.838733728346902e-06,
+      "loss": 0.6658,
+      "step": 3895
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.834182599041107e-06,
+      "loss": 0.6852,
+      "step": 3896
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.829631714522482e-06,
+      "loss": 0.7087,
+      "step": 3897
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.825081075746578e-06,
+      "loss": 0.6908,
+      "step": 3898
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.820530683668898e-06,
+      "loss": 0.6982,
+      "step": 3899
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.815980539244886e-06,
+      "loss": 0.6905,
+      "step": 3900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.811430643429939e-06,
+      "loss": 0.6711,
+      "step": 3901
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.806880997179404e-06,
+      "loss": 0.7224,
+      "step": 3902
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.802331601448567e-06,
+      "loss": 0.6809,
+      "step": 3903
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.797782457192667e-06,
+      "loss": 0.7232,
+      "step": 3904
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.79323356536689e-06,
+      "loss": 0.7175,
+      "step": 3905
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.788684926926373e-06,
+      "loss": 0.6987,
+      "step": 3906
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.784136542826192e-06,
+      "loss": 0.7125,
+      "step": 3907
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.779588414021372e-06,
+      "loss": 0.6165,
+      "step": 3908
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.77504054146689e-06,
+      "loss": 0.7284,
+      "step": 3909
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.770492926117659e-06,
+      "loss": 0.7355,
+      "step": 3910
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.765945568928548e-06,
+      "loss": 0.7421,
+      "step": 3911
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.761398470854363e-06,
+      "loss": 0.6594,
+      "step": 3912
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.756851632849865e-06,
+      "loss": 0.6877,
+      "step": 3913
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.752305055869754e-06,
+      "loss": 0.7377,
+      "step": 3914
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.747758740868677e-06,
+      "loss": 0.6,
+      "step": 3915
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.743212688801225e-06,
+      "loss": 0.7048,
+      "step": 3916
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.738666900621935e-06,
+      "loss": 0.5394,
+      "step": 3917
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.734121377285286e-06,
+      "loss": 0.7174,
+      "step": 3918
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.729576119745701e-06,
+      "loss": 0.6881,
+      "step": 3919
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.725031128957558e-06,
+      "loss": 0.7105,
+      "step": 3920
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.720486405875167e-06,
+      "loss": 0.6821,
+      "step": 3921
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.715941951452782e-06,
+      "loss": 0.5685,
+      "step": 3922
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.711397766644609e-06,
+      "loss": 0.6735,
+      "step": 3923
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.70685385240479e-06,
+      "loss": 0.6134,
+      "step": 3924
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.702310209687413e-06,
+      "loss": 0.675,
+      "step": 3925
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.697766839446504e-06,
+      "loss": 0.6581,
+      "step": 3926
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.693223742636043e-06,
+      "loss": 0.6913,
+      "step": 3927
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.688680920209944e-06,
+      "loss": 0.5865,
+      "step": 3928
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.684138373122065e-06,
+      "loss": 0.6877,
+      "step": 3929
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.679596102326206e-06,
+      "loss": 0.687,
+      "step": 3930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.675054108776112e-06,
+      "loss": 0.6954,
+      "step": 3931
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.670512393425463e-06,
+      "loss": 0.7335,
+      "step": 3932
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.665970957227888e-06,
+      "loss": 0.6355,
+      "step": 3933
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.66142980113695e-06,
+      "loss": 0.711,
+      "step": 3934
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.656888926106164e-06,
+      "loss": 0.7122,
+      "step": 3935
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.652348333088974e-06,
+      "loss": 0.6895,
+      "step": 3936
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.647808023038775e-06,
+      "loss": 0.6575,
+      "step": 3937
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.643267996908895e-06,
+      "loss": 0.5932,
+      "step": 3938
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.638728255652606e-06,
+      "loss": 0.7129,
+      "step": 3939
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.634188800223118e-06,
+      "loss": 0.689,
+      "step": 3940
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.629649631573582e-06,
+      "loss": 0.7053,
+      "step": 3941
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.62511075065709e-06,
+      "loss": 0.7121,
+      "step": 3942
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.620572158426674e-06,
+      "loss": 0.6933,
+      "step": 3943
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.616033855835304e-06,
+      "loss": 0.6905,
+      "step": 3944
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.611495843835891e-06,
+      "loss": 0.6947,
+      "step": 3945
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.606958123381278e-06,
+      "loss": 0.6822,
+      "step": 3946
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.602420695424257e-06,
+      "loss": 0.7003,
+      "step": 3947
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.597883560917547e-06,
+      "loss": 0.685,
+      "step": 3948
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.593346720813819e-06,
+      "loss": 0.6919,
+      "step": 3949
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.588810176065673e-06,
+      "loss": 0.6768,
+      "step": 3950
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.58427392762565e-06,
+      "loss": 0.6999,
+      "step": 3951
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.579737976446229e-06,
+      "loss": 0.7126,
+      "step": 3952
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.57520232347982e-06,
+      "loss": 0.6732,
+      "step": 3953
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.57066696967878e-06,
+      "loss": 0.7276,
+      "step": 3954
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.566131915995397e-06,
+      "loss": 0.6725,
+      "step": 3955
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.561597163381905e-06,
+      "loss": 0.701,
+      "step": 3956
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.557062712790459e-06,
+      "loss": 0.7009,
+      "step": 3957
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.552528565173164e-06,
+      "loss": 0.6585,
+      "step": 3958
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.54799472148206e-06,
+      "loss": 0.7034,
+      "step": 3959
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.543461182669113e-06,
+      "loss": 0.727,
+      "step": 3960
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.538927949686235e-06,
+      "loss": 0.6738,
+      "step": 3961
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.53439502348527e-06,
+      "loss": 0.6855,
+      "step": 3962
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.529862405018001e-06,
+      "loss": 0.6929,
+      "step": 3963
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.525330095236143e-06,
+      "loss": 0.6606,
+      "step": 3964
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.520798095091347e-06,
+      "loss": 0.6968,
+      "step": 3965
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.516266405535198e-06,
+      "loss": 0.6809,
+      "step": 3966
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.511735027519217e-06,
+      "loss": 0.7065,
+      "step": 3967
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.50720396199486e-06,
+      "loss": 0.6968,
+      "step": 3968
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.502673209913513e-06,
+      "loss": 0.6598,
+      "step": 3969
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.498142772226505e-06,
+      "loss": 0.705,
+      "step": 3970
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.493612649885094e-06,
+      "loss": 0.6968,
+      "step": 3971
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.489082843840469e-06,
+      "loss": 0.6788,
+      "step": 3972
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.484553355043755e-06,
+      "loss": 0.6074,
+      "step": 3973
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.480024184446014e-06,
+      "loss": 0.7096,
+      "step": 3974
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.475495332998233e-06,
+      "loss": 0.6778,
+      "step": 3975
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.470966801651342e-06,
+      "loss": 0.7319,
+      "step": 3976
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.46643859135619e-06,
+      "loss": 0.6658,
+      "step": 3977
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.461910703063579e-06,
+      "loss": 0.6763,
+      "step": 3978
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.457383137724225e-06,
+      "loss": 0.6686,
+      "step": 3979
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.452855896288782e-06,
+      "loss": 0.707,
+      "step": 3980
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.448328979707843e-06,
+      "loss": 0.6821,
+      "step": 3981
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.44380238893192e-06,
+      "loss": 0.6908,
+      "step": 3982
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.439276124911463e-06,
+      "loss": 0.6308,
+      "step": 3983
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.434750188596854e-06,
+      "loss": 0.6924,
+      "step": 3984
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.430224580938411e-06,
+      "loss": 0.7114,
+      "step": 3985
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.425699302886375e-06,
+      "loss": 0.6934,
+      "step": 3986
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.421174355390917e-06,
+      "loss": 0.7013,
+      "step": 3987
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.41664973940215e-06,
+      "loss": 0.7089,
+      "step": 3988
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.412125455870102e-06,
+      "loss": 0.6553,
+      "step": 3989
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.40760150574474e-06,
+      "loss": 0.665,
+      "step": 3990
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.403077889975959e-06,
+      "loss": 0.6751,
+      "step": 3991
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.39855460951359e-06,
+      "loss": 0.6862,
+      "step": 3992
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.394031665307383e-06,
+      "loss": 0.7223,
+      "step": 3993
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.389509058307025e-06,
+      "loss": 0.7093,
+      "step": 3994
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.38498678946213e-06,
+      "loss": 0.7275,
+      "step": 3995
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.380464859722238e-06,
+      "loss": 0.6888,
+      "step": 3996
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.375943270036822e-06,
+      "loss": 0.731,
+      "step": 3997
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.37142202135528e-06,
+      "loss": 0.6672,
+      "step": 3998
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.366901114626945e-06,
+      "loss": 0.7023,
+      "step": 3999
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.36238055080107e-06,
+      "loss": 0.7071,
+      "step": 4000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.357860330826843e-06,
+      "loss": 0.6628,
+      "step": 4001
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.353340455653374e-06,
+      "loss": 0.6816,
+      "step": 4002
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.348820926229703e-06,
+      "loss": 0.594,
+      "step": 4003
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.344301743504798e-06,
+      "loss": 0.6851,
+      "step": 4004
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.339782908427552e-06,
+      "loss": 0.6709,
+      "step": 4005
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.335264421946789e-06,
+      "loss": 0.6713,
+      "step": 4006
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.330746285011258e-06,
+      "loss": 0.6669,
+      "step": 4007
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.326228498569634e-06,
+      "loss": 0.6651,
+      "step": 4008
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.321711063570516e-06,
+      "loss": 0.7075,
+      "step": 4009
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.317193980962434e-06,
+      "loss": 0.6737,
+      "step": 4010
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.312677251693841e-06,
+      "loss": 0.7146,
+      "step": 4011
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.308160876713117e-06,
+      "loss": 0.6895,
+      "step": 4012
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.303644856968564e-06,
+      "loss": 0.7032,
+      "step": 4013
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.299129193408417e-06,
+      "loss": 0.6919,
+      "step": 4014
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.294613886980831e-06,
+      "loss": 0.7101,
+      "step": 4015
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.290098938633889e-06,
+      "loss": 0.5636,
+      "step": 4016
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.285584349315588e-06,
+      "loss": 0.6195,
+      "step": 4017
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.281070119973866e-06,
+      "loss": 0.7181,
+      "step": 4018
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.276556251556576e-06,
+      "loss": 0.699,
+      "step": 4019
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.272042745011493e-06,
+      "loss": 0.6016,
+      "step": 4020
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.267529601286325e-06,
+      "loss": 0.6546,
+      "step": 4021
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.263016821328696e-06,
+      "loss": 0.7251,
+      "step": 4022
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.258504406086158e-06,
+      "loss": 0.6559,
+      "step": 4023
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.253992356506187e-06,
+      "loss": 0.5974,
+      "step": 4024
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.249480673536174e-06,
+      "loss": 0.6926,
+      "step": 4025
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.244969358123441e-06,
+      "loss": 0.6987,
+      "step": 4026
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.240458411215231e-06,
+      "loss": 0.6778,
+      "step": 4027
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.23594783375871e-06,
+      "loss": 0.6675,
+      "step": 4028
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.231437626700968e-06,
+      "loss": 0.6653,
+      "step": 4029
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.22692779098901e-06,
+      "loss": 0.7249,
+      "step": 4030
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.222418327569777e-06,
+      "loss": 0.729,
+      "step": 4031
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.217909237390112e-06,
+      "loss": 0.7218,
+      "step": 4032
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.213400521396796e-06,
+      "loss": 0.6364,
+      "step": 4033
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.208892180536523e-06,
+      "loss": 0.7079,
+      "step": 4034
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.204384215755917e-06,
+      "loss": 0.7256,
+      "step": 4035
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.199876628001513e-06,
+      "loss": 0.6064,
+      "step": 4036
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.195369418219773e-06,
+      "loss": 0.7236,
+      "step": 4037
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.190862587357078e-06,
+      "loss": 0.7206,
+      "step": 4038
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.186356136359726e-06,
+      "loss": 0.7302,
+      "step": 4039
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.181850066173943e-06,
+      "loss": 0.612,
+      "step": 4040
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.177344377745862e-06,
+      "loss": 0.723,
+      "step": 4041
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.172839072021554e-06,
+      "loss": 0.6954,
+      "step": 4042
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.168334149946996e-06,
+      "loss": 0.6874,
+      "step": 4043
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.163829612468088e-06,
+      "loss": 0.6795,
+      "step": 4044
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.159325460530653e-06,
+      "loss": 0.6616,
+      "step": 4045
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.154821695080425e-06,
+      "loss": 0.7048,
+      "step": 4046
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.150318317063062e-06,
+      "loss": 0.7095,
+      "step": 4047
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.14581532742414e-06,
+      "loss": 0.577,
+      "step": 4048
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.14131272710916e-06,
+      "loss": 0.6697,
+      "step": 4049
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.136810517063525e-06,
+      "loss": 0.6616,
+      "step": 4050
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.132308698232574e-06,
+      "loss": 0.6808,
+      "step": 4051
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.127807271561554e-06,
+      "loss": 0.7395,
+      "step": 4052
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.123306237995628e-06,
+      "loss": 0.6964,
+      "step": 4053
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.118805598479883e-06,
+      "loss": 0.6852,
+      "step": 4054
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.11430535395932e-06,
+      "loss": 0.6584,
+      "step": 4055
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.109805505378852e-06,
+      "loss": 0.6995,
+      "step": 4056
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.105306053683323e-06,
+      "loss": 0.6468,
+      "step": 4057
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.10080699981748e-06,
+      "loss": 0.6776,
+      "step": 4058
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.096308344725993e-06,
+      "loss": 0.7355,
+      "step": 4059
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.091810089353443e-06,
+      "loss": 0.6633,
+      "step": 4060
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.087312234644333e-06,
+      "loss": 0.6748,
+      "step": 4061
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.08281478154308e-06,
+      "loss": 0.6942,
+      "step": 4062
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.078317730994013e-06,
+      "loss": 0.6848,
+      "step": 4063
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.073821083941385e-06,
+      "loss": 0.7214,
+      "step": 4064
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.069324841329354e-06,
+      "loss": 0.6265,
+      "step": 4065
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.064829004102001e-06,
+      "loss": 0.6958,
+      "step": 4066
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.060333573203318e-06,
+      "loss": 0.6883,
+      "step": 4067
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.055838549577211e-06,
+      "loss": 0.5898,
+      "step": 4068
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.051343934167504e-06,
+      "loss": 0.6914,
+      "step": 4069
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.046849727917927e-06,
+      "loss": 0.6042,
+      "step": 4070
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.042355931772142e-06,
+      "loss": 0.7121,
+      "step": 4071
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.037862546673705e-06,
+      "loss": 0.6136,
+      "step": 4072
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.033369573566097e-06,
+      "loss": 0.7216,
+      "step": 4073
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.02887701339271e-06,
+      "loss": 0.7306,
+      "step": 4074
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.024384867096846e-06,
+      "loss": 0.7034,
+      "step": 4075
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.019893135621722e-06,
+      "loss": 0.6955,
+      "step": 4076
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.015401819910472e-06,
+      "loss": 0.7151,
+      "step": 4077
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.010910920906137e-06,
+      "loss": 0.7087,
+      "step": 4078
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.006420439551676e-06,
+      "loss": 0.7161,
+      "step": 4079
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.001930376789954e-06,
+      "loss": 0.6522,
+      "step": 4080
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.997440733563756e-06,
+      "loss": 0.6251,
+      "step": 4081
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.992951510815769e-06,
+      "loss": 0.7182,
+      "step": 4082
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.9884627094886e-06,
+      "loss": 0.6453,
+      "step": 4083
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.983974330524762e-06,
+      "loss": 0.5951,
+      "step": 4084
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.979486374866684e-06,
+      "loss": 0.6944,
+      "step": 4085
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.974998843456703e-06,
+      "loss": 0.6619,
+      "step": 4086
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.97051173723707e-06,
+      "loss": 0.6639,
+      "step": 4087
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.966025057149945e-06,
+      "loss": 0.6951,
+      "step": 4088
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.961538804137393e-06,
+      "loss": 0.6669,
+      "step": 4089
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.957052979141397e-06,
+      "loss": 0.6556,
+      "step": 4090
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.95256758310385e-06,
+      "loss": 0.7234,
+      "step": 4091
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.948082616966546e-06,
+      "loss": 0.6135,
+      "step": 4092
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.943598081671202e-06,
+      "loss": 0.6489,
+      "step": 4093
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.939113978159437e-06,
+      "loss": 0.67,
+      "step": 4094
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.934630307372779e-06,
+      "loss": 0.6797,
+      "step": 4095
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.930147070252664e-06,
+      "loss": 0.6816,
+      "step": 4096
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.925664267740443e-06,
+      "loss": 0.7323,
+      "step": 4097
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.921181900777368e-06,
+      "loss": 0.6914,
+      "step": 4098
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.916699970304604e-06,
+      "loss": 0.666,
+      "step": 4099
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.912218477263228e-06,
+      "loss": 0.7033,
+      "step": 4100
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.907737422594217e-06,
+      "loss": 0.6991,
+      "step": 4101
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.903256807238464e-06,
+      "loss": 0.6768,
+      "step": 4102
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.89877663213676e-06,
+      "loss": 0.6729,
+      "step": 4103
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.894296898229815e-06,
+      "loss": 0.6644,
+      "step": 4104
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.889817606458235e-06,
+      "loss": 0.5901,
+      "step": 4105
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.885338757762538e-06,
+      "loss": 0.7396,
+      "step": 4106
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.880860353083158e-06,
+      "loss": 0.6499,
+      "step": 4107
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.876382393360422e-06,
+      "loss": 0.7093,
+      "step": 4108
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.871904879534571e-06,
+      "loss": 0.6963,
+      "step": 4109
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.86742781254575e-06,
+      "loss": 0.6699,
+      "step": 4110
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.862951193334008e-06,
+      "loss": 0.7233,
+      "step": 4111
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.858475022839304e-06,
+      "loss": 0.624,
+      "step": 4112
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.8539993020015e-06,
+      "loss": 0.5955,
+      "step": 4113
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.84952403176037e-06,
+      "loss": 0.6561,
+      "step": 4114
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.845049213055587e-06,
+      "loss": 0.6543,
+      "step": 4115
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.840574846826728e-06,
+      "loss": 0.568,
+      "step": 4116
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.836100934013278e-06,
+      "loss": 0.7227,
+      "step": 4117
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.831627475554626e-06,
+      "loss": 0.68,
+      "step": 4118
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.827154472390069e-06,
+      "loss": 0.6698,
+      "step": 4119
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.822681925458798e-06,
+      "loss": 0.6348,
+      "step": 4120
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.818209835699925e-06,
+      "loss": 0.7111,
+      "step": 4121
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.813738204052451e-06,
+      "loss": 0.7567,
+      "step": 4122
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.809267031455288e-06,
+      "loss": 0.7055,
+      "step": 4123
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.804796318847253e-06,
+      "loss": 0.74,
+      "step": 4124
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.800326067167058e-06,
+      "loss": 0.6992,
+      "step": 4125
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.795856277353324e-06,
+      "loss": 0.7244,
+      "step": 4126
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.791386950344576e-06,
+      "loss": 0.5723,
+      "step": 4127
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.786918087079242e-06,
+      "loss": 0.5846,
+      "step": 4128
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.78244968849565e-06,
+      "loss": 0.5997,
+      "step": 4129
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.77798175553203e-06,
+      "loss": 0.5403,
+      "step": 4130
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.773514289126522e-06,
+      "loss": 0.6917,
+      "step": 4131
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.769047290217155e-06,
+      "loss": 0.6844,
+      "step": 4132
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.764580759741867e-06,
+      "loss": 0.5502,
+      "step": 4133
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.760114698638501e-06,
+      "loss": 0.7325,
+      "step": 4134
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.755649107844792e-06,
+      "loss": 0.6073,
+      "step": 4135
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.751183988298388e-06,
+      "loss": 0.5716,
+      "step": 4136
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.74671934093683e-06,
+      "loss": 0.6927,
+      "step": 4137
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.742255166697564e-06,
+      "loss": 0.7072,
+      "step": 4138
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.73779146651793e-06,
+      "loss": 0.6864,
+      "step": 4139
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.733328241335175e-06,
+      "loss": 0.7239,
+      "step": 4140
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.728865492086444e-06,
+      "loss": 0.6622,
+      "step": 4141
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.724403219708781e-06,
+      "loss": 0.6977,
+      "step": 4142
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.719941425139137e-06,
+      "loss": 0.6886,
+      "step": 4143
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.71548010931435e-06,
+      "loss": 0.5541,
+      "step": 4144
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.711019273171171e-06,
+      "loss": 0.6957,
+      "step": 4145
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.706558917646237e-06,
+      "loss": 0.687,
+      "step": 4146
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.702099043676094e-06,
+      "loss": 0.6389,
+      "step": 4147
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.697639652197184e-06,
+      "loss": 0.6688,
+      "step": 4148
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.693180744145844e-06,
+      "loss": 0.5587,
+      "step": 4149
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.688722320458317e-06,
+      "loss": 0.6903,
+      "step": 4150
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.68426438207074e-06,
+      "loss": 0.7104,
+      "step": 4151
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.679806929919147e-06,
+      "loss": 0.6177,
+      "step": 4152
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.67534996493947e-06,
+      "loss": 0.7022,
+      "step": 4153
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.670893488067539e-06,
+      "loss": 0.6937,
+      "step": 4154
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.666437500239084e-06,
+      "loss": 0.7277,
+      "step": 4155
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.66198200238973e-06,
+      "loss": 0.7012,
+      "step": 4156
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.657526995455002e-06,
+      "loss": 0.7002,
+      "step": 4157
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.653072480370317e-06,
+      "loss": 0.6805,
+      "step": 4158
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.648618458070996e-06,
+      "loss": 0.7142,
+      "step": 4159
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.644164929492245e-06,
+      "loss": 0.6656,
+      "step": 4160
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.639711895569178e-06,
+      "loss": 0.7199,
+      "step": 4161
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.635259357236797e-06,
+      "loss": 0.6847,
+      "step": 4162
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.630807315430005e-06,
+      "loss": 0.6566,
+      "step": 4163
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6263557710836014e-06,
+      "loss": 0.677,
+      "step": 4164
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.621904725132277e-06,
+      "loss": 0.6945,
+      "step": 4165
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.617454178510622e-06,
+      "loss": 0.7024,
+      "step": 4166
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.613004132153113e-06,
+      "loss": 0.6993,
+      "step": 4167
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.608554586994134e-06,
+      "loss": 0.7272,
+      "step": 4168
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.604105543967957e-06,
+      "loss": 0.6681,
+      "step": 4169
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.599657004008748e-06,
+      "loss": 0.6409,
+      "step": 4170
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.595208968050567e-06,
+      "loss": 0.7283,
+      "step": 4171
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.590761437027375e-06,
+      "loss": 0.692,
+      "step": 4172
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.58631441187302e-06,
+      "loss": 0.6899,
+      "step": 4173
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.581867893521247e-06,
+      "loss": 0.5322,
+      "step": 4174
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.577421882905689e-06,
+      "loss": 0.7035,
+      "step": 4175
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.57297638095988e-06,
+      "loss": 0.6265,
+      "step": 4176
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.568531388617242e-06,
+      "loss": 0.6928,
+      "step": 4177
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.564086906811092e-06,
+      "loss": 0.6745,
+      "step": 4178
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.559642936474643e-06,
+      "loss": 0.7026,
+      "step": 4179
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.555199478540993e-06,
+      "loss": 0.6603,
+      "step": 4180
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.550756533943144e-06,
+      "loss": 0.6713,
+      "step": 4181
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5463141036139745e-06,
+      "loss": 0.7156,
+      "step": 4182
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.541872188486266e-06,
+      "loss": 0.6983,
+      "step": 4183
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.53743078949269e-06,
+      "loss": 0.6767,
+      "step": 4184
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.532989907565808e-06,
+      "loss": 0.7138,
+      "step": 4185
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.528549543638077e-06,
+      "loss": 0.7127,
+      "step": 4186
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.524109698641839e-06,
+      "loss": 0.5727,
+      "step": 4187
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.519670373509334e-06,
+      "loss": 0.5936,
+      "step": 4188
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.515231569172683e-06,
+      "loss": 0.7123,
+      "step": 4189
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.510793286563908e-06,
+      "loss": 0.6839,
+      "step": 4190
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.506355526614914e-06,
+      "loss": 0.7237,
+      "step": 4191
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.501918290257501e-06,
+      "loss": 0.6019,
+      "step": 4192
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.497481578423358e-06,
+      "loss": 0.7196,
+      "step": 4193
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.493045392044063e-06,
+      "loss": 0.7049,
+      "step": 4194
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.488609732051086e-06,
+      "loss": 0.6885,
+      "step": 4195
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4841745993757795e-06,
+      "loss": 0.5901,
+      "step": 4196
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.479739994949393e-06,
+      "loss": 0.7236,
+      "step": 4197
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.475305919703062e-06,
+      "loss": 0.7321,
+      "step": 4198
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.470872374567808e-06,
+      "loss": 0.7007,
+      "step": 4199
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.466439360474551e-06,
+      "loss": 0.572,
+      "step": 4200
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.462006878354089e-06,
+      "loss": 0.7259,
+      "step": 4201
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.457574929137114e-06,
+      "loss": 0.5799,
+      "step": 4202
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4531435137541994e-06,
+      "loss": 0.689,
+      "step": 4203
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.448712633135817e-06,
+      "loss": 0.7235,
+      "step": 4204
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.444282288212317e-06,
+      "loss": 0.6922,
+      "step": 4205
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.439852479913941e-06,
+      "loss": 0.556,
+      "step": 4206
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.435423209170819e-06,
+      "loss": 0.6676,
+      "step": 4207
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.430994476912967e-06,
+      "loss": 0.6785,
+      "step": 4208
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.426566284070291e-06,
+      "loss": 0.6933,
+      "step": 4209
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4221386315725745e-06,
+      "loss": 0.671,
+      "step": 4210
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4177115203494955e-06,
+      "loss": 0.6713,
+      "step": 4211
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.413284951330616e-06,
+      "loss": 0.676,
+      "step": 4212
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4088589254453855e-06,
+      "loss": 0.6336,
+      "step": 4213
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.404433443623138e-06,
+      "loss": 0.6207,
+      "step": 4214
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.400008506793096e-06,
+      "loss": 0.7053,
+      "step": 4215
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.395584115884364e-06,
+      "loss": 0.6919,
+      "step": 4216
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3911602718259326e-06,
+      "loss": 0.7202,
+      "step": 4217
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3867369755466775e-06,
+      "loss": 0.6492,
+      "step": 4218
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.382314227975362e-06,
+      "loss": 0.7258,
+      "step": 4219
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.377892030040631e-06,
+      "loss": 0.7318,
+      "step": 4220
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.373470382671013e-06,
+      "loss": 0.6099,
+      "step": 4221
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.369049286794927e-06,
+      "loss": 0.6941,
+      "step": 4222
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.364628743340673e-06,
+      "loss": 0.5439,
+      "step": 4223
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.360208753236433e-06,
+      "loss": 0.705,
+      "step": 4224
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.355789317410272e-06,
+      "loss": 0.5738,
+      "step": 4225
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.351370436790144e-06,
+      "loss": 0.5558,
+      "step": 4226
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.34695211230388e-06,
+      "loss": 0.712,
+      "step": 4227
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.342534344879198e-06,
+      "loss": 0.6509,
+      "step": 4228
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3381171354437e-06,
+      "loss": 0.7275,
+      "step": 4229
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.33370048492487e-06,
+      "loss": 0.6828,
+      "step": 4230
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.329284394250075e-06,
+      "loss": 0.7222,
+      "step": 4231
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.324868864346559e-06,
+      "loss": 0.6752,
+      "step": 4232
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3204538961414554e-06,
+      "loss": 0.6678,
+      "step": 4233
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.316039490561776e-06,
+      "loss": 0.7099,
+      "step": 4234
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.311625648534414e-06,
+      "loss": 0.6993,
+      "step": 4235
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3072123709861495e-06,
+      "loss": 0.6779,
+      "step": 4236
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.302799658843639e-06,
+      "loss": 0.6527,
+      "step": 4237
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2983875130334224e-06,
+      "loss": 0.5872,
+      "step": 4238
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.293975934481916e-06,
+      "loss": 0.692,
+      "step": 4239
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.289564924115424e-06,
+      "loss": 0.5908,
+      "step": 4240
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.285154482860127e-06,
+      "loss": 0.7113,
+      "step": 4241
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2807446116420854e-06,
+      "loss": 0.668,
+      "step": 4242
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.276335311387246e-06,
+      "loss": 0.6448,
+      "step": 4243
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.271926583021428e-06,
+      "loss": 0.6505,
+      "step": 4244
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.267518427470338e-06,
+      "loss": 0.6391,
+      "step": 4245
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2631108456595535e-06,
+      "loss": 0.6876,
+      "step": 4246
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.258703838514538e-06,
+      "loss": 0.597,
+      "step": 4247
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.254297406960633e-06,
+      "loss": 0.6385,
+      "step": 4248
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.249891551923056e-06,
+      "loss": 0.6738,
+      "step": 4249
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.245486274326912e-06,
+      "loss": 0.6554,
+      "step": 4250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.241081575097176e-06,
+      "loss": 0.7135,
+      "step": 4251
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.236677455158705e-06,
+      "loss": 0.709,
+      "step": 4252
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.232273915436232e-06,
+      "loss": 0.5841,
+      "step": 4253
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2278709568543725e-06,
+      "loss": 0.7042,
+      "step": 4254
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.223468580337615e-06,
+      "loss": 0.6643,
+      "step": 4255
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.219066786810331e-06,
+      "loss": 0.6974,
+      "step": 4256
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.214665577196762e-06,
+      "loss": 0.7171,
+      "step": 4257
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.21026495242104e-06,
+      "loss": 0.68,
+      "step": 4258
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.205864913407162e-06,
+      "loss": 0.6971,
+      "step": 4259
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.201465461079005e-06,
+      "loss": 0.6981,
+      "step": 4260
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.197066596360324e-06,
+      "loss": 0.7361,
+      "step": 4261
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.192668320174751e-06,
+      "loss": 0.7034,
+      "step": 4262
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.188270633445793e-06,
+      "loss": 0.7342,
+      "step": 4263
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1838735370968345e-06,
+      "loss": 0.6957,
+      "step": 4264
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.179477032051139e-06,
+      "loss": 0.6976,
+      "step": 4265
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1750811192318405e-06,
+      "loss": 0.7288,
+      "step": 4266
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1706857995619476e-06,
+      "loss": 0.74,
+      "step": 4267
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.166291073964351e-06,
+      "loss": 0.7021,
+      "step": 4268
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.161896943361811e-06,
+      "loss": 0.6883,
+      "step": 4269
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.157503408676967e-06,
+      "loss": 0.6167,
+      "step": 4270
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.153110470832327e-06,
+      "loss": 0.5543,
+      "step": 4271
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1487181307502825e-06,
+      "loss": 0.6774,
+      "step": 4272
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.144326389353095e-06,
+      "loss": 0.6926,
+      "step": 4273
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.139935247562902e-06,
+      "loss": 0.663,
+      "step": 4274
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.135544706301708e-06,
+      "loss": 0.6815,
+      "step": 4275
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1311547664913994e-06,
+      "loss": 0.7391,
+      "step": 4276
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.126765429053732e-06,
+      "loss": 0.6657,
+      "step": 4277
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.122376694910338e-06,
+      "loss": 0.6824,
+      "step": 4278
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.117988564982724e-06,
+      "loss": 0.691,
+      "step": 4279
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.113601040192265e-06,
+      "loss": 0.6613,
+      "step": 4280
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.109214121460215e-06,
+      "loss": 0.6143,
+      "step": 4281
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.104827809707691e-06,
+      "loss": 0.606,
+      "step": 4282
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1004421058556926e-06,
+      "loss": 0.7223,
+      "step": 4283
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.096057010825088e-06,
+      "loss": 0.5809,
+      "step": 4284
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0916725255366136e-06,
+      "loss": 0.5931,
+      "step": 4285
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0872886509108865e-06,
+      "loss": 0.6892,
+      "step": 4286
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.082905387868391e-06,
+      "loss": 0.6881,
+      "step": 4287
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0785227373294804e-06,
+      "loss": 0.6833,
+      "step": 4288
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.07414070021438e-06,
+      "loss": 0.6826,
+      "step": 4289
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.069759277443191e-06,
+      "loss": 0.5951,
+      "step": 4290
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0653784699358815e-06,
+      "loss": 0.6893,
+      "step": 4291
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.060998278612292e-06,
+      "loss": 0.666,
+      "step": 4292
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.056618704392128e-06,
+      "loss": 0.6599,
+      "step": 4293
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.05223974819498e-06,
+      "loss": 0.6739,
+      "step": 4294
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.047861410940296e-06,
+      "loss": 0.6788,
+      "step": 4295
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.043483693547393e-06,
+      "loss": 0.5824,
+      "step": 4296
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.039106596935465e-06,
+      "loss": 0.701,
+      "step": 4297
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.034730122023572e-06,
+      "loss": 0.667,
+      "step": 4298
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.030354269730646e-06,
+      "loss": 0.6694,
+      "step": 4299
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.025979040975483e-06,
+      "loss": 0.6772,
+      "step": 4300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.021604436676757e-06,
+      "loss": 0.7077,
+      "step": 4301
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.017230457753004e-06,
+      "loss": 0.6824,
+      "step": 4302
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.012857105122627e-06,
+      "loss": 0.6828,
+      "step": 4303
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.008484379703902e-06,
+      "loss": 0.6949,
+      "step": 4304
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.004112282414974e-06,
+      "loss": 0.5631,
+      "step": 4305
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.999740814173852e-06,
+      "loss": 0.5799,
+      "step": 4306
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.995369975898414e-06,
+      "loss": 0.7032,
+      "step": 4307
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.99099976850641e-06,
+      "loss": 0.7044,
+      "step": 4308
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.986630192915457e-06,
+      "loss": 0.6964,
+      "step": 4309
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.982261250043026e-06,
+      "loss": 0.6999,
+      "step": 4310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.977892940806474e-06,
+      "loss": 0.7086,
+      "step": 4311
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.973525266123016e-06,
+      "loss": 0.6605,
+      "step": 4312
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.969158226909732e-06,
+      "loss": 0.6939,
+      "step": 4313
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.964791824083572e-06,
+      "loss": 0.6918,
+      "step": 4314
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.960426058561354e-06,
+      "loss": 0.6974,
+      "step": 4315
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9560609312597605e-06,
+      "loss": 0.6961,
+      "step": 4316
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.951696443095331e-06,
+      "loss": 0.611,
+      "step": 4317
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.947332594984484e-06,
+      "loss": 0.7078,
+      "step": 4318
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9429693878435e-06,
+      "loss": 0.7104,
+      "step": 4319
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.938606822588523e-06,
+      "loss": 0.6712,
+      "step": 4320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.934244900135559e-06,
+      "loss": 0.6577,
+      "step": 4321
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.929883621400486e-06,
+      "loss": 0.7341,
+      "step": 4322
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.925522987299045e-06,
+      "loss": 0.7437,
+      "step": 4323
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.921162998746839e-06,
+      "loss": 0.6914,
+      "step": 4324
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.916803656659335e-06,
+      "loss": 0.7003,
+      "step": 4325
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.912444961951864e-06,
+      "loss": 0.7037,
+      "step": 4326
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.908086915539627e-06,
+      "loss": 0.6938,
+      "step": 4327
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.903729518337683e-06,
+      "loss": 0.659,
+      "step": 4328
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.899372771260957e-06,
+      "loss": 0.6546,
+      "step": 4329
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.895016675224237e-06,
+      "loss": 0.6066,
+      "step": 4330
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8906612311421775e-06,
+      "loss": 0.6769,
+      "step": 4331
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.886306439929286e-06,
+      "loss": 0.6964,
+      "step": 4332
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8819523024999454e-06,
+      "loss": 0.6845,
+      "step": 4333
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.877598819768394e-06,
+      "loss": 0.6586,
+      "step": 4334
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.873245992648734e-06,
+      "loss": 0.6654,
+      "step": 4335
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.868893822054929e-06,
+      "loss": 0.691,
+      "step": 4336
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.864542308900809e-06,
+      "loss": 0.7088,
+      "step": 4337
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.860191454100065e-06,
+      "loss": 0.7116,
+      "step": 4338
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.855841258566241e-06,
+      "loss": 0.5629,
+      "step": 4339
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.851491723212753e-06,
+      "loss": 0.6705,
+      "step": 4340
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.847142848952877e-06,
+      "loss": 0.7295,
+      "step": 4341
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8427946366997445e-06,
+      "loss": 0.6146,
+      "step": 4342
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.838447087366351e-06,
+      "loss": 0.6994,
+      "step": 4343
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.834100201865556e-06,
+      "loss": 0.7263,
+      "step": 4344
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.829753981110079e-06,
+      "loss": 0.6712,
+      "step": 4345
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.825408426012492e-06,
+      "loss": 0.7146,
+      "step": 4346
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.821063537485237e-06,
+      "loss": 0.6445,
+      "step": 4347
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.81671931644061e-06,
+      "loss": 0.6759,
+      "step": 4348
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.812375763790769e-06,
+      "loss": 0.5876,
+      "step": 4349
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.808032880447731e-06,
+      "loss": 0.6321,
+      "step": 4350
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.803690667323377e-06,
+      "loss": 0.6479,
+      "step": 4351
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.799349125329444e-06,
+      "loss": 0.6949,
+      "step": 4352
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7950082553775185e-06,
+      "loss": 0.689,
+      "step": 4353
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.790668058379062e-06,
+      "loss": 0.6877,
+      "step": 4354
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.786328535245386e-06,
+      "loss": 0.6588,
+      "step": 4355
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.781989686887662e-06,
+      "loss": 0.6807,
+      "step": 4356
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.777651514216917e-06,
+      "loss": 0.6855,
+      "step": 4357
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7733140181440446e-06,
+      "loss": 0.7017,
+      "step": 4358
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.768977199579791e-06,
+      "loss": 0.6942,
+      "step": 4359
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.764641059434751e-06,
+      "loss": 0.6298,
+      "step": 4360
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.760305598619394e-06,
+      "loss": 0.7081,
+      "step": 4361
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7559708180440356e-06,
+      "loss": 0.6896,
+      "step": 4362
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.75163671861885e-06,
+      "loss": 0.6924,
+      "step": 4363
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.747303301253872e-06,
+      "loss": 0.6614,
+      "step": 4364
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.74297056685899e-06,
+      "loss": 0.7047,
+      "step": 4365
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.738638516343957e-06,
+      "loss": 0.6949,
+      "step": 4366
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.734307150618362e-06,
+      "loss": 0.6989,
+      "step": 4367
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.729976470591674e-06,
+      "loss": 0.6646,
+      "step": 4368
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.725646477173204e-06,
+      "loss": 0.6764,
+      "step": 4369
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.721317171272124e-06,
+      "loss": 0.6749,
+      "step": 4370
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.716988553797459e-06,
+      "loss": 0.6715,
+      "step": 4371
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.712660625658087e-06,
+      "loss": 0.6945,
+      "step": 4372
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.708333387762753e-06,
+      "loss": 0.6936,
+      "step": 4373
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.704006841020048e-06,
+      "loss": 0.6699,
+      "step": 4374
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.699680986338411e-06,
+      "loss": 0.6984,
+      "step": 4375
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.69535582462615e-06,
+      "loss": 0.698,
+      "step": 4376
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.691031356791417e-06,
+      "loss": 0.5851,
+      "step": 4377
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.686707583742227e-06,
+      "loss": 0.5712,
+      "step": 4378
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.682384506386439e-06,
+      "loss": 0.6802,
+      "step": 4379
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.678062125631777e-06,
+      "loss": 0.575,
+      "step": 4380
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6737404423858135e-06,
+      "loss": 0.6895,
+      "step": 4381
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.669419457555968e-06,
+      "loss": 0.6412,
+      "step": 4382
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6650991720495225e-06,
+      "loss": 0.6747,
+      "step": 4383
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.660779586773609e-06,
+      "loss": 0.6961,
+      "step": 4384
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.656460702635213e-06,
+      "loss": 0.6516,
+      "step": 4385
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.652142520541171e-06,
+      "loss": 0.5794,
+      "step": 4386
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.647825041398177e-06,
+      "loss": 0.639,
+      "step": 4387
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.643508266112777e-06,
+      "loss": 0.6594,
+      "step": 4388
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.639192195591353e-06,
+      "loss": 0.6608,
+      "step": 4389
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.634876830740163e-06,
+      "loss": 0.7147,
+      "step": 4390
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.630562172465302e-06,
+      "loss": 0.5753,
+      "step": 4391
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.626248221672724e-06,
+      "loss": 0.6747,
+      "step": 4392
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.621934979268225e-06,
+      "loss": 0.7048,
+      "step": 4393
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.617622446157465e-06,
+      "loss": 0.6971,
+      "step": 4394
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6133106232459495e-06,
+      "loss": 0.6828,
+      "step": 4395
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6089995114390225e-06,
+      "loss": 0.6638,
+      "step": 4396
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6046891116419e-06,
+      "loss": 0.5718,
+      "step": 4397
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.600379424759636e-06,
+      "loss": 0.7266,
+      "step": 4398
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.596070451697138e-06,
+      "loss": 0.6636,
+      "step": 4399
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.591762193359158e-06,
+      "loss": 0.6073,
+      "step": 4400
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.587454650650309e-06,
+      "loss": 0.6758,
+      "step": 4401
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.58314782447505e-06,
+      "loss": 0.6916,
+      "step": 4402
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.578841715737676e-06,
+      "loss": 0.7135,
+      "step": 4403
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.574536325342351e-06,
+      "loss": 0.6667,
+      "step": 4404
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.570231654193078e-06,
+      "loss": 0.6305,
+      "step": 4405
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.565927703193708e-06,
+      "loss": 0.663,
+      "step": 4406
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.561624473247945e-06,
+      "loss": 0.6973,
+      "step": 4407
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.557321965259343e-06,
+      "loss": 0.651,
+      "step": 4408
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.553020180131302e-06,
+      "loss": 0.5985,
+      "step": 4409
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.548719118767059e-06,
+      "loss": 0.7081,
+      "step": 4410
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.544418782069721e-06,
+      "loss": 0.573,
+      "step": 4411
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.540119170942229e-06,
+      "loss": 0.7355,
+      "step": 4412
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.535820286287371e-06,
+      "loss": 0.6519,
+      "step": 4413
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.531522129007788e-06,
+      "loss": 0.5542,
+      "step": 4414
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.527224700005963e-06,
+      "loss": 0.6056,
+      "step": 4415
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.522928000184239e-06,
+      "loss": 0.5807,
+      "step": 4416
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5186320304447805e-06,
+      "loss": 0.6602,
+      "step": 4417
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5143367916896235e-06,
+      "loss": 0.6901,
+      "step": 4418
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.510042284820641e-06,
+      "loss": 0.7203,
+      "step": 4419
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.505748510739549e-06,
+      "loss": 0.7096,
+      "step": 4420
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.501455470347917e-06,
+      "loss": 0.6701,
+      "step": 4421
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.497163164547151e-06,
+      "loss": 0.6817,
+      "step": 4422
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.492871594238514e-06,
+      "loss": 0.6243,
+      "step": 4423
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.488580760323109e-06,
+      "loss": 0.6685,
+      "step": 4424
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.484290663701877e-06,
+      "loss": 0.6649,
+      "step": 4425
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.480001305275618e-06,
+      "loss": 0.6629,
+      "step": 4426
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.47571268594497e-06,
+      "loss": 0.7031,
+      "step": 4427
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4714248066104134e-06,
+      "loss": 0.6732,
+      "step": 4428
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.467137668172274e-06,
+      "loss": 0.709,
+      "step": 4429
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.46285127153073e-06,
+      "loss": 0.6619,
+      "step": 4430
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.458565617585802e-06,
+      "loss": 0.7195,
+      "step": 4431
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.454280707237335e-06,
+      "loss": 0.6527,
+      "step": 4432
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.449996541385044e-06,
+      "loss": 0.7068,
+      "step": 4433
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.445713120928476e-06,
+      "loss": 0.6605,
+      "step": 4434
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.441430446767023e-06,
+      "loss": 0.6547,
+      "step": 4435
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.437148519799916e-06,
+      "loss": 0.6757,
+      "step": 4436
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.432867340926238e-06,
+      "loss": 0.6395,
+      "step": 4437
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.428586911044912e-06,
+      "loss": 0.7045,
+      "step": 4438
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4243072310546905e-06,
+      "loss": 0.685,
+      "step": 4439
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.42002830185419e-06,
+      "loss": 0.7066,
+      "step": 4440
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.415750124341855e-06,
+      "loss": 0.6879,
+      "step": 4441
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.411472699415977e-06,
+      "loss": 0.7006,
+      "step": 4442
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.407196027974687e-06,
+      "loss": 0.7028,
+      "step": 4443
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.402920110915963e-06,
+      "loss": 0.6795,
+      "step": 4444
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3986449491376246e-06,
+      "loss": 0.6782,
+      "step": 4445
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.394370543537316e-06,
+      "loss": 0.6702,
+      "step": 4446
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.390096895012547e-06,
+      "loss": 0.6434,
+      "step": 4447
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.385824004460655e-06,
+      "loss": 0.6738,
+      "step": 4448
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.381551872778819e-06,
+      "loss": 0.7083,
+      "step": 4449
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.377280500864061e-06,
+      "loss": 0.703,
+      "step": 4450
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.373009889613242e-06,
+      "loss": 0.708,
+      "step": 4451
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.368740039923072e-06,
+      "loss": 0.6465,
+      "step": 4452
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.364470952690078e-06,
+      "loss": 0.7226,
+      "step": 4453
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.360202628810652e-06,
+      "loss": 0.6855,
+      "step": 4454
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3559350691810165e-06,
+      "loss": 0.5722,
+      "step": 4455
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.351668274697229e-06,
+      "loss": 0.6728,
+      "step": 4456
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.347402246255192e-06,
+      "loss": 0.7408,
+      "step": 4457
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.343136984750642e-06,
+      "loss": 0.6394,
+      "step": 4458
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.338872491079167e-06,
+      "loss": 0.6076,
+      "step": 4459
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3346087661361725e-06,
+      "loss": 0.7308,
+      "step": 4460
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.33034581081692e-06,
+      "loss": 0.6735,
+      "step": 4461
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.326083626016506e-06,
+      "loss": 0.7278,
+      "step": 4462
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.321822212629862e-06,
+      "loss": 0.576,
+      "step": 4463
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.317561571551758e-06,
+      "loss": 0.7156,
+      "step": 4464
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3133017036768e-06,
+      "loss": 0.7034,
+      "step": 4465
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.309042609899443e-06,
+      "loss": 0.75,
+      "step": 4466
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.304784291113959e-06,
+      "loss": 0.672,
+      "step": 4467
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.300526748214477e-06,
+      "loss": 0.676,
+      "step": 4468
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.296269982094953e-06,
+      "loss": 0.7018,
+      "step": 4469
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.292013993649182e-06,
+      "loss": 0.6739,
+      "step": 4470
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.287758783770794e-06,
+      "loss": 0.7163,
+      "step": 4471
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.283504353353258e-06,
+      "loss": 0.7044,
+      "step": 4472
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2792507032898815e-06,
+      "loss": 0.6616,
+      "step": 4473
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.274997834473807e-06,
+      "loss": 0.7005,
+      "step": 4474
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.270745747798001e-06,
+      "loss": 0.6511,
+      "step": 4475
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.266494444155283e-06,
+      "loss": 0.6904,
+      "step": 4476
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2622439244383e-06,
+      "loss": 0.646,
+      "step": 4477
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.257994189539538e-06,
+      "loss": 0.693,
+      "step": 4478
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.253745240351309e-06,
+      "loss": 0.723,
+      "step": 4479
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.249497077765774e-06,
+      "loss": 0.6674,
+      "step": 4480
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.24524970267492e-06,
+      "loss": 0.6851,
+      "step": 4481
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.241003115970564e-06,
+      "loss": 0.6783,
+      "step": 4482
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.236757318544371e-06,
+      "loss": 0.6529,
+      "step": 4483
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.232512311287828e-06,
+      "loss": 0.6798,
+      "step": 4484
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.228268095092263e-06,
+      "loss": 0.6637,
+      "step": 4485
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.224024670848834e-06,
+      "loss": 0.6181,
+      "step": 4486
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2197820394485386e-06,
+      "loss": 0.642,
+      "step": 4487
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.215540201782207e-06,
+      "loss": 0.7064,
+      "step": 4488
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.211299158740486e-06,
+      "loss": 0.6266,
+      "step": 4489
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.20705891121388e-06,
+      "loss": 0.6486,
+      "step": 4490
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.202819460092713e-06,
+      "loss": 0.6456,
+      "step": 4491
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.198580806267145e-06,
+      "loss": 0.6785,
+      "step": 4492
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1943429506271655e-06,
+      "loss": 0.6906,
+      "step": 4493
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1901058940626e-06,
+      "loss": 0.7045,
+      "step": 4494
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.185869637463111e-06,
+      "loss": 0.6487,
+      "step": 4495
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.181634181718177e-06,
+      "loss": 0.6837,
+      "step": 4496
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1773995277171235e-06,
+      "loss": 0.628,
+      "step": 4497
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.173165676349103e-06,
+      "loss": 0.6988,
+      "step": 4498
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.168932628503099e-06,
+      "loss": 0.6701,
+      "step": 4499
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.164700385067924e-06,
+      "loss": 0.5935,
+      "step": 4500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.160468946932225e-06,
+      "loss": 0.6621,
+      "step": 4501
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.156238314984485e-06,
+      "loss": 0.6535,
+      "step": 4502
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.152008490112999e-06,
+      "loss": 0.571,
+      "step": 4503
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.147779473205915e-06,
+      "loss": 0.5839,
+      "step": 4504
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1435512651512e-06,
+      "loss": 0.6177,
+      "step": 4505
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.139323866836648e-06,
+      "loss": 0.6937,
+      "step": 4506
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.135097279149893e-06,
+      "loss": 0.5573,
+      "step": 4507
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1308715029783885e-06,
+      "loss": 0.6923,
+      "step": 4508
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.126646539209431e-06,
+      "loss": 0.7084,
+      "step": 4509
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.122422388730129e-06,
+      "loss": 0.6731,
+      "step": 4510
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1181990524274294e-06,
+      "loss": 0.6928,
+      "step": 4511
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.113976531188112e-06,
+      "loss": 0.6922,
+      "step": 4512
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.109754825898783e-06,
+      "loss": 0.7128,
+      "step": 4513
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.105533937445871e-06,
+      "loss": 0.6524,
+      "step": 4514
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.101313866715637e-06,
+      "loss": 0.7349,
+      "step": 4515
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.09709461459418e-06,
+      "loss": 0.7046,
+      "step": 4516
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.092876181967408e-06,
+      "loss": 0.7338,
+      "step": 4517
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.088658569721069e-06,
+      "loss": 0.7264,
+      "step": 4518
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.08444177874074e-06,
+      "loss": 0.5794,
+      "step": 4519
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.08022580991182e-06,
+      "loss": 0.6575,
+      "step": 4520
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0760106641195406e-06,
+      "loss": 0.664,
+      "step": 4521
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.071796342248952e-06,
+      "loss": 0.6888,
+      "step": 4522
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0675828451849484e-06,
+      "loss": 0.5909,
+      "step": 4523
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.063370173812228e-06,
+      "loss": 0.608,
+      "step": 4524
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.059158329015329e-06,
+      "loss": 0.6839,
+      "step": 4525
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.054947311678619e-06,
+      "loss": 0.6976,
+      "step": 4526
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.050737122686283e-06,
+      "loss": 0.68,
+      "step": 4527
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.046527762922339e-06,
+      "loss": 0.667,
+      "step": 4528
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.042319233270627e-06,
+      "loss": 0.6896,
+      "step": 4529
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.038111534614811e-06,
+      "loss": 0.6109,
+      "step": 4530
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.033904667838392e-06,
+      "loss": 0.6268,
+      "step": 4531
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.029698633824677e-06,
+      "loss": 0.7464,
+      "step": 4532
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.025493433456813e-06,
+      "loss": 0.6735,
+      "step": 4533
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.021289067617768e-06,
+      "loss": 0.6798,
+      "step": 4534
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.017085537190336e-06,
+      "loss": 0.6477,
+      "step": 4535
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.012882843057131e-06,
+      "loss": 0.6906,
+      "step": 4536
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0086809861005945e-06,
+      "loss": 0.5737,
+      "step": 4537
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.004479967203e-06,
+      "loss": 0.6929,
+      "step": 4538
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0002797872464235e-06,
+      "loss": 0.6683,
+      "step": 4539
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.996080447112788e-06,
+      "loss": 0.6847,
+      "step": 4540
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.991881947683828e-06,
+      "loss": 0.7098,
+      "step": 4541
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.987684289841107e-06,
+      "loss": 0.6786,
+      "step": 4542
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.983487474466006e-06,
+      "loss": 0.6728,
+      "step": 4543
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.979291502439728e-06,
+      "loss": 0.6773,
+      "step": 4544
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.975096374643317e-06,
+      "loss": 0.6765,
+      "step": 4545
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.970902091957609e-06,
+      "loss": 0.695,
+      "step": 4546
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.966708655263289e-06,
+      "loss": 0.6358,
+      "step": 4547
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.962516065440853e-06,
+      "loss": 0.6983,
+      "step": 4548
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.958324323370619e-06,
+      "loss": 0.6562,
+      "step": 4549
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.954133429932731e-06,
+      "loss": 0.6842,
+      "step": 4550
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.949943386007151e-06,
+      "loss": 0.6223,
+      "step": 4551
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.94575419247367e-06,
+      "loss": 0.7025,
+      "step": 4552
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.941565850211886e-06,
+      "loss": 0.6816,
+      "step": 4553
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.937378360101229e-06,
+      "loss": 0.7083,
+      "step": 4554
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.933191723020952e-06,
+      "loss": 0.6737,
+      "step": 4555
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.929005939850123e-06,
+      "loss": 0.6795,
+      "step": 4556
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.924821011467632e-06,
+      "loss": 0.5617,
+      "step": 4557
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.920636938752189e-06,
+      "loss": 0.5596,
+      "step": 4558
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.916453722582334e-06,
+      "loss": 0.7059,
+      "step": 4559
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9122713638364075e-06,
+      "loss": 0.6666,
+      "step": 4560
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.908089863392583e-06,
+      "loss": 0.7277,
+      "step": 4561
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.903909222128857e-06,
+      "loss": 0.652,
+      "step": 4562
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.899729440923038e-06,
+      "loss": 0.6561,
+      "step": 4563
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.895550520652755e-06,
+      "loss": 0.6732,
+      "step": 4564
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.891372462195459e-06,
+      "loss": 0.7023,
+      "step": 4565
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.887195266428424e-06,
+      "loss": 0.7062,
+      "step": 4566
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.883018934228728e-06,
+      "loss": 0.6879,
+      "step": 4567
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.87884346647328e-06,
+      "loss": 0.7175,
+      "step": 4568
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.874668864038808e-06,
+      "loss": 0.6962,
+      "step": 4569
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.870495127801853e-06,
+      "loss": 0.6655,
+      "step": 4570
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.866322258638777e-06,
+      "loss": 0.673,
+      "step": 4571
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.862150257425758e-06,
+      "loss": 0.6387,
+      "step": 4572
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.857979125038795e-06,
+      "loss": 0.6918,
+      "step": 4573
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.853808862353699e-06,
+      "loss": 0.6865,
+      "step": 4574
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.849639470246102e-06,
+      "loss": 0.6498,
+      "step": 4575
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.845470949591457e-06,
+      "loss": 0.7127,
+      "step": 4576
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.841303301265027e-06,
+      "loss": 0.6873,
+      "step": 4577
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.837136526141896e-06,
+      "loss": 0.6529,
+      "step": 4578
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.832970625096961e-06,
+      "loss": 0.722,
+      "step": 4579
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.828805599004943e-06,
+      "loss": 0.6508,
+      "step": 4580
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.824641448740373e-06,
+      "loss": 0.658,
+      "step": 4581
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.820478175177593e-06,
+      "loss": 0.6501,
+      "step": 4582
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.816315779190775e-06,
+      "loss": 0.7085,
+      "step": 4583
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.812154261653894e-06,
+      "loss": 0.6755,
+      "step": 4584
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.807993623440746e-06,
+      "loss": 0.5629,
+      "step": 4585
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.803833865424949e-06,
+      "loss": 0.689,
+      "step": 4586
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.79967498847992e-06,
+      "loss": 0.6403,
+      "step": 4587
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7955169934789064e-06,
+      "loss": 0.6453,
+      "step": 4588
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.791359881294962e-06,
+      "loss": 0.6671,
+      "step": 4589
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.787203652800955e-06,
+      "loss": 0.6985,
+      "step": 4590
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.783048308869574e-06,
+      "loss": 0.7071,
+      "step": 4591
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.778893850373315e-06,
+      "loss": 0.6727,
+      "step": 4592
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.774740278184492e-06,
+      "loss": 0.6821,
+      "step": 4593
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.770587593175237e-06,
+      "loss": 0.643,
+      "step": 4594
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.766435796217489e-06,
+      "loss": 0.6781,
+      "step": 4595
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7622848881829965e-06,
+      "loss": 0.6612,
+      "step": 4596
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.758134869943335e-06,
+      "loss": 0.6626,
+      "step": 4597
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.753985742369881e-06,
+      "loss": 0.705,
+      "step": 4598
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.749837506333828e-06,
+      "loss": 0.604,
+      "step": 4599
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.74569016270619e-06,
+      "loss": 0.6448,
+      "step": 4600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7415437123577776e-06,
+      "loss": 0.5567,
+      "step": 4601
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.73739815615923e-06,
+      "loss": 0.6386,
+      "step": 4602
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7332534949809885e-06,
+      "loss": 0.6417,
+      "step": 4603
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.729109729693305e-06,
+      "loss": 0.6503,
+      "step": 4604
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.724966861166256e-06,
+      "loss": 0.6863,
+      "step": 4605
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.720824890269712e-06,
+      "loss": 0.6637,
+      "step": 4606
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.71668381787337e-06,
+      "loss": 0.6576,
+      "step": 4607
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.712543644846735e-06,
+      "loss": 0.6938,
+      "step": 4608
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.708404372059118e-06,
+      "loss": 0.6899,
+      "step": 4609
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.704266000379641e-06,
+      "loss": 0.6696,
+      "step": 4610
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.700128530677246e-06,
+      "loss": 0.652,
+      "step": 4611
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.695991963820671e-06,
+      "loss": 0.7231,
+      "step": 4612
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.691856300678478e-06,
+      "loss": 0.741,
+      "step": 4613
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6877215421190354e-06,
+      "loss": 0.6581,
+      "step": 4614
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6835876890105166e-06,
+      "loss": 0.6582,
+      "step": 4615
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.679454742220912e-06,
+      "loss": 0.6912,
+      "step": 4616
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.675322702618018e-06,
+      "loss": 0.6084,
+      "step": 4617
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.671191571069432e-06,
+      "loss": 0.6572,
+      "step": 4618
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.667061348442583e-06,
+      "loss": 0.6872,
+      "step": 4619
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.662932035604683e-06,
+      "loss": 0.7107,
+      "step": 4620
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.658803633422773e-06,
+      "loss": 0.6549,
+      "step": 4621
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.654676142763696e-06,
+      "loss": 0.6639,
+      "step": 4622
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.650549564494101e-06,
+      "loss": 0.6342,
+      "step": 4623
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.646423899480444e-06,
+      "loss": 0.6338,
+      "step": 4624
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.642299148588999e-06,
+      "loss": 0.6605,
+      "step": 4625
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.638175312685835e-06,
+      "loss": 0.5859,
+      "step": 4626
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.634052392636842e-06,
+      "loss": 0.674,
+      "step": 4627
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.629930389307705e-06,
+      "loss": 0.6733,
+      "step": 4628
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.625809303563925e-06,
+      "loss": 0.7016,
+      "step": 4629
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.621689136270814e-06,
+      "loss": 0.7026,
+      "step": 4630
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.61756988829348e-06,
+      "loss": 0.7162,
+      "step": 4631
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.61345156049684e-06,
+      "loss": 0.7051,
+      "step": 4632
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.609334153745628e-06,
+      "loss": 0.656,
+      "step": 4633
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.605217668904371e-06,
+      "loss": 0.6502,
+      "step": 4634
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.601102106837411e-06,
+      "loss": 0.6917,
+      "step": 4635
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5969874684089e-06,
+      "loss": 0.6898,
+      "step": 4636
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.592873754482781e-06,
+      "loss": 0.6794,
+      "step": 4637
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.588760965922824e-06,
+      "loss": 0.6842,
+      "step": 4638
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.584649103592583e-06,
+      "loss": 0.686,
+      "step": 4639
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.580538168355429e-06,
+      "loss": 0.5509,
+      "step": 4640
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.576428161074541e-06,
+      "loss": 0.6781,
+      "step": 4641
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.572319082612894e-06,
+      "loss": 0.6595,
+      "step": 4642
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.568210933833274e-06,
+      "loss": 0.6799,
+      "step": 4643
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5641037155982744e-06,
+      "loss": 0.6437,
+      "step": 4644
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.55999742877029e-06,
+      "loss": 0.6705,
+      "step": 4645
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.555892074211512e-06,
+      "loss": 0.715,
+      "step": 4646
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.551787652783951e-06,
+      "loss": 0.7042,
+      "step": 4647
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.547684165349408e-06,
+      "loss": 0.6661,
+      "step": 4648
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.543581612769498e-06,
+      "loss": 0.6353,
+      "step": 4649
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.539479995905639e-06,
+      "loss": 0.5647,
+      "step": 4650
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5353793156190405e-06,
+      "loss": 0.6912,
+      "step": 4651
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.531279572770731e-06,
+      "loss": 0.6926,
+      "step": 4652
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.527180768221536e-06,
+      "loss": 0.6713,
+      "step": 4653
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.523082902832076e-06,
+      "loss": 0.6512,
+      "step": 4654
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.518985977462788e-06,
+      "loss": 0.6981,
+      "step": 4655
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5148899929738995e-06,
+      "loss": 0.6587,
+      "step": 4656
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.510794950225448e-06,
+      "loss": 0.6541,
+      "step": 4657
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.506700850077279e-06,
+      "loss": 0.6971,
+      "step": 4658
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.502607693389026e-06,
+      "loss": 0.6952,
+      "step": 4659
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.498515481020126e-06,
+      "loss": 0.6626,
+      "step": 4660
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.494424213829832e-06,
+      "loss": 0.6822,
+      "step": 4661
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.490333892677182e-06,
+      "loss": 0.7027,
+      "step": 4662
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.486244518421028e-06,
+      "loss": 0.7021,
+      "step": 4663
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.48215609192001e-06,
+      "loss": 0.7256,
+      "step": 4664
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.478068614032584e-06,
+      "loss": 0.6932,
+      "step": 4665
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.473982085616998e-06,
+      "loss": 0.7236,
+      "step": 4666
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.469896507531302e-06,
+      "loss": 0.6761,
+      "step": 4667
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.465811880633343e-06,
+      "loss": 0.5706,
+      "step": 4668
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.461728205780777e-06,
+      "loss": 0.6484,
+      "step": 4669
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4576454838310486e-06,
+      "loss": 0.7251,
+      "step": 4670
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.453563715641413e-06,
+      "loss": 0.6615,
+      "step": 4671
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.449482902068923e-06,
+      "loss": 0.7083,
+      "step": 4672
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.445403043970427e-06,
+      "loss": 0.6504,
+      "step": 4673
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.441324142202568e-06,
+      "loss": 0.6385,
+      "step": 4674
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.437246197621805e-06,
+      "loss": 0.6943,
+      "step": 4675
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.433169211084377e-06,
+      "loss": 0.6933,
+      "step": 4676
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.429093183446339e-06,
+      "loss": 0.6579,
+      "step": 4677
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.425018115563526e-06,
+      "loss": 0.6636,
+      "step": 4678
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.420944008291589e-06,
+      "loss": 0.6605,
+      "step": 4679
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.416870862485973e-06,
+      "loss": 0.5966,
+      "step": 4680
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.412798679001912e-06,
+      "loss": 0.7022,
+      "step": 4681
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4087274586944415e-06,
+      "loss": 0.6644,
+      "step": 4682
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.404657202418406e-06,
+      "loss": 0.6714,
+      "step": 4683
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.400587911028431e-06,
+      "loss": 0.7167,
+      "step": 4684
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.39651958537895e-06,
+      "loss": 0.6718,
+      "step": 4685
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3924522263241965e-06,
+      "loss": 0.5915,
+      "step": 4686
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.388385834718187e-06,
+      "loss": 0.6491,
+      "step": 4687
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.384320411414752e-06,
+      "loss": 0.6557,
+      "step": 4688
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.380255957267504e-06,
+      "loss": 0.6702,
+      "step": 4689
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3761924731298565e-06,
+      "loss": 0.6598,
+      "step": 4690
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.372129959855028e-06,
+      "loss": 0.6554,
+      "step": 4691
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.368068418296018e-06,
+      "loss": 0.6551,
+      "step": 4692
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.364007849305635e-06,
+      "loss": 0.6644,
+      "step": 4693
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.359948253736481e-06,
+      "loss": 0.619,
+      "step": 4694
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.355889632440948e-06,
+      "loss": 0.706,
+      "step": 4695
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.351831986271223e-06,
+      "loss": 0.6505,
+      "step": 4696
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.347775316079298e-06,
+      "loss": 0.6629,
+      "step": 4697
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.343719622716946e-06,
+      "loss": 0.711,
+      "step": 4698
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.339664907035752e-06,
+      "loss": 0.6874,
+      "step": 4699
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3356111698870785e-06,
+      "loss": 0.6224,
+      "step": 4700
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.331558412122092e-06,
+      "loss": 0.6356,
+      "step": 4701
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.327506634591757e-06,
+      "loss": 0.6404,
+      "step": 4702
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.323455838146825e-06,
+      "loss": 0.7192,
+      "step": 4703
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.319406023637837e-06,
+      "loss": 0.6777,
+      "step": 4704
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.315357191915141e-06,
+      "loss": 0.6624,
+      "step": 4705
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3113093438288675e-06,
+      "loss": 0.6669,
+      "step": 4706
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.307262480228947e-06,
+      "loss": 0.6891,
+      "step": 4707
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.303216601965107e-06,
+      "loss": 0.5751,
+      "step": 4708
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2991717098868534e-06,
+      "loss": 0.5878,
+      "step": 4709
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.295127804843494e-06,
+      "loss": 0.6691,
+      "step": 4710
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.291084887684137e-06,
+      "loss": 0.6569,
+      "step": 4711
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.287042959257667e-06,
+      "loss": 0.6588,
+      "step": 4712
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.283002020412777e-06,
+      "loss": 0.7067,
+      "step": 4713
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.278962071997936e-06,
+      "loss": 0.6922,
+      "step": 4714
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.274923114861419e-06,
+      "loss": 0.5627,
+      "step": 4715
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.270885149851294e-06,
+      "loss": 0.6503,
+      "step": 4716
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.266848177815402e-06,
+      "loss": 0.678,
+      "step": 4717
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2628121996013906e-06,
+      "loss": 0.6837,
+      "step": 4718
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.258777216056703e-06,
+      "loss": 0.6817,
+      "step": 4719
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.254743228028557e-06,
+      "loss": 0.6434,
+      "step": 4720
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.250710236363976e-06,
+      "loss": 0.6745,
+      "step": 4721
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.246678241909772e-06,
+      "loss": 0.6785,
+      "step": 4722
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.24264724551254e-06,
+      "loss": 0.7056,
+      "step": 4723
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.238617248018667e-06,
+      "loss": 0.6924,
+      "step": 4724
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2345882502743415e-06,
+      "loss": 0.5614,
+      "step": 4725
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.230560253125527e-06,
+      "loss": 0.7115,
+      "step": 4726
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.226533257417989e-06,
+      "loss": 0.6707,
+      "step": 4727
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2225072639972705e-06,
+      "loss": 0.6915,
+      "step": 4728
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.218482273708716e-06,
+      "loss": 0.6513,
+      "step": 4729
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.214458287397457e-06,
+      "loss": 0.6876,
+      "step": 4730
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.210435305908408e-06,
+      "loss": 0.6622,
+      "step": 4731
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2064133300862725e-06,
+      "loss": 0.6679,
+      "step": 4732
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.202392360775556e-06,
+      "loss": 0.6823,
+      "step": 4733
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.198372398820533e-06,
+      "loss": 0.6634,
+      "step": 4734
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1943534450652785e-06,
+      "loss": 0.6986,
+      "step": 4735
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.190335500353664e-06,
+      "loss": 0.596,
+      "step": 4736
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.186318565529325e-06,
+      "loss": 0.5621,
+      "step": 4737
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.182302641435711e-06,
+      "loss": 0.6726,
+      "step": 4738
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.178287728916042e-06,
+      "loss": 0.6466,
+      "step": 4739
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.174273828813325e-06,
+      "loss": 0.6908,
+      "step": 4740
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.170260941970371e-06,
+      "loss": 0.6813,
+      "step": 4741
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.166249069229759e-06,
+      "loss": 0.6492,
+      "step": 4742
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.162238211433867e-06,
+      "loss": 0.6867,
+      "step": 4743
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.158228369424861e-06,
+      "loss": 0.5618,
+      "step": 4744
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.154219544044684e-06,
+      "loss": 0.7114,
+      "step": 4745
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1502117361350675e-06,
+      "loss": 0.6619,
+      "step": 4746
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1462049465375406e-06,
+      "loss": 0.5258,
+      "step": 4747
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.142199176093404e-06,
+      "loss": 0.5974,
+      "step": 4748
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.138194425643757e-06,
+      "loss": 0.6636,
+      "step": 4749
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.134190696029472e-06,
+      "loss": 0.6428,
+      "step": 4750
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1301879880912175e-06,
+      "loss": 0.6888,
+      "step": 4751
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1261863026694455e-06,
+      "loss": 0.6843,
+      "step": 4752
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.122185640604391e-06,
+      "loss": 0.6832,
+      "step": 4753
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.118186002736069e-06,
+      "loss": 0.662,
+      "step": 4754
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.114187389904295e-06,
+      "loss": 0.6828,
+      "step": 4755
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.110189802948649e-06,
+      "loss": 0.544,
+      "step": 4756
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.10619324270851e-06,
+      "loss": 0.6219,
+      "step": 4757
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.102197710023044e-06,
+      "loss": 0.566,
+      "step": 4758
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.098203205731187e-06,
+      "loss": 0.6826,
+      "step": 4759
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0942097306716664e-06,
+      "loss": 0.7068,
+      "step": 4760
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.090217285682999e-06,
+      "loss": 0.6693,
+      "step": 4761
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.086225871603474e-06,
+      "loss": 0.6659,
+      "step": 4762
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.082235489271179e-06,
+      "loss": 0.6789,
+      "step": 4763
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.078246139523966e-06,
+      "loss": 0.6551,
+      "step": 4764
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.074257823199485e-06,
+      "loss": 0.701,
+      "step": 4765
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.070270541135171e-06,
+      "loss": 0.5787,
+      "step": 4766
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.066284294168224e-06,
+      "loss": 0.6914,
+      "step": 4767
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.062299083135642e-06,
+      "loss": 0.7064,
+      "step": 4768
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.058314908874206e-06,
+      "loss": 0.6596,
+      "step": 4769
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.054331772220469e-06,
+      "loss": 0.6411,
+      "step": 4770
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.050349674010773e-06,
+      "loss": 0.6681,
+      "step": 4771
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.046368615081246e-06,
+      "loss": 0.6668,
+      "step": 4772
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0423885962677885e-06,
+      "loss": 0.713,
+      "step": 4773
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0384096184060835e-06,
+      "loss": 0.6675,
+      "step": 4774
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.034431682331607e-06,
+      "loss": 0.6709,
+      "step": 4775
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.030454788879597e-06,
+      "loss": 0.6197,
+      "step": 4776
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.026478938885093e-06,
+      "loss": 0.6612,
+      "step": 4777
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.022504133182901e-06,
+      "loss": 0.6954,
+      "step": 4778
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.018530372607611e-06,
+      "loss": 0.7018,
+      "step": 4779
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.014557657993604e-06,
+      "loss": 0.6646,
+      "step": 4780
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.010585990175027e-06,
+      "loss": 0.6342,
+      "step": 4781
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.006615369985808e-06,
+      "loss": 0.6724,
+      "step": 4782
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.002645798259667e-06,
+      "loss": 0.7064,
+      "step": 4783
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9986772758300905e-06,
+      "loss": 0.6847,
+      "step": 4784
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9947098035303586e-06,
+      "loss": 0.6279,
+      "step": 4785
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.990743382193516e-06,
+      "loss": 0.663,
+      "step": 4786
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.986778012652395e-06,
+      "loss": 0.7039,
+      "step": 4787
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.982813695739612e-06,
+      "loss": 0.7237,
+      "step": 4788
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.978850432287551e-06,
+      "loss": 0.5527,
+      "step": 4789
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.974888223128378e-06,
+      "loss": 0.6757,
+      "step": 4790
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.970927069094046e-06,
+      "loss": 0.7157,
+      "step": 4791
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.966966971016274e-06,
+      "loss": 0.6723,
+      "step": 4792
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.963007929726568e-06,
+      "loss": 0.6626,
+      "step": 4793
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9590499460562125e-06,
+      "loss": 0.671,
+      "step": 4794
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9550930208362645e-06,
+      "loss": 0.7081,
+      "step": 4795
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.951137154897558e-06,
+      "loss": 0.6702,
+      "step": 4796
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.947182349070715e-06,
+      "loss": 0.6967,
+      "step": 4797
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.94322860418612e-06,
+      "loss": 0.6313,
+      "step": 4798
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.939275921073949e-06,
+      "loss": 0.7161,
+      "step": 4799
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.935324300564143e-06,
+      "loss": 0.6969,
+      "step": 4800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.931373743486427e-06,
+      "loss": 0.6905,
+      "step": 4801
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9274242506703115e-06,
+      "loss": 0.6903,
+      "step": 4802
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.923475822945055e-06,
+      "loss": 0.6892,
+      "step": 4803
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.91952846113972e-06,
+      "loss": 0.6729,
+      "step": 4804
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9155821660831396e-06,
+      "loss": 0.6717,
+      "step": 4805
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9116369386039096e-06,
+      "loss": 0.6803,
+      "step": 4806
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.907692779530418e-06,
+      "loss": 0.5589,
+      "step": 4807
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.903749689690821e-06,
+      "loss": 0.6689,
+      "step": 4808
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.899807669913051e-06,
+      "loss": 0.6858,
+      "step": 4809
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.89586672102481e-06,
+      "loss": 0.6756,
+      "step": 4810
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.891926843853588e-06,
+      "loss": 0.6845,
+      "step": 4811
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.887988039226636e-06,
+      "loss": 0.5858,
+      "step": 4812
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.884050307970994e-06,
+      "loss": 0.671,
+      "step": 4813
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.880113650913462e-06,
+      "loss": 0.6833,
+      "step": 4814
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8761780688806225e-06,
+      "loss": 0.6347,
+      "step": 4815
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.872243562698843e-06,
+      "loss": 0.6548,
+      "step": 4816
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.868310133194235e-06,
+      "loss": 0.7046,
+      "step": 4817
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.864377781192712e-06,
+      "loss": 0.5618,
+      "step": 4818
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.860446507519953e-06,
+      "loss": 0.682,
+      "step": 4819
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.856516313001404e-06,
+      "loss": 0.7079,
+      "step": 4820
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8525871984622965e-06,
+      "loss": 0.6814,
+      "step": 4821
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.84865916472762e-06,
+      "loss": 0.6795,
+      "step": 4822
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.844732212622154e-06,
+      "loss": 0.6954,
+      "step": 4823
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8408063429704344e-06,
+      "loss": 0.6471,
+      "step": 4824
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8368815565967845e-06,
+      "loss": 0.6702,
+      "step": 4825
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.832957854325286e-06,
+      "loss": 0.6697,
+      "step": 4826
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.82903523697981e-06,
+      "loss": 0.6594,
+      "step": 4827
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.82511370538398e-06,
+      "loss": 0.6921,
+      "step": 4828
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8211932603612056e-06,
+      "loss": 0.6552,
+      "step": 4829
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.817273902734669e-06,
+      "loss": 0.657,
+      "step": 4830
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.813355633327317e-06,
+      "loss": 0.6843,
+      "step": 4831
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.809438452961863e-06,
+      "loss": 0.6001,
+      "step": 4832
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.805522362460809e-06,
+      "loss": 0.6463,
+      "step": 4833
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.80160736264641e-06,
+      "loss": 0.6552,
+      "step": 4834
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.797693454340707e-06,
+      "loss": 0.7065,
+      "step": 4835
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7937806383655e-06,
+      "loss": 0.6886,
+      "step": 4836
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.789868915542365e-06,
+      "loss": 0.6591,
+      "step": 4837
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.785958286692658e-06,
+      "loss": 0.6594,
+      "step": 4838
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.782048752637478e-06,
+      "loss": 0.5162,
+      "step": 4839
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7781403141977224e-06,
+      "loss": 0.6866,
+      "step": 4840
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.774232972194048e-06,
+      "loss": 0.6169,
+      "step": 4841
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.770326727446877e-06,
+      "loss": 0.5375,
+      "step": 4842
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.766421580776406e-06,
+      "loss": 0.7134,
+      "step": 4843
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.762517533002605e-06,
+      "loss": 0.6428,
+      "step": 4844
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.758614584945205e-06,
+      "loss": 0.6724,
+      "step": 4845
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.754712737423706e-06,
+      "loss": 0.6797,
+      "step": 4846
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.750811991257388e-06,
+      "loss": 0.6723,
+      "step": 4847
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.746912347265284e-06,
+      "loss": 0.6825,
+      "step": 4848
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.743013806266214e-06,
+      "loss": 0.6536,
+      "step": 4849
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.739116369078746e-06,
+      "loss": 0.6184,
+      "step": 4850
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7352200365212305e-06,
+      "loss": 0.6919,
+      "step": 4851
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7313248094117914e-06,
+      "loss": 0.6645,
+      "step": 4852
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.727430688568294e-06,
+      "loss": 0.6812,
+      "step": 4853
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.723537674808398e-06,
+      "loss": 0.6786,
+      "step": 4854
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.719645768949524e-06,
+      "loss": 0.6625,
+      "step": 4855
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.715754971808849e-06,
+      "loss": 0.6524,
+      "step": 4856
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.71186528420333e-06,
+      "loss": 0.6936,
+      "step": 4857
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.70797670694969e-06,
+      "loss": 0.6457,
+      "step": 4858
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.70408924086441e-06,
+      "loss": 0.6848,
+      "step": 4859
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.700202886763741e-06,
+      "loss": 0.6635,
+      "step": 4860
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.69631764546371e-06,
+      "loss": 0.5391,
+      "step": 4861
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6924335177800925e-06,
+      "loss": 0.6953,
+      "step": 4862
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6885505045284515e-06,
+      "loss": 0.598,
+      "step": 4863
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.684668606524097e-06,
+      "loss": 0.6994,
+      "step": 4864
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.680787824582113e-06,
+      "loss": 0.66,
+      "step": 4865
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.676908159517361e-06,
+      "loss": 0.6603,
+      "step": 4866
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.673029612144437e-06,
+      "loss": 0.6726,
+      "step": 4867
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.669152183277731e-06,
+      "loss": 0.6564,
+      "step": 4868
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.665275873731392e-06,
+      "loss": 0.6986,
+      "step": 4869
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.661400684319322e-06,
+      "loss": 0.5747,
+      "step": 4870
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.657526615855204e-06,
+      "loss": 0.6897,
+      "step": 4871
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.653653669152469e-06,
+      "loss": 0.6341,
+      "step": 4872
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.649781845024333e-06,
+      "loss": 0.7086,
+      "step": 4873
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.645911144283751e-06,
+      "loss": 0.6916,
+      "step": 4874
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.642041567743468e-06,
+      "loss": 0.6507,
+      "step": 4875
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.638173116215972e-06,
+      "loss": 0.6425,
+      "step": 4876
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.634305790513529e-06,
+      "loss": 0.6738,
+      "step": 4877
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.630439591448158e-06,
+      "loss": 0.6055,
+      "step": 4878
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.626574519831648e-06,
+      "loss": 0.6947,
+      "step": 4879
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.622710576475556e-06,
+      "loss": 0.5641,
+      "step": 4880
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.618847762191189e-06,
+      "loss": 0.6414,
+      "step": 4881
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.614986077789622e-06,
+      "loss": 0.7045,
+      "step": 4882
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.611125524081701e-06,
+      "loss": 0.653,
+      "step": 4883
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.607266101878022e-06,
+      "loss": 0.7077,
+      "step": 4884
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.603407811988956e-06,
+      "loss": 0.7004,
+      "step": 4885
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.599550655224622e-06,
+      "loss": 0.5862,
+      "step": 4886
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.595694632394914e-06,
+      "loss": 0.7134,
+      "step": 4887
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.591839744309487e-06,
+      "loss": 0.7019,
+      "step": 4888
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.587985991777744e-06,
+      "loss": 0.5567,
+      "step": 4889
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.58413337560886e-06,
+      "loss": 0.6921,
+      "step": 4890
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.580281896611779e-06,
+      "loss": 0.6494,
+      "step": 4891
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.576431555595189e-06,
+      "loss": 0.7201,
+      "step": 4892
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.57258235336755e-06,
+      "loss": 0.7148,
+      "step": 4893
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.568734290737086e-06,
+      "loss": 0.7149,
+      "step": 4894
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.564887368511772e-06,
+      "loss": 0.662,
+      "step": 4895
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.561041587499344e-06,
+      "loss": 0.7106,
+      "step": 4896
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.557196948507311e-06,
+      "loss": 0.6996,
+      "step": 4897
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5533534523429234e-06,
+      "loss": 0.7132,
+      "step": 4898
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.549511099813211e-06,
+      "loss": 0.6207,
+      "step": 4899
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.545669891724946e-06,
+      "loss": 0.6849,
+      "step": 4900
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.541829828884674e-06,
+      "loss": 0.6846,
+      "step": 4901
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5379909120987e-06,
+      "loss": 0.598,
+      "step": 4902
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.534153142173068e-06,
+      "loss": 0.6698,
+      "step": 4903
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.530316519913605e-06,
+      "loss": 0.6665,
+      "step": 4904
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.526481046125891e-06,
+      "loss": 0.6031,
+      "step": 4905
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.522646721615256e-06,
+      "loss": 0.6701,
+      "step": 4906
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5188135471868e-06,
+      "loss": 0.6025,
+      "step": 4907
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.51498152364537e-06,
+      "loss": 0.6239,
+      "step": 4908
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.511150651795584e-06,
+      "loss": 0.6744,
+      "step": 4909
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.507320932441807e-06,
+      "loss": 0.6895,
+      "step": 4910
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.503492366388171e-06,
+      "loss": 0.6918,
+      "step": 4911
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.499664954438556e-06,
+      "loss": 0.7172,
+      "step": 4912
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4958386973966125e-06,
+      "loss": 0.6798,
+      "step": 4913
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.492013596065734e-06,
+      "loss": 0.6974,
+      "step": 4914
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.488189651249082e-06,
+      "loss": 0.6825,
+      "step": 4915
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.48436686374958e-06,
+      "loss": 0.5817,
+      "step": 4916
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.480545234369884e-06,
+      "loss": 0.6583,
+      "step": 4917
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4767247639124315e-06,
+      "loss": 0.6464,
+      "step": 4918
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.472905453179411e-06,
+      "loss": 0.6649,
+      "step": 4919
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.46908730297276e-06,
+      "loss": 0.6719,
+      "step": 4920
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.465270314094181e-06,
+      "loss": 0.641,
+      "step": 4921
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4614544873451246e-06,
+      "loss": 0.5813,
+      "step": 4922
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.457639823526807e-06,
+      "loss": 0.6816,
+      "step": 4923
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4538263234401915e-06,
+      "loss": 0.6514,
+      "step": 4924
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.450013987885996e-06,
+      "loss": 0.6854,
+      "step": 4925
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4462028176647034e-06,
+      "loss": 0.6619,
+      "step": 4926
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.442392813576548e-06,
+      "loss": 0.5832,
+      "step": 4927
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.438583976421513e-06,
+      "loss": 0.6597,
+      "step": 4928
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.434776306999345e-06,
+      "loss": 0.6594,
+      "step": 4929
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4309698061095455e-06,
+      "loss": 0.6612,
+      "step": 4930
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.427164474551362e-06,
+      "loss": 0.68,
+      "step": 4931
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.423360313123799e-06,
+      "loss": 0.6624,
+      "step": 4932
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.419557322625625e-06,
+      "loss": 0.6475,
+      "step": 4933
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.415755503855348e-06,
+      "loss": 0.6538,
+      "step": 4934
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.411954857611245e-06,
+      "loss": 0.6558,
+      "step": 4935
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.408155384691332e-06,
+      "loss": 0.6732,
+      "step": 4936
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.404357085893389e-06,
+      "loss": 0.6763,
+      "step": 4937
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.400559962014955e-06,
+      "loss": 0.6803,
+      "step": 4938
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3967640138533e-06,
+      "loss": 0.6667,
+      "step": 4939
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.392969242205464e-06,
+      "loss": 0.665,
+      "step": 4940
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.389175647868245e-06,
+      "loss": 0.5876,
+      "step": 4941
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.385383231638174e-06,
+      "loss": 0.6602,
+      "step": 4942
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3815919943115574e-06,
+      "loss": 0.6265,
+      "step": 4943
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.377801936684434e-06,
+      "loss": 0.6308,
+      "step": 4944
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.374013059552611e-06,
+      "loss": 0.6624,
+      "step": 4945
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.370225363711633e-06,
+      "loss": 0.699,
+      "step": 4946
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.366438849956812e-06,
+      "loss": 0.6912,
+      "step": 4947
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.362653519083197e-06,
+      "loss": 0.6713,
+      "step": 4948
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3588693718856e-06,
+      "loss": 0.5572,
+      "step": 4949
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.355086409158577e-06,
+      "loss": 0.6682,
+      "step": 4950
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3513046316964395e-06,
+      "loss": 0.6812,
+      "step": 4951
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.347524040293255e-06,
+      "loss": 0.6445,
+      "step": 4952
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.343744635742824e-06,
+      "loss": 0.6263,
+      "step": 4953
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.339966418838715e-06,
+      "loss": 0.7061,
+      "step": 4954
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.336189390374247e-06,
+      "loss": 0.6791,
+      "step": 4955
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.332413551142476e-06,
+      "loss": 0.7022,
+      "step": 4956
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.328638901936223e-06,
+      "loss": 0.66,
+      "step": 4957
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.324865443548047e-06,
+      "loss": 0.6701,
+      "step": 4958
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.321093176770269e-06,
+      "loss": 0.6948,
+      "step": 4959
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.31732210239495e-06,
+      "loss": 0.7001,
+      "step": 4960
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.313552221213901e-06,
+      "loss": 0.6704,
+      "step": 4961
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3097835340186885e-06,
+      "loss": 0.6853,
+      "step": 4962
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.30601604160063e-06,
+      "loss": 0.7271,
+      "step": 4963
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.302249744750778e-06,
+      "loss": 0.5232,
+      "step": 4964
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2984846442599495e-06,
+      "loss": 0.6789,
+      "step": 4965
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.29472074091871e-06,
+      "loss": 0.6265,
+      "step": 4966
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.290958035517354e-06,
+      "loss": 0.6507,
+      "step": 4967
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.287196528845945e-06,
+      "loss": 0.6398,
+      "step": 4968
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.283436221694293e-06,
+      "loss": 0.6566,
+      "step": 4969
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.279677114851942e-06,
+      "loss": 0.6945,
+      "step": 4970
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.275919209108202e-06,
+      "loss": 0.6235,
+      "step": 4971
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.272162505252115e-06,
+      "loss": 0.6884,
+      "step": 4972
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.268407004072485e-06,
+      "loss": 0.545,
+      "step": 4973
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2646527063578504e-06,
+      "loss": 0.6415,
+      "step": 4974
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.260899612896502e-06,
+      "loss": 0.6591,
+      "step": 4975
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.257147724476479e-06,
+      "loss": 0.6736,
+      "step": 4976
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.253397041885574e-06,
+      "loss": 0.587,
+      "step": 4977
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.249647565911309e-06,
+      "loss": 0.6894,
+      "step": 4978
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.245899297340973e-06,
+      "loss": 0.6269,
+      "step": 4979
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.242152236961583e-06,
+      "loss": 0.6801,
+      "step": 4980
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2384063855599164e-06,
+      "loss": 0.6438,
+      "step": 4981
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.234661743922488e-06,
+      "loss": 0.6778,
+      "step": 4982
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.230918312835566e-06,
+      "loss": 0.5307,
+      "step": 4983
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.227176093085153e-06,
+      "loss": 0.6876,
+      "step": 4984
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.223435085457012e-06,
+      "loss": 0.5691,
+      "step": 4985
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2196952907366375e-06,
+      "loss": 0.6454,
+      "step": 4986
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.215956709709279e-06,
+      "loss": 0.7302,
+      "step": 4987
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2122193431599356e-06,
+      "loss": 0.6551,
+      "step": 4988
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.208483191873328e-06,
+      "loss": 0.6753,
+      "step": 4989
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.204748256633946e-06,
+      "loss": 0.6628,
+      "step": 4990
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.201014538226019e-06,
+      "loss": 0.5669,
+      "step": 4991
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1972820374335085e-06,
+      "loss": 0.6547,
+      "step": 4992
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.19355075504014e-06,
+      "loss": 0.6672,
+      "step": 4993
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.189820691829361e-06,
+      "loss": 0.6947,
+      "step": 4994
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.186091848584385e-06,
+      "loss": 0.6601,
+      "step": 4995
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.182364226088153e-06,
+      "loss": 0.6574,
+      "step": 4996
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.178637825123353e-06,
+      "loss": 0.5663,
+      "step": 4997
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1749126464724224e-06,
+      "loss": 0.6816,
+      "step": 4998
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.171188690917541e-06,
+      "loss": 0.6749,
+      "step": 4999
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1674659592406245e-06,
+      "loss": 0.647,
+      "step": 5000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.163744452223338e-06,
+      "loss": 0.5998,
+      "step": 5001
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.160024170647097e-06,
+      "loss": 0.5498,
+      "step": 5002
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.156305115293033e-06,
+      "loss": 0.7166,
+      "step": 5003
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.152587286942048e-06,
+      "loss": 0.5813,
+      "step": 5004
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.148870686374778e-06,
+      "loss": 0.6672,
+      "step": 5005
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1451553143715925e-06,
+      "loss": 0.683,
+      "step": 5006
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.141441171712618e-06,
+      "loss": 0.6785,
+      "step": 5007
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1377282591777045e-06,
+      "loss": 0.6698,
+      "step": 5008
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.134016577546463e-06,
+      "loss": 0.7066,
+      "step": 5009
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.130306127598234e-06,
+      "loss": 0.5679,
+      "step": 5010
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.126596910112098e-06,
+      "loss": 0.5624,
+      "step": 5011
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.122888925866883e-06,
+      "loss": 0.6623,
+      "step": 5012
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.11918217564116e-06,
+      "loss": 0.6934,
+      "step": 5013
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1154766602132315e-06,
+      "loss": 0.7008,
+      "step": 5014
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.111772380361149e-06,
+      "loss": 0.6438,
+      "step": 5015
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1080693368627075e-06,
+      "loss": 0.6981,
+      "step": 5016
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.104367530495424e-06,
+      "loss": 0.6639,
+      "step": 5017
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1006669620365755e-06,
+      "loss": 0.5458,
+      "step": 5018
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.096967632263174e-06,
+      "loss": 0.7093,
+      "step": 5019
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.093269541951964e-06,
+      "loss": 0.6898,
+      "step": 5020
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.089572691879441e-06,
+      "loss": 0.6796,
+      "step": 5021
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.085877082821827e-06,
+      "loss": 0.6263,
+      "step": 5022
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.082182715555098e-06,
+      "loss": 0.6416,
+      "step": 5023
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.078489590854958e-06,
+      "loss": 0.7132,
+      "step": 5024
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0747977094968505e-06,
+      "loss": 0.703,
+      "step": 5025
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.071107072255967e-06,
+      "loss": 0.6822,
+      "step": 5026
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.067417679907232e-06,
+      "loss": 0.7023,
+      "step": 5027
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.063729533225303e-06,
+      "loss": 0.6868,
+      "step": 5028
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.060042632984591e-06,
+      "loss": 0.6576,
+      "step": 5029
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.056356979959227e-06,
+      "loss": 0.7107,
+      "step": 5030
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0526725749230965e-06,
+      "loss": 0.6642,
+      "step": 5031
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.048989418649808e-06,
+      "loss": 0.6667,
+      "step": 5032
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.045307511912723e-06,
+      "loss": 0.6836,
+      "step": 5033
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.041626855484926e-06,
+      "loss": 0.6648,
+      "step": 5034
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.037947450139252e-06,
+      "loss": 0.6528,
+      "step": 5035
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.034269296648262e-06,
+      "loss": 0.5621,
+      "step": 5036
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0305923957842615e-06,
+      "loss": 0.6517,
+      "step": 5037
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0269167483193e-06,
+      "loss": 0.6807,
+      "step": 5038
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.023242355025138e-06,
+      "loss": 0.6177,
+      "step": 5039
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.019569216673297e-06,
+      "loss": 0.6976,
+      "step": 5040
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.015897334035033e-06,
+      "loss": 0.6438,
+      "step": 5041
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.012226707881324e-06,
+      "loss": 0.706,
+      "step": 5042
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.008557338982899e-06,
+      "loss": 0.6938,
+      "step": 5043
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.004889228110212e-06,
+      "loss": 0.6623,
+      "step": 5044
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.001222376033464e-06,
+      "loss": 0.7003,
+      "step": 5045
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.997556783522583e-06,
+      "loss": 0.7148,
+      "step": 5046
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.993892451347229e-06,
+      "loss": 0.6725,
+      "step": 5047
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9902293802768085e-06,
+      "loss": 0.6842,
+      "step": 5048
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.986567571080462e-06,
+      "loss": 0.6419,
+      "step": 5049
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.982907024527053e-06,
+      "loss": 0.6455,
+      "step": 5050
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.979247741385193e-06,
+      "loss": 0.6936,
+      "step": 5051
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.975589722423228e-06,
+      "loss": 0.6638,
+      "step": 5052
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.971932968409221e-06,
+      "loss": 0.6584,
+      "step": 5053
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.96827748011099e-06,
+      "loss": 0.6767,
+      "step": 5054
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.964623258296082e-06,
+      "loss": 0.7106,
+      "step": 5055
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.960970303731768e-06,
+      "loss": 0.6578,
+      "step": 5056
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9573186171850685e-06,
+      "loss": 0.5984,
+      "step": 5057
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.953668199422722e-06,
+      "loss": 0.6565,
+      "step": 5058
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9500190512112145e-06,
+      "loss": 0.6754,
+      "step": 5059
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9463711733167585e-06,
+      "loss": 0.6313,
+      "step": 5060
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.942724566505294e-06,
+      "loss": 0.6337,
+      "step": 5061
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.939079231542505e-06,
+      "loss": 0.6566,
+      "step": 5062
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.935435169193808e-06,
+      "loss": 0.6542,
+      "step": 5063
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.931792380224341e-06,
+      "loss": 0.6524,
+      "step": 5064
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.928150865398989e-06,
+      "loss": 0.6832,
+      "step": 5065
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.924510625482358e-06,
+      "loss": 0.6648,
+      "step": 5066
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.92087166123879e-06,
+      "loss": 0.555,
+      "step": 5067
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.917233973432359e-06,
+      "loss": 0.6417,
+      "step": 5068
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.91359756282688e-06,
+      "loss": 0.7058,
+      "step": 5069
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.909962430185881e-06,
+      "loss": 0.6561,
+      "step": 5070
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.90632857627264e-06,
+      "loss": 0.6797,
+      "step": 5071
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9026960018501535e-06,
+      "loss": 0.6588,
+      "step": 5072
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8990647076811615e-06,
+      "loss": 0.6678,
+      "step": 5073
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.895434694528123e-06,
+      "loss": 0.6974,
+      "step": 5074
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.891805963153231e-06,
+      "loss": 0.6672,
+      "step": 5075
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.888178514318416e-06,
+      "loss": 0.5944,
+      "step": 5076
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.884552348785338e-06,
+      "loss": 0.569,
+      "step": 5077
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.880927467315376e-06,
+      "loss": 0.6862,
+      "step": 5078
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.877303870669657e-06,
+      "loss": 0.7023,
+      "step": 5079
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8736815596090205e-06,
+      "loss": 0.6344,
+      "step": 5080
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.870060534894052e-06,
+      "loss": 0.6859,
+      "step": 5081
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.866440797285058e-06,
+      "loss": 0.7125,
+      "step": 5082
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.86282234754207e-06,
+      "loss": 0.6672,
+      "step": 5083
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.859205186424862e-06,
+      "loss": 0.7549,
+      "step": 5084
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.855589314692931e-06,
+      "loss": 0.6387,
+      "step": 5085
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.851974733105498e-06,
+      "loss": 0.6584,
+      "step": 5086
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.848361442421521e-06,
+      "loss": 0.7071,
+      "step": 5087
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8447494433996925e-06,
+      "loss": 0.6975,
+      "step": 5088
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.84113873679841e-06,
+      "loss": 0.6233,
+      "step": 5089
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8375293233758224e-06,
+      "loss": 0.6829,
+      "step": 5090
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8339212038898045e-06,
+      "loss": 0.6706,
+      "step": 5091
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.830314379097946e-06,
+      "loss": 0.5896,
+      "step": 5092
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.826708849757581e-06,
+      "loss": 0.6708,
+      "step": 5093
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.823104616625758e-06,
+      "loss": 0.7016,
+      "step": 5094
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.819501680459264e-06,
+      "loss": 0.6387,
+      "step": 5095
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.815900042014607e-06,
+      "loss": 0.6943,
+      "step": 5096
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8122997020480224e-06,
+      "loss": 0.7154,
+      "step": 5097
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8087006613154764e-06,
+      "loss": 0.5687,
+      "step": 5098
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8051029205726652e-06,
+      "loss": 0.635,
+      "step": 5099
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8015064805750025e-06,
+      "loss": 0.6475,
+      "step": 5100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.797911342077638e-06,
+      "loss": 0.637,
+      "step": 5101
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7943175058354444e-06,
+      "loss": 0.6774,
+      "step": 5102
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.790724972603015e-06,
+      "loss": 0.6838,
+      "step": 5103
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.787133743134681e-06,
+      "loss": 0.6987,
+      "step": 5104
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7835438181844963e-06,
+      "loss": 0.687,
+      "step": 5105
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7799551985062323e-06,
+      "loss": 0.6476,
+      "step": 5106
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7763678848534e-06,
+      "loss": 0.6083,
+      "step": 5107
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7727818779792234e-06,
+      "loss": 0.6955,
+      "step": 5108
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7691971786366643e-06,
+      "loss": 0.6713,
+      "step": 5109
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7656137875783983e-06,
+      "loss": 0.5704,
+      "step": 5110
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.76203170555683e-06,
+      "loss": 0.6927,
+      "step": 5111
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7584509333240937e-06,
+      "loss": 0.6794,
+      "step": 5112
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7548714716320477e-06,
+      "loss": 0.6516,
+      "step": 5113
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.751293321232269e-06,
+      "loss": 0.6526,
+      "step": 5114
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.747716482876068e-06,
+      "loss": 0.6668,
+      "step": 5115
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7441409573144725e-06,
+      "loss": 0.7044,
+      "step": 5116
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.740566745298233e-06,
+      "loss": 0.6777,
+      "step": 5117
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7369938475778355e-06,
+      "loss": 0.6326,
+      "step": 5118
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.733422264903476e-06,
+      "loss": 0.625,
+      "step": 5119
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7298519980250847e-06,
+      "loss": 0.7299,
+      "step": 5120
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.726283047692315e-06,
+      "loss": 0.6623,
+      "step": 5121
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7227154146545343e-06,
+      "loss": 0.6675,
+      "step": 5122
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7191490996608483e-06,
+      "loss": 0.6997,
+      "step": 5123
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7155841034600715e-06,
+      "loss": 0.6425,
+      "step": 5124
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7120204268007464e-06,
+      "loss": 0.6187,
+      "step": 5125
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.708458070431142e-06,
+      "loss": 0.6835,
+      "step": 5126
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7048970350992518e-06,
+      "loss": 0.6311,
+      "step": 5127
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7013373215527815e-06,
+      "loss": 0.6731,
+      "step": 5128
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6977789305391708e-06,
+      "loss": 0.7001,
+      "step": 5129
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6942218628055716e-06,
+      "loss": 0.682,
+      "step": 5130
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.690666119098868e-06,
+      "loss": 0.5721,
+      "step": 5131
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6871117001656596e-06,
+      "loss": 0.6513,
+      "step": 5132
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6835586067522645e-06,
+      "loss": 0.675,
+      "step": 5133
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6800068396047293e-06,
+      "loss": 0.6484,
+      "step": 5134
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.676456399468826e-06,
+      "loss": 0.5699,
+      "step": 5135
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6729072870900352e-06,
+      "loss": 0.7178,
+      "step": 5136
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6693595032135664e-06,
+      "loss": 0.667,
+      "step": 5137
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.665813048584359e-06,
+      "loss": 0.6445,
+      "step": 5138
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6622679239470483e-06,
+      "loss": 0.6461,
+      "step": 5139
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.658724130046012e-06,
+      "loss": 0.6551,
+      "step": 5140
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.655181667625346e-06,
+      "loss": 0.575,
+      "step": 5141
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6516405374288565e-06,
+      "loss": 0.6573,
+      "step": 5142
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6481007402000823e-06,
+      "loss": 0.6661,
+      "step": 5143
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6445622766822687e-06,
+      "loss": 0.6834,
+      "step": 5144
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6410251476183965e-06,
+      "loss": 0.6278,
+      "step": 5145
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6374893537511536e-06,
+      "loss": 0.665,
+      "step": 5146
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.63395489582295e-06,
+      "loss": 0.6726,
+      "step": 5147
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6304217745759185e-06,
+      "loss": 0.6236,
+      "step": 5148
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.626889990751916e-06,
+      "loss": 0.6872,
+      "step": 5149
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.623359545092504e-06,
+      "loss": 0.6,
+      "step": 5150
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6198304383389806e-06,
+      "loss": 0.6576,
+      "step": 5151
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.616302671232348e-06,
+      "loss": 0.5528,
+      "step": 5152
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.612776244513331e-06,
+      "loss": 0.6107,
+      "step": 5153
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6092511589223765e-06,
+      "loss": 0.696,
+      "step": 5154
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6057274151996546e-06,
+      "loss": 0.5845,
+      "step": 5155
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6022050140850385e-06,
+      "loss": 0.7045,
+      "step": 5156
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.598683956318135e-06,
+      "loss": 0.7179,
+      "step": 5157
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.595164242638256e-06,
+      "loss": 0.5585,
+      "step": 5158
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.591645873784443e-06,
+      "loss": 0.7111,
+      "step": 5159
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.588128850495448e-06,
+      "loss": 0.6609,
+      "step": 5160
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5846131735097366e-06,
+      "loss": 0.597,
+      "step": 5161
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5810988435655004e-06,
+      "loss": 0.6687,
+      "step": 5162
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5775858614006497e-06,
+      "loss": 0.6699,
+      "step": 5163
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5740742277527984e-06,
+      "loss": 0.6638,
+      "step": 5164
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.570563943359293e-06,
+      "loss": 0.6927,
+      "step": 5165
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5670550089571866e-06,
+      "loss": 0.6794,
+      "step": 5166
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.563547425283248e-06,
+      "loss": 0.507,
+      "step": 5167
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.560041193073972e-06,
+      "loss": 0.6175,
+      "step": 5168
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.556536313065557e-06,
+      "loss": 0.6794,
+      "step": 5169
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5530327859939296e-06,
+      "loss": 0.7088,
+      "step": 5170
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.549530612594727e-06,
+      "loss": 0.6672,
+      "step": 5171
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.546029793603297e-06,
+      "loss": 0.7142,
+      "step": 5172
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5425303297547163e-06,
+      "loss": 0.6576,
+      "step": 5173
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5390322217837623e-06,
+      "loss": 0.6839,
+      "step": 5174
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5355354704249336e-06,
+      "loss": 0.5402,
+      "step": 5175
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5320400764124452e-06,
+      "loss": 0.7231,
+      "step": 5176
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.528546040480232e-06,
+      "loss": 0.621,
+      "step": 5177
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.525053363361931e-06,
+      "loss": 0.6492,
+      "step": 5178
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.521562045790907e-06,
+      "loss": 0.5348,
+      "step": 5179
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5180720885002274e-06,
+      "loss": 0.691,
+      "step": 5180
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5145834922226875e-06,
+      "loss": 0.6679,
+      "step": 5181
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5110962576907838e-06,
+      "loss": 0.6502,
+      "step": 5182
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5076103856367303e-06,
+      "loss": 0.6648,
+      "step": 5183
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5041258767924615e-06,
+      "loss": 0.6754,
+      "step": 5184
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.500642731889623e-06,
+      "loss": 0.6882,
+      "step": 5185
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.497160951659566e-06,
+      "loss": 0.6312,
+      "step": 5186
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.493680536833367e-06,
+      "loss": 0.6662,
+      "step": 5187
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.490201488141809e-06,
+      "loss": 0.5842,
+      "step": 5188
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.486723806315384e-06,
+      "loss": 0.6846,
+      "step": 5189
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.483247492084307e-06,
+      "loss": 0.6717,
+      "step": 5190
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.479772546178505e-06,
+      "loss": 0.6706,
+      "step": 5191
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.476298969327604e-06,
+      "loss": 0.572,
+      "step": 5192
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4728267622609623e-06,
+      "loss": 0.5971,
+      "step": 5193
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4693559257076324e-06,
+      "loss": 0.6019,
+      "step": 5194
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.465886460396395e-06,
+      "loss": 0.65,
+      "step": 5195
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.462418367055732e-06,
+      "loss": 0.5682,
+      "step": 5196
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.458951646413835e-06,
+      "loss": 0.549,
+      "step": 5197
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4554862991986184e-06,
+      "loss": 0.672,
+      "step": 5198
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.452022326137705e-06,
+      "loss": 0.6988,
+      "step": 5199
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4485597279584203e-06,
+      "loss": 0.6768,
+      "step": 5200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.445098505387814e-06,
+      "loss": 0.645,
+      "step": 5201
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.441638659152636e-06,
+      "loss": 0.69,
+      "step": 5202
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.438180189979351e-06,
+      "loss": 0.5882,
+      "step": 5203
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4347230985941395e-06,
+      "loss": 0.6762,
+      "step": 5204
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4312673857228827e-06,
+      "loss": 0.5797,
+      "step": 5205
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.427813052091182e-06,
+      "loss": 0.6703,
+      "step": 5206
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4243600984243475e-06,
+      "loss": 0.6547,
+      "step": 5207
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.420908525447392e-06,
+      "loss": 0.6352,
+      "step": 5208
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4174583338850476e-06,
+      "loss": 0.6876,
+      "step": 5209
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4140095244617533e-06,
+      "loss": 0.666,
+      "step": 5210
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4105620979016497e-06,
+      "loss": 0.6787,
+      "step": 5211
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.407116054928601e-06,
+      "loss": 0.6234,
+      "step": 5212
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4036713962661747e-06,
+      "loss": 0.6802,
+      "step": 5213
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4002281226376432e-06,
+      "loss": 0.6795,
+      "step": 5214
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3967862347659974e-06,
+      "loss": 0.6055,
+      "step": 5215
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.393345733373928e-06,
+      "loss": 0.5824,
+      "step": 5216
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3899066191838382e-06,
+      "loss": 0.6355,
+      "step": 5217
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3864688929178423e-06,
+      "loss": 0.5652,
+      "step": 5218
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3830325552977584e-06,
+      "loss": 0.6581,
+      "step": 5219
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3795976070451173e-06,
+      "loss": 0.6709,
+      "step": 5220
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3761640488811607e-06,
+      "loss": 0.6665,
+      "step": 5221
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3727318815268264e-06,
+      "loss": 0.6731,
+      "step": 5222
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3693011057027758e-06,
+      "loss": 0.6642,
+      "step": 5223
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3658717221293656e-06,
+      "loss": 0.5949,
+      "step": 5224
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.362443731526663e-06,
+      "loss": 0.6784,
+      "step": 5225
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3590171346144483e-06,
+      "loss": 0.5567,
+      "step": 5226
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.355591932112208e-06,
+      "loss": 0.7024,
+      "step": 5227
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3521681247391267e-06,
+      "loss": 0.6847,
+      "step": 5228
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3487457132141087e-06,
+      "loss": 0.6617,
+      "step": 5229
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.345324698255753e-06,
+      "loss": 0.6873,
+      "step": 5230
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3419050805823796e-06,
+      "loss": 0.6177,
+      "step": 5231
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3384868609120013e-06,
+      "loss": 0.6356,
+      "step": 5232
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3350700399623416e-06,
+      "loss": 0.6832,
+      "step": 5233
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3316546184508347e-06,
+      "loss": 0.674,
+      "step": 5234
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3282405970946207e-06,
+      "loss": 0.6645,
+      "step": 5235
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3248279766105386e-06,
+      "loss": 0.6515,
+      "step": 5236
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3214167577151412e-06,
+      "loss": 0.6894,
+      "step": 5237
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3180069411246827e-06,
+      "loss": 0.6398,
+      "step": 5238
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3145985275551194e-06,
+      "loss": 0.5693,
+      "step": 5239
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.311191517722124e-06,
+      "loss": 0.6378,
+      "step": 5240
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3077859123410617e-06,
+      "loss": 0.6397,
+      "step": 5241
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3043817121270115e-06,
+      "loss": 0.5791,
+      "step": 5242
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.300978917794758e-06,
+      "loss": 0.7,
+      "step": 5243
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.297577530058782e-06,
+      "loss": 0.5836,
+      "step": 5244
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2941775496332807e-06,
+      "loss": 0.6691,
+      "step": 5245
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.290778977232144e-06,
+      "loss": 0.6926,
+      "step": 5246
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2873818135689707e-06,
+      "loss": 0.6003,
+      "step": 5247
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2839860593570673e-06,
+      "loss": 0.6288,
+      "step": 5248
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2805917153094445e-06,
+      "loss": 0.5659,
+      "step": 5249
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2771987821388073e-06,
+      "loss": 0.6605,
+      "step": 5250
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2738072605575797e-06,
+      "loss": 0.6401,
+      "step": 5251
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.270417151277876e-06,
+      "loss": 0.5734,
+      "step": 5252
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.267028455011515e-06,
+      "loss": 0.6172,
+      "step": 5253
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.263641172470031e-06,
+      "loss": 0.6677,
+      "step": 5254
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2602553043646456e-06,
+      "loss": 0.6075,
+      "step": 5255
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2568708514062954e-06,
+      "loss": 0.5994,
+      "step": 5256
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.253487814305617e-06,
+      "loss": 0.6929,
+      "step": 5257
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2501061937729416e-06,
+      "loss": 0.6393,
+      "step": 5258
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.246725990518318e-06,
+      "loss": 0.7177,
+      "step": 5259
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2433472052514836e-06,
+      "loss": 0.6693,
+      "step": 5260
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2399698386818814e-06,
+      "loss": 0.6985,
+      "step": 5261
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.236593891518661e-06,
+      "loss": 0.5357,
+      "step": 5262
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2332193644706744e-06,
+      "loss": 0.6672,
+      "step": 5263
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2298462582464674e-06,
+      "loss": 0.6549,
+      "step": 5264
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.226474573554298e-06,
+      "loss": 0.5543,
+      "step": 5265
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2231043111021164e-06,
+      "loss": 0.6568,
+      "step": 5266
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.219735471597576e-06,
+      "loss": 0.6945,
+      "step": 5267
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.216368055748039e-06,
+      "loss": 0.6908,
+      "step": 5268
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2130020642605573e-06,
+      "loss": 0.5713,
+      "step": 5269
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2096374978418933e-06,
+      "loss": 0.6452,
+      "step": 5270
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2062743571985087e-06,
+      "loss": 0.6713,
+      "step": 5271
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2029126430365576e-06,
+      "loss": 0.5559,
+      "step": 5272
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1995523560619056e-06,
+      "loss": 0.7292,
+      "step": 5273
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.196193496980112e-06,
+      "loss": 0.6538,
+      "step": 5274
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1928360664964353e-06,
+      "loss": 0.582,
+      "step": 5275
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.189480065315841e-06,
+      "loss": 0.6419,
+      "step": 5276
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.186125494142984e-06,
+      "loss": 0.6782,
+      "step": 5277
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1827723536822285e-06,
+      "loss": 0.6944,
+      "step": 5278
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1794206446376396e-06,
+      "loss": 0.6144,
+      "step": 5279
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.176070367712969e-06,
+      "loss": 0.5838,
+      "step": 5280
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1727215236116814e-06,
+      "loss": 0.5461,
+      "step": 5281
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1693741130369337e-06,
+      "loss": 0.5716,
+      "step": 5282
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.166028136691579e-06,
+      "loss": 0.634,
+      "step": 5283
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1626835952781763e-06,
+      "loss": 0.6251,
+      "step": 5284
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1593404894989856e-06,
+      "loss": 0.7019,
+      "step": 5285
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.15599882005595e-06,
+      "loss": 0.5726,
+      "step": 5286
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.152658587650731e-06,
+      "loss": 0.6254,
+      "step": 5287
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.149319792984674e-06,
+      "loss": 0.6939,
+      "step": 5288
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1459824367588245e-06,
+      "loss": 0.6942,
+      "step": 5289
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1426465196739342e-06,
+      "loss": 0.6717,
+      "step": 5290
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.13931204243044e-06,
+      "loss": 0.6588,
+      "step": 5291
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.135979005728489e-06,
+      "loss": 0.6602,
+      "step": 5292
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1326474102679226e-06,
+      "loss": 0.6744,
+      "step": 5293
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.12931725674827e-06,
+      "loss": 0.5714,
+      "step": 5294
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1259885458687713e-06,
+      "loss": 0.692,
+      "step": 5295
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1226612783283563e-06,
+      "loss": 0.5689,
+      "step": 5296
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1193354548256462e-06,
+      "loss": 0.6507,
+      "step": 5297
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.116011076058971e-06,
+      "loss": 0.6629,
+      "step": 5298
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1126881427263556e-06,
+      "loss": 0.5673,
+      "step": 5299
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.10936665552551e-06,
+      "loss": 0.5439,
+      "step": 5300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.106046615153855e-06,
+      "loss": 0.6692,
+      "step": 5301
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1027280223084968e-06,
+      "loss": 0.6493,
+      "step": 5302
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.099410877686241e-06,
+      "loss": 0.5408,
+      "step": 5303
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.096095181983594e-06,
+      "loss": 0.694,
+      "step": 5304
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0927809358967477e-06,
+      "loss": 0.627,
+      "step": 5305
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0894681401216e-06,
+      "loss": 0.6154,
+      "step": 5306
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.086156795353742e-06,
+      "loss": 0.6818,
+      "step": 5307
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.082846902288452e-06,
+      "loss": 0.7185,
+      "step": 5308
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.079538461620718e-06,
+      "loss": 0.6594,
+      "step": 5309
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.076231474045208e-06,
+      "loss": 0.688,
+      "step": 5310
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0729259402562918e-06,
+      "loss": 0.7112,
+      "step": 5311
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0696218609480356e-06,
+      "loss": 0.6533,
+      "step": 5312
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0663192368142003e-06,
+      "loss": 0.6661,
+      "step": 5313
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0630180685482337e-06,
+      "loss": 0.5481,
+      "step": 5314
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.059718356843291e-06,
+      "loss": 0.6288,
+      "step": 5315
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0564201023922092e-06,
+      "loss": 0.6385,
+      "step": 5316
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.053123305887522e-06,
+      "loss": 0.6916,
+      "step": 5317
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.049827968021466e-06,
+      "loss": 0.6541,
+      "step": 5318
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.046534089485956e-06,
+      "loss": 0.6845,
+      "step": 5319
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0432416709726154e-06,
+      "loss": 0.6655,
+      "step": 5320
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0399507131727557e-06,
+      "loss": 0.6836,
+      "step": 5321
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0366612167773747e-06,
+      "loss": 0.689,
+      "step": 5322
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.033373182477176e-06,
+      "loss": 0.7013,
+      "step": 5323
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.030086610962546e-06,
+      "loss": 0.5731,
+      "step": 5324
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0268015029235655e-06,
+      "loss": 0.6791,
+      "step": 5325
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0235178590500148e-06,
+      "loss": 0.6825,
+      "step": 5326
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.020235680031357e-06,
+      "loss": 0.6936,
+      "step": 5327
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0169549665567543e-06,
+      "loss": 0.5538,
+      "step": 5328
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0136757193150634e-06,
+      "loss": 0.6821,
+      "step": 5329
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0103979389948236e-06,
+      "loss": 0.6721,
+      "step": 5330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0071216262842772e-06,
+      "loss": 0.6507,
+      "step": 5331
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0038467818713514e-06,
+      "loss": 0.6844,
+      "step": 5332
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.000573406443661e-06,
+      "loss": 0.6568,
+      "step": 5333
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.997301500688522e-06,
+      "loss": 0.5661,
+      "step": 5334
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.994031065292943e-06,
+      "loss": 0.7028,
+      "step": 5335
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9907621009436095e-06,
+      "loss": 0.6643,
+      "step": 5336
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.987494608326915e-06,
+      "loss": 0.6473,
+      "step": 5337
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9842285881289333e-06,
+      "loss": 0.7048,
+      "step": 5338
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.980964041035428e-06,
+      "loss": 0.7143,
+      "step": 5339
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9777009677318645e-06,
+      "loss": 0.5886,
+      "step": 5340
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.974439368903387e-06,
+      "loss": 0.6768,
+      "step": 5341
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9711792452348344e-06,
+      "loss": 0.7246,
+      "step": 5342
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9679205974107416e-06,
+      "loss": 0.6427,
+      "step": 5343
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9646634261153217e-06,
+      "loss": 0.7137,
+      "step": 5344
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9614077320324895e-06,
+      "loss": 0.6713,
+      "step": 5345
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.958153515845844e-06,
+      "loss": 0.5765,
+      "step": 5346
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9549007782386684e-06,
+      "loss": 0.531,
+      "step": 5347
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9516495198939454e-06,
+      "loss": 0.5785,
+      "step": 5348
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9483997414943465e-06,
+      "loss": 0.5655,
+      "step": 5349
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.945151443722224e-06,
+      "loss": 0.6737,
+      "step": 5350
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9419046272596276e-06,
+      "loss": 0.6581,
+      "step": 5351
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.938659292788293e-06,
+      "loss": 0.6556,
+      "step": 5352
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9354154409896386e-06,
+      "loss": 0.6466,
+      "step": 5353
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.932173072544784e-06,
+      "loss": 0.6846,
+      "step": 5354
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9289321881345257e-06,
+      "loss": 0.6734,
+      "step": 5355
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.925692788439356e-06,
+      "loss": 0.6683,
+      "step": 5356
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9224548741394563e-06,
+      "loss": 0.6479,
+      "step": 5357
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9192184459146857e-06,
+      "loss": 0.6909,
+      "step": 5358
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.915983504444606e-06,
+      "loss": 0.6685,
+      "step": 5359
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.912750050408456e-06,
+      "loss": 0.673,
+      "step": 5360
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.909518084485161e-06,
+      "loss": 0.7063,
+      "step": 5361
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9062876073533454e-06,
+      "loss": 0.6689,
+      "step": 5362
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.903058619691307e-06,
+      "loss": 0.6638,
+      "step": 5363
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8998311221770415e-06,
+      "loss": 0.6631,
+      "step": 5364
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8966051154882314e-06,
+      "loss": 0.6883,
+      "step": 5365
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.893380600302239e-06,
+      "loss": 0.6694,
+      "step": 5366
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8901575772961134e-06,
+      "loss": 0.5545,
+      "step": 5367
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8869360471466013e-06,
+      "loss": 0.6583,
+      "step": 5368
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.883716010530122e-06,
+      "loss": 0.6713,
+      "step": 5369
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8804974681227905e-06,
+      "loss": 0.6188,
+      "step": 5370
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.87728042060041e-06,
+      "loss": 0.6547,
+      "step": 5371
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.874064868638459e-06,
+      "loss": 0.6868,
+      "step": 5372
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.870850812912113e-06,
+      "loss": 0.6004,
+      "step": 5373
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8676382540962267e-06,
+      "loss": 0.6941,
+      "step": 5374
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.864427192865339e-06,
+      "loss": 0.6673,
+      "step": 5375
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8612176298936834e-06,
+      "loss": 0.6735,
+      "step": 5376
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8580095658551686e-06,
+      "loss": 0.677,
+      "step": 5377
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8548030014233953e-06,
+      "loss": 0.6957,
+      "step": 5378
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8515979372716494e-06,
+      "loss": 0.7037,
+      "step": 5379
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.848394374072898e-06,
+      "loss": 0.6661,
+      "step": 5380
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8451923124997917e-06,
+      "loss": 0.6766,
+      "step": 5381
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.841991753224673e-06,
+      "loss": 0.6279,
+      "step": 5382
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.83879269691956e-06,
+      "loss": 0.6747,
+      "step": 5383
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8355951442561624e-06,
+      "loss": 0.6548,
+      "step": 5384
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8323990959058746e-06,
+      "loss": 0.6305,
+      "step": 5385
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.829204552539766e-06,
+      "loss": 0.6494,
+      "step": 5386
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.826011514828604e-06,
+      "loss": 0.6189,
+      "step": 5387
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.822819983442827e-06,
+      "loss": 0.69,
+      "step": 5388
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.81962995905256e-06,
+      "loss": 0.6775,
+      "step": 5389
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8164414423276197e-06,
+      "loss": 0.6117,
+      "step": 5390
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8132544339374944e-06,
+      "loss": 0.5832,
+      "step": 5391
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8100689345513653e-06,
+      "loss": 0.6519,
+      "step": 5392
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.806884944838094e-06,
+      "loss": 0.6836,
+      "step": 5393
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8037024654662204e-06,
+      "loss": 0.6824,
+      "step": 5394
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.800521497103975e-06,
+      "loss": 0.6451,
+      "step": 5395
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7973420404192664e-06,
+      "loss": 0.6568,
+      "step": 5396
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7941640960796813e-06,
+      "loss": 0.6559,
+      "step": 5397
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7909876647525013e-06,
+      "loss": 0.6634,
+      "step": 5398
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7878127471046755e-06,
+      "loss": 0.5968,
+      "step": 5399
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7846393438028484e-06,
+      "loss": 0.6959,
+      "step": 5400
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7814674555133415e-06,
+      "loss": 0.6365,
+      "step": 5401
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.778297082902155e-06,
+      "loss": 0.6724,
+      "step": 5402
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.775128226634971e-06,
+      "loss": 0.6269,
+      "step": 5403
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.771960887377162e-06,
+      "loss": 0.6483,
+      "step": 5404
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.768795065793768e-06,
+      "loss": 0.6138,
+      "step": 5405
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7656307625495218e-06,
+      "loss": 0.6916,
+      "step": 5406
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7624679783088357e-06,
+      "loss": 0.6599,
+      "step": 5407
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7593067137357964e-06,
+      "loss": 0.6561,
+      "step": 5408
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.75614696949418e-06,
+      "loss": 0.6865,
+      "step": 5409
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.752988746247438e-06,
+      "loss": 0.5717,
+      "step": 5410
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.749832044658701e-06,
+      "loss": 0.6446,
+      "step": 5411
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7466768653907873e-06,
+      "loss": 0.6552,
+      "step": 5412
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.743523209106187e-06,
+      "loss": 0.697,
+      "step": 5413
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7403710764670753e-06,
+      "loss": 0.6666,
+      "step": 5414
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.737220468135312e-06,
+      "loss": 0.5672,
+      "step": 5415
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7340713847724286e-06,
+      "loss": 0.5249,
+      "step": 5416
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7309238270396353e-06,
+      "loss": 0.618,
+      "step": 5417
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7277777955978324e-06,
+      "loss": 0.7289,
+      "step": 5418
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.724633291107588e-06,
+      "loss": 0.671,
+      "step": 5419
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7214903142291583e-06,
+      "loss": 0.7041,
+      "step": 5420
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7183488656224765e-06,
+      "loss": 0.6843,
+      "step": 5421
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.715208945947151e-06,
+      "loss": 0.6307,
+      "step": 5422
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.712070555862475e-06,
+      "loss": 0.6807,
+      "step": 5423
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7089336960274158e-06,
+      "loss": 0.5654,
+      "step": 5424
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7057983671006196e-06,
+      "loss": 0.6497,
+      "step": 5425
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.702664569740416e-06,
+      "loss": 0.6521,
+      "step": 5426
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.699532304604806e-06,
+      "loss": 0.5879,
+      "step": 5427
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.696401572351476e-06,
+      "loss": 0.6286,
+      "step": 5428
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6932723736377876e-06,
+      "loss": 0.6999,
+      "step": 5429
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6901447091207777e-06,
+      "loss": 0.6723,
+      "step": 5430
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6870185794571623e-06,
+      "loss": 0.7007,
+      "step": 5431
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6838939853033406e-06,
+      "loss": 0.6532,
+      "step": 5432
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.680770927315379e-06,
+      "loss": 0.6835,
+      "step": 5433
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.67764940614903e-06,
+      "loss": 0.5787,
+      "step": 5434
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.674529422459725e-06,
+      "loss": 0.5452,
+      "step": 5435
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.671410976902562e-06,
+      "loss": 0.5803,
+      "step": 5436
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6682940701323267e-06,
+      "loss": 0.5798,
+      "step": 5437
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.665178702803476e-06,
+      "loss": 0.6492,
+      "step": 5438
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6620648755701417e-06,
+      "loss": 0.673,
+      "step": 5439
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6589525890861413e-06,
+      "loss": 0.5726,
+      "step": 5440
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.655841844004956e-06,
+      "loss": 0.6497,
+      "step": 5441
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.652732640979754e-06,
+      "loss": 0.6273,
+      "step": 5442
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.649624980663379e-06,
+      "loss": 0.661,
+      "step": 5443
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.646518863708342e-06,
+      "loss": 0.6535,
+      "step": 5444
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6434142907668413e-06,
+      "loss": 0.6653,
+      "step": 5445
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6403112624907424e-06,
+      "loss": 0.6766,
+      "step": 5446
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6372097795315864e-06,
+      "loss": 0.6689,
+      "step": 5447
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6341098425405987e-06,
+      "loss": 0.6093,
+      "step": 5448
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.631011452168669e-06,
+      "loss": 0.5992,
+      "step": 5449
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6279146090663708e-06,
+      "loss": 0.707,
+      "step": 5450
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6248193138839516e-06,
+      "loss": 0.6526,
+      "step": 5451
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6217255672713304e-06,
+      "loss": 0.6226,
+      "step": 5452
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6186333698780975e-06,
+      "loss": 0.6334,
+      "step": 5453
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6155427223535303e-06,
+      "loss": 0.5954,
+      "step": 5454
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.612453625346567e-06,
+      "loss": 0.6419,
+      "step": 5455
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6093660795058297e-06,
+      "loss": 0.6956,
+      "step": 5456
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6062800854796134e-06,
+      "loss": 0.6454,
+      "step": 5457
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6031956439158814e-06,
+      "loss": 0.656,
+      "step": 5458
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6001127554622806e-06,
+      "loss": 0.6521,
+      "step": 5459
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.597031420766123e-06,
+      "loss": 0.6246,
+      "step": 5460
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5939516404743948e-06,
+      "loss": 0.6737,
+      "step": 5461
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5908734152337657e-06,
+      "loss": 0.7051,
+      "step": 5462
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.587796745690565e-06,
+      "loss": 0.6766,
+      "step": 5463
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5847216324908053e-06,
+      "loss": 0.6622,
+      "step": 5464
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5816480762801734e-06,
+      "loss": 0.6236,
+      "step": 5465
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5785760777040204e-06,
+      "loss": 0.6632,
+      "step": 5466
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.575505637407374e-06,
+      "loss": 0.6538,
+      "step": 5467
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5724367560349393e-06,
+      "loss": 0.6408,
+      "step": 5468
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.569369434231088e-06,
+      "loss": 0.6852,
+      "step": 5469
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5663036726398683e-06,
+      "loss": 0.6289,
+      "step": 5470
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5632394719050026e-06,
+      "loss": 0.6997,
+      "step": 5471
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.560176832669875e-06,
+      "loss": 0.6784,
+      "step": 5472
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.557115755577557e-06,
+      "loss": 0.6451,
+      "step": 5473
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5540562412707814e-06,
+      "loss": 0.654,
+      "step": 5474
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5509982903919507e-06,
+      "loss": 0.6737,
+      "step": 5475
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.547941903583152e-06,
+      "loss": 0.6958,
+      "step": 5476
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5448870814861304e-06,
+      "loss": 0.6783,
+      "step": 5477
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.541833824742309e-06,
+      "loss": 0.5789,
+      "step": 5478
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5387821339927866e-06,
+      "loss": 0.6715,
+      "step": 5479
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5357320098783233e-06,
+      "loss": 0.5873,
+      "step": 5480
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.532683453039353e-06,
+      "loss": 0.671,
+      "step": 5481
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.529636464115989e-06,
+      "loss": 0.6612,
+      "step": 5482
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.526591043748001e-06,
+      "loss": 0.6615,
+      "step": 5483
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5235471925748446e-06,
+      "loss": 0.6702,
+      "step": 5484
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5205049112356327e-06,
+      "loss": 0.6353,
+      "step": 5485
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.517464200369156e-06,
+      "loss": 0.5739,
+      "step": 5486
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5144250606138785e-06,
+      "loss": 0.5948,
+      "step": 5487
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.511387492607925e-06,
+      "loss": 0.5537,
+      "step": 5488
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5083514969890933e-06,
+      "loss": 0.6597,
+      "step": 5489
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5053170743948587e-06,
+      "loss": 0.572,
+      "step": 5490
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.502284225462354e-06,
+      "loss": 0.6558,
+      "step": 5491
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.499252950828389e-06,
+      "loss": 0.6246,
+      "step": 5492
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.496223251129447e-06,
+      "loss": 0.6463,
+      "step": 5493
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4931951270016673e-06,
+      "loss": 0.6812,
+      "step": 5494
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.490168579080875e-06,
+      "loss": 0.6633,
+      "step": 5495
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4871436080025513e-06,
+      "loss": 0.6636,
+      "step": 5496
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4841202144018463e-06,
+      "loss": 0.5993,
+      "step": 5497
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4810983989135917e-06,
+      "loss": 0.6505,
+      "step": 5498
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4780781621722715e-06,
+      "loss": 0.6719,
+      "step": 5499
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4750595048120508e-06,
+      "loss": 0.6675,
+      "step": 5500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.472042427466759e-06,
+      "loss": 0.6871,
+      "step": 5501
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.469026930769892e-06,
+      "loss": 0.6698,
+      "step": 5502
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4660130153546113e-06,
+      "loss": 0.6287,
+      "step": 5503
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4630006818537556e-06,
+      "loss": 0.6686,
+      "step": 5504
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.459989930899821e-06,
+      "loss": 0.5659,
+      "step": 5505
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4569807631249777e-06,
+      "loss": 0.6545,
+      "step": 5506
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4539731791610666e-06,
+      "loss": 0.6076,
+      "step": 5507
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4509671796395827e-06,
+      "loss": 0.6649,
+      "step": 5508
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.447962765191706e-06,
+      "loss": 0.6566,
+      "step": 5509
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.44495993644827e-06,
+      "loss": 0.6559,
+      "step": 5510
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.441958694039778e-06,
+      "loss": 0.6772,
+      "step": 5511
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4389590385964066e-06,
+      "loss": 0.6729,
+      "step": 5512
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.435960970747989e-06,
+      "loss": 0.6436,
+      "step": 5513
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4329644911240356e-06,
+      "loss": 0.621,
+      "step": 5514
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.42996960035372e-06,
+      "loss": 0.5745,
+      "step": 5515
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.426976299065876e-06,
+      "loss": 0.6828,
+      "step": 5516
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.423984587889009e-06,
+      "loss": 0.6328,
+      "step": 5517
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.420994467451294e-06,
+      "loss": 0.6489,
+      "step": 5518
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.418005938380561e-06,
+      "loss": 0.6707,
+      "step": 5519
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4150190013043207e-06,
+      "loss": 0.6949,
+      "step": 5520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4120336568497338e-06,
+      "loss": 0.6788,
+      "step": 5521
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4090499056436377e-06,
+      "loss": 0.6403,
+      "step": 5522
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.406067748312536e-06,
+      "loss": 0.6177,
+      "step": 5523
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.403087185482589e-06,
+      "loss": 0.697,
+      "step": 5524
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.400108217779624e-06,
+      "loss": 0.6731,
+      "step": 5525
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3971308458291443e-06,
+      "loss": 0.6596,
+      "step": 5526
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.394155070256301e-06,
+      "loss": 0.6743,
+      "step": 5527
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3911808916859225e-06,
+      "loss": 0.7143,
+      "step": 5528
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3882083107425037e-06,
+      "loss": 0.673,
+      "step": 5529
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3852373280501917e-06,
+      "loss": 0.6088,
+      "step": 5530
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3822679442328055e-06,
+      "loss": 0.6404,
+      "step": 5531
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3793001599138334e-06,
+      "loss": 0.5943,
+      "step": 5532
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.376333975716414e-06,
+      "loss": 0.6555,
+      "step": 5533
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.373369392263367e-06,
+      "loss": 0.6346,
+      "step": 5534
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3704064101771594e-06,
+      "loss": 0.6424,
+      "step": 5535
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3674450300799333e-06,
+      "loss": 0.6465,
+      "step": 5536
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.364485252593494e-06,
+      "loss": 0.6782,
+      "step": 5537
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.361527078339304e-06,
+      "loss": 0.6767,
+      "step": 5538
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.35857050793849e-06,
+      "loss": 0.6591,
+      "step": 5539
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.355615542011851e-06,
+      "loss": 0.7301,
+      "step": 5540
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3526621811798343e-06,
+      "loss": 0.5502,
+      "step": 5541
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3497104260625623e-06,
+      "loss": 0.7005,
+      "step": 5542
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3467602772798194e-06,
+      "loss": 0.6539,
+      "step": 5543
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3438117354510436e-06,
+      "loss": 0.6863,
+      "step": 5544
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3408648011953483e-06,
+      "loss": 0.6478,
+      "step": 5545
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.337919475131498e-06,
+      "loss": 0.5581,
+      "step": 5546
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.334975757877922e-06,
+      "loss": 0.7229,
+      "step": 5547
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3320336500527195e-06,
+      "loss": 0.6741,
+      "step": 5548
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3290931522736404e-06,
+      "loss": 0.6948,
+      "step": 5549
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.326154265158105e-06,
+      "loss": 0.6579,
+      "step": 5550
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.323216989323195e-06,
+      "loss": 0.6834,
+      "step": 5551
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3202813253856502e-06,
+      "loss": 0.6326,
+      "step": 5552
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3173472739618687e-06,
+      "loss": 0.6223,
+      "step": 5553
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.314414835667921e-06,
+      "loss": 0.6887,
+      "step": 5554
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.311484011119527e-06,
+      "loss": 0.6491,
+      "step": 5555
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3085548009320778e-06,
+      "loss": 0.6613,
+      "step": 5556
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3056272057206153e-06,
+      "loss": 0.5651,
+      "step": 5557
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.302701226099853e-06,
+      "loss": 0.5516,
+      "step": 5558
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.29977686268416e-06,
+      "loss": 0.6379,
+      "step": 5559
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2968541160875646e-06,
+      "loss": 0.6873,
+      "step": 5560
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2939329869237536e-06,
+      "loss": 0.6643,
+      "step": 5561
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2910134758060844e-06,
+      "loss": 0.6891,
+      "step": 5562
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2880955833475617e-06,
+      "loss": 0.6075,
+      "step": 5563
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.28517931016086e-06,
+      "loss": 0.6277,
+      "step": 5564
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2822646568583127e-06,
+      "loss": 0.6885,
+      "step": 5565
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2793516240519087e-06,
+      "loss": 0.6904,
+      "step": 5566
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2764402123532946e-06,
+      "loss": 0.6855,
+      "step": 5567
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2735304223737863e-06,
+      "loss": 0.682,
+      "step": 5568
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2706222547243508e-06,
+      "loss": 0.6566,
+      "step": 5569
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.267715710015619e-06,
+      "loss": 0.6227,
+      "step": 5570
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2648107888578763e-06,
+      "loss": 0.5712,
+      "step": 5571
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2619074918610717e-06,
+      "loss": 0.6323,
+      "step": 5572
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2590058196348185e-06,
+      "loss": 0.6868,
+      "step": 5573
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2561057727883705e-06,
+      "loss": 0.5713,
+      "step": 5574
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2532073519306575e-06,
+      "loss": 0.6219,
+      "step": 5575
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2503105576702635e-06,
+      "loss": 0.6866,
+      "step": 5576
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.247415390615426e-06,
+      "loss": 0.6482,
+      "step": 5577
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.244521851374045e-06,
+      "loss": 0.5923,
+      "step": 5578
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2416299405536835e-06,
+      "loss": 0.6557,
+      "step": 5579
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.238739658761553e-06,
+      "loss": 0.6251,
+      "step": 5580
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.235851006604525e-06,
+      "loss": 0.6347,
+      "step": 5581
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.232963984689135e-06,
+      "loss": 0.6602,
+      "step": 5582
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2300785936215687e-06,
+      "loss": 0.6551,
+      "step": 5583
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2271948340076764e-06,
+      "loss": 0.6465,
+      "step": 5584
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.224312706452959e-06,
+      "loss": 0.6452,
+      "step": 5585
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.22143221156258e-06,
+      "loss": 0.5846,
+      "step": 5586
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2185533499413593e-06,
+      "loss": 0.5888,
+      "step": 5587
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.215676122193773e-06,
+      "loss": 0.7052,
+      "step": 5588
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2128005289239474e-06,
+      "loss": 0.6844,
+      "step": 5589
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.209926570735681e-06,
+      "loss": 0.6198,
+      "step": 5590
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.207054248232412e-06,
+      "loss": 0.6866,
+      "step": 5591
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.204183562017248e-06,
+      "loss": 0.6781,
+      "step": 5592
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2013145126929493e-06,
+      "loss": 0.6813,
+      "step": 5593
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1984471008619257e-06,
+      "loss": 0.5887,
+      "step": 5594
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.195581327126256e-06,
+      "loss": 0.6782,
+      "step": 5595
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1927171920876644e-06,
+      "loss": 0.6761,
+      "step": 5596
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.189854696347531e-06,
+      "loss": 0.6653,
+      "step": 5597
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1869938405069014e-06,
+      "loss": 0.5772,
+      "step": 5598
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1841346251664654e-06,
+      "loss": 0.666,
+      "step": 5599
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1812770509265767e-06,
+      "loss": 0.6968,
+      "step": 5600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1784211183872426e-06,
+      "loss": 0.6764,
+      "step": 5601
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1755668281481235e-06,
+      "loss": 0.6355,
+      "step": 5602
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.172714180808533e-06,
+      "loss": 0.6873,
+      "step": 5603
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.169863176967446e-06,
+      "loss": 0.6728,
+      "step": 5604
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.167013817223487e-06,
+      "loss": 0.6976,
+      "step": 5605
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.164166102174938e-06,
+      "loss": 0.6676,
+      "step": 5606
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1613200324197347e-06,
+      "loss": 0.6421,
+      "step": 5607
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.158475608555467e-06,
+      "loss": 0.6626,
+      "step": 5608
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1556328311793837e-06,
+      "loss": 0.6357,
+      "step": 5609
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1527917008883804e-06,
+      "loss": 0.6412,
+      "step": 5610
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.149952218279009e-06,
+      "loss": 0.6534,
+      "step": 5611
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.147114383947481e-06,
+      "loss": 0.6724,
+      "step": 5612
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1442781984896522e-06,
+      "loss": 0.6403,
+      "step": 5613
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1414436625010413e-06,
+      "loss": 0.6829,
+      "step": 5614
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1386107765768196e-06,
+      "loss": 0.6258,
+      "step": 5615
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1357795413118065e-06,
+      "loss": 0.6562,
+      "step": 5616
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.132949957300474e-06,
+      "loss": 0.6782,
+      "step": 5617
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.130122025136958e-06,
+      "loss": 0.6193,
+      "step": 5618
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1272957454150344e-06,
+      "loss": 0.6682,
+      "step": 5619
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.124471118728144e-06,
+      "loss": 0.6648,
+      "step": 5620
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1216481456693684e-06,
+      "loss": 0.6525,
+      "step": 5621
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1188268268314527e-06,
+      "loss": 0.6607,
+      "step": 5622
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1160071628067956e-06,
+      "loss": 0.6889,
+      "step": 5623
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1131891541874316e-06,
+      "loss": 0.6744,
+      "step": 5624
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.110372801565065e-06,
+      "loss": 0.7114,
+      "step": 5625
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.107558105531048e-06,
+      "loss": 0.6509,
+      "step": 5626
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.10474506667638e-06,
+      "loss": 0.5859,
+      "step": 5627
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1019336855917184e-06,
+      "loss": 0.6768,
+      "step": 5628
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.099123962867372e-06,
+      "loss": 0.6507,
+      "step": 5629
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0963158990932964e-06,
+      "loss": 0.6441,
+      "step": 5630
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0935094948590996e-06,
+      "loss": 0.6074,
+      "step": 5631
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.09070475075405e-06,
+      "loss": 0.724,
+      "step": 5632
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.087901667367054e-06,
+      "loss": 0.6807,
+      "step": 5633
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.085100245286683e-06,
+      "loss": 0.6661,
+      "step": 5634
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0823004851011454e-06,
+      "loss": 0.6819,
+      "step": 5635
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0795023873983135e-06,
+      "loss": 0.6657,
+      "step": 5636
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.076705952765705e-06,
+      "loss": 0.6805,
+      "step": 5637
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0739111817904867e-06,
+      "loss": 0.6517,
+      "step": 5638
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.071118075059476e-06,
+      "loss": 0.6786,
+      "step": 5639
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0683266331591466e-06,
+      "loss": 0.6866,
+      "step": 5640
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.065536856675615e-06,
+      "loss": 0.6808,
+      "step": 5641
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0627487461946572e-06,
+      "loss": 0.6345,
+      "step": 5642
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0599623023016856e-06,
+      "loss": 0.6554,
+      "step": 5643
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.057177525581777e-06,
+      "loss": 0.6482,
+      "step": 5644
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.054394416619654e-06,
+      "loss": 0.6813,
+      "step": 5645
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.051612975999684e-06,
+      "loss": 0.71,
+      "step": 5646
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0488332043058834e-06,
+      "loss": 0.6942,
+      "step": 5647
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0460551021219297e-06,
+      "loss": 0.665,
+      "step": 5648
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.043278670031136e-06,
+      "loss": 0.5803,
+      "step": 5649
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0405039086164722e-06,
+      "loss": 0.6809,
+      "step": 5650
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0377308184605604e-06,
+      "loss": 0.6929,
+      "step": 5651
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.034959400145665e-06,
+      "loss": 0.6749,
+      "step": 5652
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.032189654253698e-06,
+      "loss": 0.6792,
+      "step": 5653
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.029421581366231e-06,
+      "loss": 0.5789,
+      "step": 5654
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0266551820644696e-06,
+      "loss": 0.6312,
+      "step": 5655
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.023890456929284e-06,
+      "loss": 0.6192,
+      "step": 5656
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.021127406541178e-06,
+      "loss": 0.6559,
+      "step": 5657
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0183660314803123e-06,
+      "loss": 0.6683,
+      "step": 5658
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0156063323265006e-06,
+      "loss": 0.6134,
+      "step": 5659
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.012848309659188e-06,
+      "loss": 0.6266,
+      "step": 5660
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.010091964057481e-06,
+      "loss": 0.5512,
+      "step": 5661
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0073372961001346e-06,
+      "loss": 0.6798,
+      "step": 5662
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0045843063655424e-06,
+      "loss": 0.6583,
+      "step": 5663
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.001832995431753e-06,
+      "loss": 0.6397,
+      "step": 5664
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9990833638764618e-06,
+      "loss": 0.6676,
+      "step": 5665
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9963354122770107e-06,
+      "loss": 0.5709,
+      "step": 5666
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9935891412103824e-06,
+      "loss": 0.643,
+      "step": 5667
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9908445512532192e-06,
+      "loss": 0.6987,
+      "step": 5668
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.988101642981799e-06,
+      "loss": 0.6214,
+      "step": 5669
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.985360416972055e-06,
+      "loss": 0.67,
+      "step": 5670
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.982620873799561e-06,
+      "loss": 0.6446,
+      "step": 5671
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9798830140395407e-06,
+      "loss": 0.6077,
+      "step": 5672
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.977146838266869e-06,
+      "loss": 0.653,
+      "step": 5673
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9744123470560517e-06,
+      "loss": 0.6632,
+      "step": 5674
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9716795409812563e-06,
+      "loss": 0.5704,
+      "step": 5675
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9689484206162933e-06,
+      "loss": 0.635,
+      "step": 5676
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9662189865346137e-06,
+      "loss": 0.6385,
+      "step": 5677
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.963491239309322e-06,
+      "loss": 0.6341,
+      "step": 5678
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9607651795131587e-06,
+      "loss": 0.6529,
+      "step": 5679
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9580408077185207e-06,
+      "loss": 0.6704,
+      "step": 5680
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.955318124497443e-06,
+      "loss": 0.674,
+      "step": 5681
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9525971304216095e-06,
+      "loss": 0.6565,
+      "step": 5682
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9498778260623475e-06,
+      "loss": 0.6887,
+      "step": 5683
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.947160211990633e-06,
+      "loss": 0.6945,
+      "step": 5684
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.94444428877708e-06,
+      "loss": 0.6357,
+      "step": 5685
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.941730056991954e-06,
+      "loss": 0.6616,
+      "step": 5686
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9390175172051682e-06,
+      "loss": 0.5474,
+      "step": 5687
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.936306669986271e-06,
+      "loss": 0.6009,
+      "step": 5688
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9335975159044584e-06,
+      "loss": 0.5547,
+      "step": 5689
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.930890055528577e-06,
+      "loss": 0.6467,
+      "step": 5690
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9281842894271097e-06,
+      "loss": 0.6577,
+      "step": 5691
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9254802181681907e-06,
+      "loss": 0.5958,
+      "step": 5692
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.922777842319591e-06,
+      "loss": 0.6481,
+      "step": 5693
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9200771624487326e-06,
+      "loss": 0.6777,
+      "step": 5694
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.917378179122683e-06,
+      "loss": 0.6653,
+      "step": 5695
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9146808929081373e-06,
+      "loss": 0.6738,
+      "step": 5696
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9119853043714533e-06,
+      "loss": 0.6662,
+      "step": 5697
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9092914140786255e-06,
+      "loss": 0.6688,
+      "step": 5698
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9065992225952867e-06,
+      "loss": 0.6681,
+      "step": 5699
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.903908730486721e-06,
+      "loss": 0.6748,
+      "step": 5700
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9012199383178532e-06,
+      "loss": 0.6599,
+      "step": 5701
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8985328466532482e-06,
+      "loss": 0.688,
+      "step": 5702
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8958474560571139e-06,
+      "loss": 0.7057,
+      "step": 5703
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8931637670933067e-06,
+      "loss": 0.6163,
+      "step": 5704
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8904817803253184e-06,
+      "loss": 0.6974,
+      "step": 5705
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8878014963162904e-06,
+      "loss": 0.6719,
+      "step": 5706
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8851229156289986e-06,
+      "loss": 0.5575,
+      "step": 5707
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.882446038825868e-06,
+      "loss": 0.7039,
+      "step": 5708
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8797708664689695e-06,
+      "loss": 0.6261,
+      "step": 5709
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8770973991199991e-06,
+      "loss": 0.5322,
+      "step": 5710
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8744256373403113e-06,
+      "loss": 0.5434,
+      "step": 5711
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8717555816908995e-06,
+      "loss": 0.6617,
+      "step": 5712
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8690872327323905e-06,
+      "loss": 0.673,
+      "step": 5713
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8664205910250632e-06,
+      "loss": 0.6434,
+      "step": 5714
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.863755657128835e-06,
+      "loss": 0.6891,
+      "step": 5715
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8610924316032596e-06,
+      "loss": 0.6567,
+      "step": 5716
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8584309150075353e-06,
+      "loss": 0.6601,
+      "step": 5717
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8557711079005046e-06,
+      "loss": 0.6525,
+      "step": 5718
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8531130108406447e-06,
+      "loss": 0.6739,
+      "step": 5719
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8504566243860822e-06,
+      "loss": 0.6645,
+      "step": 5720
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8478019490945743e-06,
+      "loss": 0.5745,
+      "step": 5721
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8451489855235261e-06,
+      "loss": 0.6342,
+      "step": 5722
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8424977342299887e-06,
+      "loss": 0.7106,
+      "step": 5723
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8398481957706337e-06,
+      "loss": 0.6852,
+      "step": 5724
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8372003707017937e-06,
+      "loss": 0.6689,
+      "step": 5725
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.834554259579433e-06,
+      "loss": 0.5623,
+      "step": 5726
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8319098629591547e-06,
+      "loss": 0.6681,
+      "step": 5727
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8292671813962071e-06,
+      "loss": 0.6618,
+      "step": 5728
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8266262154454706e-06,
+      "loss": 0.6514,
+      "step": 5729
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.823986965661474e-06,
+      "loss": 0.6497,
+      "step": 5730
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8213494325983782e-06,
+      "loss": 0.5674,
+      "step": 5731
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8187136168099917e-06,
+      "loss": 0.7116,
+      "step": 5732
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8160795188497527e-06,
+      "loss": 0.6487,
+      "step": 5733
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8134471392707487e-06,
+      "loss": 0.6575,
+      "step": 5734
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8108164786256953e-06,
+      "loss": 0.6931,
+      "step": 5735
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8081875374669588e-06,
+      "loss": 0.6226,
+      "step": 5736
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8055603163465396e-06,
+      "loss": 0.6475,
+      "step": 5737
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8029348158160742e-06,
+      "loss": 0.68,
+      "step": 5738
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8003110364268383e-06,
+      "loss": 0.6602,
+      "step": 5739
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7976889787297526e-06,
+      "loss": 0.6803,
+      "step": 5740
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7950686432753662e-06,
+      "loss": 0.6262,
+      "step": 5741
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.792450030613877e-06,
+      "loss": 0.5873,
+      "step": 5742
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7898331412951108e-06,
+      "loss": 0.5487,
+      "step": 5743
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7872179758685415e-06,
+      "loss": 0.6686,
+      "step": 5744
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7846045348832796e-06,
+      "loss": 0.6544,
+      "step": 5745
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.781992818888061e-06,
+      "loss": 0.6444,
+      "step": 5746
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.779382828431273e-06,
+      "loss": 0.6832,
+      "step": 5747
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7767745640609401e-06,
+      "loss": 0.6547,
+      "step": 5748
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7741680263247164e-06,
+      "loss": 0.7239,
+      "step": 5749
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.771563215769897e-06,
+      "loss": 0.6781,
+      "step": 5750
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7689601329434214e-06,
+      "loss": 0.6656,
+      "step": 5751
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7663587783918545e-06,
+      "loss": 0.6586,
+      "step": 5752
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.763759152661403e-06,
+      "loss": 0.7104,
+      "step": 5753
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7611612562979163e-06,
+      "loss": 0.6492,
+      "step": 5754
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7585650898468698e-06,
+      "loss": 0.6129,
+      "step": 5755
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7559706538533882e-06,
+      "loss": 0.6757,
+      "step": 5756
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7533779488622193e-06,
+      "loss": 0.6752,
+      "step": 5757
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7507869754177587e-06,
+      "loss": 0.6627,
+      "step": 5758
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7481977340640387e-06,
+      "loss": 0.6403,
+      "step": 5759
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7456102253447128e-06,
+      "loss": 0.6806,
+      "step": 5760
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.743024449803087e-06,
+      "loss": 0.6432,
+      "step": 5761
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7404404079821002e-06,
+      "loss": 0.6561,
+      "step": 5762
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.737858100424319e-06,
+      "loss": 0.6335,
+      "step": 5763
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.735277527671957e-06,
+      "loss": 0.6119,
+      "step": 5764
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.732698690266853e-06,
+      "loss": 0.6851,
+      "step": 5765
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.730121588750493e-06,
+      "loss": 0.5266,
+      "step": 5766
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7275462236639851e-06,
+      "loss": 0.6874,
+      "step": 5767
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7249725955480856e-06,
+      "loss": 0.712,
+      "step": 5768
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7224007049431757e-06,
+      "loss": 0.6915,
+      "step": 5769
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.71983055238928e-06,
+      "loss": 0.6436,
+      "step": 5770
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7172621384260514e-06,
+      "loss": 0.6122,
+      "step": 5771
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7146954635927827e-06,
+      "loss": 0.716,
+      "step": 5772
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7121305284284029e-06,
+      "loss": 0.6709,
+      "step": 5773
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7095673334714658e-06,
+      "loss": 0.6474,
+      "step": 5774
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7070058792601685e-06,
+      "loss": 0.6879,
+      "step": 5775
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.704446166332344e-06,
+      "loss": 0.639,
+      "step": 5776
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7018881952254529e-06,
+      "loss": 0.5948,
+      "step": 5777
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6993319664765972e-06,
+      "loss": 0.6541,
+      "step": 5778
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6967774806225034e-06,
+      "loss": 0.6518,
+      "step": 5779
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6942247381995446e-06,
+      "loss": 0.6663,
+      "step": 5780
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6916737397437178e-06,
+      "loss": 0.704,
+      "step": 5781
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6891244857906553e-06,
+      "loss": 0.6865,
+      "step": 5782
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6865769768756269e-06,
+      "loss": 0.6818,
+      "step": 5783
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.684031213533538e-06,
+      "loss": 0.6361,
+      "step": 5784
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6814871962989154e-06,
+      "loss": 0.6093,
+      "step": 5785
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6789449257059344e-06,
+      "loss": 0.656,
+      "step": 5786
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6764044022883953e-06,
+      "loss": 0.5396,
+      "step": 5787
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6738656265797316e-06,
+      "loss": 0.6449,
+      "step": 5788
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6713285991130102e-06,
+      "loss": 0.6505,
+      "step": 5789
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.668793320420935e-06,
+      "loss": 0.6645,
+      "step": 5790
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6662597910358348e-06,
+      "loss": 0.6469,
+      "step": 5791
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6637280114896815e-06,
+      "loss": 0.6573,
+      "step": 5792
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.661197982314068e-06,
+      "loss": 0.6547,
+      "step": 5793
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6586697040402289e-06,
+      "loss": 0.6653,
+      "step": 5794
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6561431771990321e-06,
+      "loss": 0.6611,
+      "step": 5795
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6536184023209645e-06,
+      "loss": 0.6491,
+      "step": 5796
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6510953799361584e-06,
+      "loss": 0.6606,
+      "step": 5797
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.648574110574377e-06,
+      "loss": 0.6507,
+      "step": 5798
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6460545947650063e-06,
+      "loss": 0.6575,
+      "step": 5799
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.643536833037075e-06,
+      "loss": 0.6397,
+      "step": 5800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.641020825919235e-06,
+      "loss": 0.6622,
+      "step": 5801
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6385065739397787e-06,
+      "loss": 0.6596,
+      "step": 5802
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.635994077626617e-06,
+      "loss": 0.6326,
+      "step": 5803
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6334833375073068e-06,
+      "loss": 0.6326,
+      "step": 5804
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.630974354109025e-06,
+      "loss": 0.6549,
+      "step": 5805
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6284671279585862e-06,
+      "loss": 0.6319,
+      "step": 5806
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6259616595824312e-06,
+      "loss": 0.659,
+      "step": 5807
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.623457949506635e-06,
+      "loss": 0.6777,
+      "step": 5808
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.620955998256909e-06,
+      "loss": 0.6565,
+      "step": 5809
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6184558063585787e-06,
+      "loss": 0.6563,
+      "step": 5810
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6159573743366152e-06,
+      "loss": 0.67,
+      "step": 5811
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.613460702715618e-06,
+      "loss": 0.5988,
+      "step": 5812
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6109657920198086e-06,
+      "loss": 0.7034,
+      "step": 5813
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6084726427730501e-06,
+      "loss": 0.6639,
+      "step": 5814
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.605981255498824e-06,
+      "loss": 0.6655,
+      "step": 5815
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6034916307202553e-06,
+      "loss": 0.6545,
+      "step": 5816
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6010037689600876e-06,
+      "loss": 0.6132,
+      "step": 5817
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5985176707406947e-06,
+      "loss": 0.6022,
+      "step": 5818
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5960333365840886e-06,
+      "loss": 0.6742,
+      "step": 5819
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5935507670119066e-06,
+      "loss": 0.6373,
+      "step": 5820
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5910699625454106e-06,
+      "loss": 0.6467,
+      "step": 5821
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5885909237054985e-06,
+      "loss": 0.6198,
+      "step": 5822
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5861136510127006e-06,
+      "loss": 0.6571,
+      "step": 5823
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5836381449871596e-06,
+      "loss": 0.6736,
+      "step": 5824
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5811644061486653e-06,
+      "loss": 0.7155,
+      "step": 5825
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.578692435016631e-06,
+      "loss": 0.6787,
+      "step": 5826
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.576222232110093e-06,
+      "loss": 0.6639,
+      "step": 5827
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5737537979477258e-06,
+      "loss": 0.6596,
+      "step": 5828
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.571287133047822e-06,
+      "loss": 0.6613,
+      "step": 5829
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5688222379283136e-06,
+      "loss": 0.6911,
+      "step": 5830
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5663591131067534e-06,
+      "loss": 0.6715,
+      "step": 5831
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5638977591003234e-06,
+      "loss": 0.6557,
+      "step": 5832
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5614381764258346e-06,
+      "loss": 0.6881,
+      "step": 5833
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5589803655997316e-06,
+      "loss": 0.6671,
+      "step": 5834
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.556524327138077e-06,
+      "loss": 0.6412,
+      "step": 5835
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5540700615565697e-06,
+      "loss": 0.6095,
+      "step": 5836
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5516175693705282e-06,
+      "loss": 0.6318,
+      "step": 5837
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5491668510949099e-06,
+      "loss": 0.6587,
+      "step": 5838
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5467179072442862e-06,
+      "loss": 0.6461,
+      "step": 5839
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5442707383328693e-06,
+      "loss": 0.6806,
+      "step": 5840
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5418253448744857e-06,
+      "loss": 0.6556,
+      "step": 5841
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5393817273826028e-06,
+      "loss": 0.6429,
+      "step": 5842
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5369398863703011e-06,
+      "loss": 0.6481,
+      "step": 5843
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5344998223502994e-06,
+      "loss": 0.5827,
+      "step": 5844
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5320615358349412e-06,
+      "loss": 0.7008,
+      "step": 5845
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5296250273361867e-06,
+      "loss": 0.6723,
+      "step": 5846
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.527190297365636e-06,
+      "loss": 0.6653,
+      "step": 5847
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.524757346434511e-06,
+      "loss": 0.6743,
+      "step": 5848
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5223261750536566e-06,
+      "loss": 0.6542,
+      "step": 5849
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.51989678373355e-06,
+      "loss": 0.66,
+      "step": 5850
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5174691729842873e-06,
+      "loss": 0.6457,
+      "step": 5851
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5150433433155997e-06,
+      "loss": 0.6251,
+      "step": 5852
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.512619295236837e-06,
+      "loss": 0.677,
+      "step": 5853
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5101970292569757e-06,
+      "loss": 0.6689,
+      "step": 5854
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5077765458846227e-06,
+      "loss": 0.6863,
+      "step": 5855
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5053578456280094e-06,
+      "loss": 0.6841,
+      "step": 5856
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.502940928994988e-06,
+      "loss": 0.6447,
+      "step": 5857
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5005257964930397e-06,
+      "loss": 0.6869,
+      "step": 5858
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4981124486292775e-06,
+      "loss": 0.65,
+      "step": 5859
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4957008859104226e-06,
+      "loss": 0.6418,
+      "step": 5860
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.493291108842837e-06,
+      "loss": 0.6363,
+      "step": 5861
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4908831179325046e-06,
+      "loss": 0.6498,
+      "step": 5862
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4884769136850275e-06,
+      "loss": 0.579,
+      "step": 5863
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4860724966056428e-06,
+      "loss": 0.6663,
+      "step": 5864
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4836698671992022e-06,
+      "loss": 0.563,
+      "step": 5865
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4812690259701912e-06,
+      "loss": 0.6428,
+      "step": 5866
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.478869973422713e-06,
+      "loss": 0.6314,
+      "step": 5867
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4764727100604947e-06,
+      "loss": 0.6497,
+      "step": 5868
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.474077236386895e-06,
+      "loss": 0.6528,
+      "step": 5869
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4716835529048924e-06,
+      "loss": 0.67,
+      "step": 5870
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.469291660117087e-06,
+      "loss": 0.5722,
+      "step": 5871
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4669015585257074e-06,
+      "loss": 0.6575,
+      "step": 5872
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4645132486326075e-06,
+      "loss": 0.6741,
+      "step": 5873
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.462126730939254e-06,
+      "loss": 0.6486,
+      "step": 5874
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4597420059467493e-06,
+      "loss": 0.6589,
+      "step": 5875
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4573590741558164e-06,
+      "loss": 0.6783,
+      "step": 5876
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4549779360667981e-06,
+      "loss": 0.6285,
+      "step": 5877
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4525985921796648e-06,
+      "loss": 0.639,
+      "step": 5878
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4502210429940067e-06,
+      "loss": 0.6483,
+      "step": 5879
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.447845289009041e-06,
+      "loss": 0.6576,
+      "step": 5880
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4454713307236045e-06,
+      "loss": 0.6846,
+      "step": 5881
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4430991686361551e-06,
+      "loss": 0.6926,
+      "step": 5882
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4407288032447809e-06,
+      "loss": 0.6957,
+      "step": 5883
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4383602350471881e-06,
+      "loss": 0.6879,
+      "step": 5884
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4359934645407036e-06,
+      "loss": 0.7027,
+      "step": 5885
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4336284922222832e-06,
+      "loss": 0.6411,
+      "step": 5886
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4312653185884962e-06,
+      "loss": 0.6718,
+      "step": 5887
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4289039441355434e-06,
+      "loss": 0.6772,
+      "step": 5888
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4265443693592406e-06,
+      "loss": 0.5304,
+      "step": 5889
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4241865947550315e-06,
+      "loss": 0.5647,
+      "step": 5890
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4218306208179744e-06,
+      "loss": 0.7258,
+      "step": 5891
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4194764480427604e-06,
+      "loss": 0.6802,
+      "step": 5892
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4171240769236905e-06,
+      "loss": 0.664,
+      "step": 5893
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4147735079546954e-06,
+      "loss": 0.6622,
+      "step": 5894
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4124247416293301e-06,
+      "loss": 0.656,
+      "step": 5895
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.410077778440756e-06,
+      "loss": 0.6341,
+      "step": 5896
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.407732618881772e-06,
+      "loss": 0.6555,
+      "step": 5897
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4053892634447919e-06,
+      "loss": 0.6922,
+      "step": 5898
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4030477126218488e-06,
+      "loss": 0.6621,
+      "step": 5899
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4007079669046042e-06,
+      "loss": 0.7068,
+      "step": 5900
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.398370026784328e-06,
+      "loss": 0.5634,
+      "step": 5901
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3960338927519257e-06,
+      "loss": 0.6352,
+      "step": 5902
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3936995652979135e-06,
+      "loss": 0.621,
+      "step": 5903
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3913670449124273e-06,
+      "loss": 0.674,
+      "step": 5904
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3890363320852319e-06,
+      "loss": 0.6606,
+      "step": 5905
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3867074273057092e-06,
+      "loss": 0.6931,
+      "step": 5906
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3843803310628557e-06,
+      "loss": 0.6166,
+      "step": 5907
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3820550438452952e-06,
+      "loss": 0.6212,
+      "step": 5908
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3797315661412735e-06,
+      "loss": 0.6915,
+      "step": 5909
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.377409898438643e-06,
+      "loss": 0.6721,
+      "step": 5910
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3750900412248912e-06,
+      "loss": 0.6739,
+      "step": 5911
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.37277199498712e-06,
+      "loss": 0.6547,
+      "step": 5912
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3704557602120472e-06,
+      "loss": 0.5678,
+      "step": 5913
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3681413373860175e-06,
+      "loss": 0.6718,
+      "step": 5914
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3658287269949867e-06,
+      "loss": 0.6403,
+      "step": 5915
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3635179295245394e-06,
+      "loss": 0.6882,
+      "step": 5916
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3612089454598721e-06,
+      "loss": 0.677,
+      "step": 5917
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3589017752858014e-06,
+      "loss": 0.6782,
+      "step": 5918
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3565964194867664e-06,
+      "loss": 0.6542,
+      "step": 5919
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3542928785468268e-06,
+      "loss": 0.636,
+      "step": 5920
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3519911529496531e-06,
+      "loss": 0.6696,
+      "step": 5921
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3496912431785436e-06,
+      "loss": 0.6968,
+      "step": 5922
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3473931497164106e-06,
+      "loss": 0.6601,
+      "step": 5923
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3450968730457813e-06,
+      "loss": 0.6059,
+      "step": 5924
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3428024136488093e-06,
+      "loss": 0.6815,
+      "step": 5925
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3405097720072656e-06,
+      "loss": 0.6012,
+      "step": 5926
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.338218948602532e-06,
+      "loss": 0.639,
+      "step": 5927
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3359299439156192e-06,
+      "loss": 0.6721,
+      "step": 5928
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3336427584271449e-06,
+      "loss": 0.6444,
+      "step": 5929
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3313573926173563e-06,
+      "loss": 0.5523,
+      "step": 5930
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3290738469661079e-06,
+      "loss": 0.6488,
+      "step": 5931
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3267921219528778e-06,
+      "loss": 0.667,
+      "step": 5932
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3245122180567604e-06,
+      "loss": 0.6808,
+      "step": 5933
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3222341357564706e-06,
+      "loss": 0.6601,
+      "step": 5934
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3199578755303354e-06,
+      "loss": 0.5521,
+      "step": 5935
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3176834378563042e-06,
+      "loss": 0.5546,
+      "step": 5936
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.31541082321194e-06,
+      "loss": 0.6559,
+      "step": 5937
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.313140032074427e-06,
+      "loss": 0.6886,
+      "step": 5938
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3108710649205625e-06,
+      "loss": 0.5837,
+      "step": 5939
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.30860392222676e-06,
+      "loss": 0.5625,
+      "step": 5940
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3063386044690563e-06,
+      "loss": 0.6853,
+      "step": 5941
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.304075112123101e-06,
+      "loss": 0.6809,
+      "step": 5942
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3018134456641574e-06,
+      "loss": 0.6894,
+      "step": 5943
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.299553605567111e-06,
+      "loss": 0.6338,
+      "step": 5944
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2972955923064645e-06,
+      "loss": 0.6048,
+      "step": 5945
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2950394063563265e-06,
+      "loss": 0.6736,
+      "step": 5946
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2927850481904324e-06,
+      "loss": 0.6227,
+      "step": 5947
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2905325182821337e-06,
+      "loss": 0.6385,
+      "step": 5948
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.288281817104391e-06,
+      "loss": 0.6158,
+      "step": 5949
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2860329451297892e-06,
+      "loss": 0.6691,
+      "step": 5950
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.28378590283052e-06,
+      "loss": 0.6633,
+      "step": 5951
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2815406906784011e-06,
+      "loss": 0.6487,
+      "step": 5952
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2792973091448569e-06,
+      "loss": 0.6781,
+      "step": 5953
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2770557587009302e-06,
+      "loss": 0.6773,
+      "step": 5954
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2748160398172827e-06,
+      "loss": 0.6339,
+      "step": 5955
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2725781529641901e-06,
+      "loss": 0.5894,
+      "step": 5956
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.270342098611539e-06,
+      "loss": 0.6146,
+      "step": 5957
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2681078772288392e-06,
+      "loss": 0.5444,
+      "step": 5958
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2658754892852076e-06,
+      "loss": 0.6289,
+      "step": 5959
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2636449352493795e-06,
+      "loss": 0.6773,
+      "step": 5960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2614162155897058e-06,
+      "loss": 0.6552,
+      "step": 5961
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2591893307741542e-06,
+      "loss": 0.6598,
+      "step": 5962
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2569642812703008e-06,
+      "loss": 0.663,
+      "step": 5963
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2547410675453442e-06,
+      "loss": 0.6897,
+      "step": 5964
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2525196900660897e-06,
+      "loss": 0.7215,
+      "step": 5965
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2503001492989642e-06,
+      "loss": 0.571,
+      "step": 5966
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2480824457100039e-06,
+      "loss": 0.6376,
+      "step": 5967
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2458665797648583e-06,
+      "loss": 0.5555,
+      "step": 5968
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2436525519287979e-06,
+      "loss": 0.6862,
+      "step": 5969
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2414403626667026e-06,
+      "loss": 0.651,
+      "step": 5970
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2392300124430645e-06,
+      "loss": 0.6719,
+      "step": 5971
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2370215017219945e-06,
+      "loss": 0.6372,
+      "step": 5972
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2348148309672126e-06,
+      "loss": 0.6718,
+      "step": 5973
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2326100006420539e-06,
+      "loss": 0.6582,
+      "step": 5974
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.230407011209469e-06,
+      "loss": 0.6738,
+      "step": 5975
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2282058631320193e-06,
+      "loss": 0.6208,
+      "step": 5976
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2260065568718805e-06,
+      "loss": 0.6563,
+      "step": 5977
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2238090928908453e-06,
+      "loss": 0.6649,
+      "step": 5978
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2216134716503114e-06,
+      "loss": 0.6646,
+      "step": 5979
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.219419693611299e-06,
+      "loss": 0.6425,
+      "step": 5980
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.217227759234434e-06,
+      "loss": 0.6874,
+      "step": 5981
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.215037668979956e-06,
+      "loss": 0.5559,
+      "step": 5982
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2128494233077204e-06,
+      "loss": 0.6825,
+      "step": 5983
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2106630226771976e-06,
+      "loss": 0.691,
+      "step": 5984
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2084784675474614e-06,
+      "loss": 0.648,
+      "step": 5985
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2062957583772095e-06,
+      "loss": 0.6259,
+      "step": 5986
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2041148956247416e-06,
+      "loss": 0.6734,
+      "step": 5987
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2019358797479774e-06,
+      "loss": 0.6509,
+      "step": 5988
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1997587112044451e-06,
+      "loss": 0.6956,
+      "step": 5989
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1975833904512823e-06,
+      "loss": 0.6609,
+      "step": 5990
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.195409917945245e-06,
+      "loss": 0.5822,
+      "step": 5991
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1932382941427012e-06,
+      "loss": 0.6361,
+      "step": 5992
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1910685194996209e-06,
+      "loss": 0.6699,
+      "step": 5993
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.188900594471597e-06,
+      "loss": 0.6311,
+      "step": 5994
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1867345195138324e-06,
+      "loss": 0.6914,
+      "step": 5995
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1845702950811322e-06,
+      "loss": 0.6655,
+      "step": 5996
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1824079216279216e-06,
+      "loss": 0.6511,
+      "step": 5997
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1802473996082375e-06,
+      "loss": 0.6567,
+      "step": 5998
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1780887294757226e-06,
+      "loss": 0.6788,
+      "step": 5999
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1759319116836376e-06,
+      "loss": 0.6629,
+      "step": 6000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.173776946684846e-06,
+      "loss": 0.6419,
+      "step": 6001
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1716238349318298e-06,
+      "loss": 0.6175,
+      "step": 6002
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1694725768766779e-06,
+      "loss": 0.6604,
+      "step": 6003
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1673231729710898e-06,
+      "loss": 0.6909,
+      "step": 6004
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.165175623666377e-06,
+      "loss": 0.6898,
+      "step": 6005
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1630299294134638e-06,
+      "loss": 0.6592,
+      "step": 6006
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1608860906628794e-06,
+      "loss": 0.6495,
+      "step": 6007
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1587441078647688e-06,
+      "loss": 0.6596,
+      "step": 6008
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1566039814688857e-06,
+      "loss": 0.624,
+      "step": 6009
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1544657119245883e-06,
+      "loss": 0.6282,
+      "step": 6010
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1523292996808532e-06,
+      "loss": 0.7035,
+      "step": 6011
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1501947451862671e-06,
+      "loss": 0.6743,
+      "step": 6012
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.148062048889017e-06,
+      "loss": 0.6395,
+      "step": 6013
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1459312112369125e-06,
+      "loss": 0.6974,
+      "step": 6014
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1438022326773602e-06,
+      "loss": 0.6346,
+      "step": 6015
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1416751136573878e-06,
+      "loss": 0.5596,
+      "step": 6016
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1395498546236262e-06,
+      "loss": 0.6784,
+      "step": 6017
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1374264560223125e-06,
+      "loss": 0.67,
+      "step": 6018
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1353049182993014e-06,
+      "loss": 0.6552,
+      "step": 6019
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1331852419000555e-06,
+      "loss": 0.5468,
+      "step": 6020
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1310674272696387e-06,
+      "loss": 0.6595,
+      "step": 6021
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1289514748527342e-06,
+      "loss": 0.5739,
+      "step": 6022
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1268373850936265e-06,
+      "loss": 0.6623,
+      "step": 6023
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1247251584362118e-06,
+      "loss": 0.5994,
+      "step": 6024
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1226147953239974e-06,
+      "loss": 0.6723,
+      "step": 6025
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1205062962000945e-06,
+      "loss": 0.651,
+      "step": 6026
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1183996615072267e-06,
+      "loss": 0.6793,
+      "step": 6027
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.116294891687727e-06,
+      "loss": 0.6076,
+      "step": 6028
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1141919871835306e-06,
+      "loss": 0.6307,
+      "step": 6029
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1120909484361897e-06,
+      "loss": 0.6746,
+      "step": 6030
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1099917758868595e-06,
+      "loss": 0.6893,
+      "step": 6031
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1078944699762995e-06,
+      "loss": 0.5814,
+      "step": 6032
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1057990311448863e-06,
+      "loss": 0.6098,
+      "step": 6033
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1037054598326003e-06,
+      "loss": 0.6413,
+      "step": 6034
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1016137564790265e-06,
+      "loss": 0.6295,
+      "step": 6035
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0995239215233644e-06,
+      "loss": 0.6596,
+      "step": 6036
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0974359554044133e-06,
+      "loss": 0.6755,
+      "step": 6037
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0953498585605905e-06,
+      "loss": 0.6321,
+      "step": 6038
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0932656314299096e-06,
+      "loss": 0.6882,
+      "step": 6039
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.091183274449995e-06,
+      "loss": 0.6546,
+      "step": 6040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.089102788058084e-06,
+      "loss": 0.6895,
+      "step": 6041
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.087024172691018e-06,
+      "loss": 0.6647,
+      "step": 6042
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0849474287852403e-06,
+      "loss": 0.6868,
+      "step": 6043
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0828725567768116e-06,
+      "loss": 0.6806,
+      "step": 6044
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0807995571013897e-06,
+      "loss": 0.6441,
+      "step": 6045
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0787284301942413e-06,
+      "loss": 0.6607,
+      "step": 6046
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0766591764902446e-06,
+      "loss": 0.627,
+      "step": 6047
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0745917964238838e-06,
+      "loss": 0.6689,
+      "step": 6048
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0725262904292432e-06,
+      "loss": 0.6938,
+      "step": 6049
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0704626589400213e-06,
+      "loss": 0.644,
+      "step": 6050
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0684009023895169e-06,
+      "loss": 0.6288,
+      "step": 6051
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0663410212106407e-06,
+      "loss": 0.6714,
+      "step": 6052
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0642830158359064e-06,
+      "loss": 0.7019,
+      "step": 6053
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0622268866974305e-06,
+      "loss": 0.6946,
+      "step": 6054
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0601726342269424e-06,
+      "loss": 0.5718,
+      "step": 6055
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.058120258855776e-06,
+      "loss": 0.6892,
+      "step": 6056
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.056069761014865e-06,
+      "loss": 0.6664,
+      "step": 6057
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.054021141134759e-06,
+      "loss": 0.6144,
+      "step": 6058
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0519743996456033e-06,
+      "loss": 0.6609,
+      "step": 6059
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0499295369771523e-06,
+      "loss": 0.5874,
+      "step": 6060
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.04788655355877e-06,
+      "loss": 0.6433,
+      "step": 6061
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0458454498194203e-06,
+      "loss": 0.6651,
+      "step": 6062
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0438062261876735e-06,
+      "loss": 0.6883,
+      "step": 6063
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0417688830917117e-06,
+      "loss": 0.5433,
+      "step": 6064
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0397334209593102e-06,
+      "loss": 0.7067,
+      "step": 6065
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0376998402178606e-06,
+      "loss": 0.6499,
+      "step": 6066
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0356681412943537e-06,
+      "loss": 0.6295,
+      "step": 6067
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0336383246153824e-06,
+      "loss": 0.6443,
+      "step": 6068
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0316103906071507e-06,
+      "loss": 0.636,
+      "step": 6069
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0295843396954675e-06,
+      "loss": 0.6333,
+      "step": 6070
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0275601723057394e-06,
+      "loss": 0.6431,
+      "step": 6071
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0255378888629853e-06,
+      "loss": 0.6415,
+      "step": 6072
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0235174897918232e-06,
+      "loss": 0.6591,
+      "step": 6073
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.021498975516475e-06,
+      "loss": 0.6644,
+      "step": 6074
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0194823464607729e-06,
+      "loss": 0.6092,
+      "step": 6075
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0174676030481446e-06,
+      "loss": 0.6561,
+      "step": 6076
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0154547457016305e-06,
+      "loss": 0.6665,
+      "step": 6077
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0134437748438708e-06,
+      "loss": 0.5704,
+      "step": 6078
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0114346908971085e-06,
+      "loss": 0.6307,
+      "step": 6079
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0094274942831938e-06,
+      "loss": 0.6404,
+      "step": 6080
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0074221854235777e-06,
+      "loss": 0.6656,
+      "step": 6081
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0054187647393122e-06,
+      "loss": 0.6648,
+      "step": 6082
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0034172326510605e-06,
+      "loss": 0.6375,
+      "step": 6083
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0014175895790857e-06,
+      "loss": 0.6453,
+      "step": 6084
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.994198359432506e-07,
+      "loss": 0.6656,
+      "step": 6085
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.974239721630275e-07,
+      "loss": 0.6768,
+      "step": 6086
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.954299986574856e-07,
+      "loss": 0.6526,
+      "step": 6087
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.934379158453034e-07,
+      "loss": 0.6604,
+      "step": 6088
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.914477241447584e-07,
+      "loss": 0.6851,
+      "step": 6089
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.894594239737286e-07,
+      "loss": 0.5886,
+      "step": 6090
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.874730157497026e-07,
+      "loss": 0.64,
+      "step": 6091
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.854884998897662e-07,
+      "loss": 0.6644,
+      "step": 6092
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.835058768106077e-07,
+      "loss": 0.5589,
+      "step": 6093
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.815251469285213e-07,
+      "loss": 0.7086,
+      "step": 6094
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.795463106594016e-07,
+      "loss": 0.6609,
+      "step": 6095
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.775693684187437e-07,
+      "loss": 0.6736,
+      "step": 6096
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.755943206216489e-07,
+      "loss": 0.6819,
+      "step": 6097
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.736211676828156e-07,
+      "loss": 0.6827,
+      "step": 6098
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.716499100165512e-07,
+      "loss": 0.5996,
+      "step": 6099
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.69680548036761e-07,
+      "loss": 0.6375,
+      "step": 6100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.677130821569502e-07,
+      "loss": 0.5381,
+      "step": 6101
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.657475127902327e-07,
+      "loss": 0.6405,
+      "step": 6102
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.637838403493173e-07,
+      "loss": 0.6226,
+      "step": 6103
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.61822065246516e-07,
+      "loss": 0.5842,
+      "step": 6104
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.598621878937441e-07,
+      "loss": 0.6255,
+      "step": 6105
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.579042087025202e-07,
+      "loss": 0.6889,
+      "step": 6106
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.5594812808396e-07,
+      "loss": 0.5451,
+      "step": 6107
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.539939464487834e-07,
+      "loss": 0.6475,
+      "step": 6108
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.520416642073116e-07,
+      "loss": 0.6293,
+      "step": 6109
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.500912817694618e-07,
+      "loss": 0.65,
+      "step": 6110
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.481427995447623e-07,
+      "loss": 0.6608,
+      "step": 6111
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.461962179423323e-07,
+      "loss": 0.6399,
+      "step": 6112
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.442515373708982e-07,
+      "loss": 0.681,
+      "step": 6113
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.423087582387869e-07,
+      "loss": 0.661,
+      "step": 6114
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.403678809539218e-07,
+      "loss": 0.5664,
+      "step": 6115
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.384289059238327e-07,
+      "loss": 0.6671,
+      "step": 6116
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.36491833555645e-07,
+      "loss": 0.5971,
+      "step": 6117
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.345566642560855e-07,
+      "loss": 0.512,
+      "step": 6118
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.326233984314836e-07,
+      "loss": 0.6562,
+      "step": 6119
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.306920364877692e-07,
+      "loss": 0.5435,
+      "step": 6120
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.2876257883047e-07,
+      "loss": 0.6773,
+      "step": 6121
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.268350258647163e-07,
+      "loss": 0.5915,
+      "step": 6122
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.249093779952356e-07,
+      "loss": 0.5691,
+      "step": 6123
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.229856356263567e-07,
+      "loss": 0.6128,
+      "step": 6124
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.210637991620109e-07,
+      "loss": 0.6603,
+      "step": 6125
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.191438690057231e-07,
+      "loss": 0.6797,
+      "step": 6126
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.172258455606253e-07,
+      "loss": 0.7054,
+      "step": 6127
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.153097292294466e-07,
+      "loss": 0.643,
+      "step": 6128
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.133955204145107e-07,
+      "loss": 0.6821,
+      "step": 6129
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.114832195177481e-07,
+      "loss": 0.6569,
+      "step": 6130
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.095728269406856e-07,
+      "loss": 0.6022,
+      "step": 6131
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.076643430844445e-07,
+      "loss": 0.5328,
+      "step": 6132
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.057577683497565e-07,
+      "loss": 0.6545,
+      "step": 6133
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.038531031369402e-07,
+      "loss": 0.6819,
+      "step": 6134
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.019503478459213e-07,
+      "loss": 0.5488,
+      "step": 6135
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.000495028762246e-07,
+      "loss": 0.5428,
+      "step": 6136
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.981505686269676e-07,
+      "loss": 0.649,
+      "step": 6137
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.962535454968724e-07,
+      "loss": 0.636,
+      "step": 6138
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.943584338842581e-07,
+      "loss": 0.6495,
+      "step": 6139
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.924652341870388e-07,
+      "loss": 0.5902,
+      "step": 6140
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.90573946802733e-07,
+      "loss": 0.6312,
+      "step": 6141
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.886845721284565e-07,
+      "loss": 0.6631,
+      "step": 6142
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.867971105609186e-07,
+      "loss": 0.6423,
+      "step": 6143
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.849115624964344e-07,
+      "loss": 0.6689,
+      "step": 6144
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.830279283309106e-07,
+      "loss": 0.6372,
+      "step": 6145
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.811462084598532e-07,
+      "loss": 0.5851,
+      "step": 6146
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.792664032783715e-07,
+      "loss": 0.6394,
+      "step": 6147
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.773885131811654e-07,
+      "loss": 0.6784,
+      "step": 6148
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.755125385625373e-07,
+      "loss": 0.721,
+      "step": 6149
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.7363847981639e-07,
+      "loss": 0.6436,
+      "step": 6150
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.71766337336215e-07,
+      "loss": 0.6429,
+      "step": 6151
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.698961115151106e-07,
+      "loss": 0.5341,
+      "step": 6152
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.680278027457689e-07,
+      "loss": 0.6445,
+      "step": 6153
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.661614114204753e-07,
+      "loss": 0.6891,
+      "step": 6154
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.642969379311206e-07,
+      "loss": 0.6407,
+      "step": 6155
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.624343826691895e-07,
+      "loss": 0.6077,
+      "step": 6156
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.605737460257602e-07,
+      "loss": 0.6817,
+      "step": 6157
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.587150283915146e-07,
+      "loss": 0.6769,
+      "step": 6158
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.568582301567285e-07,
+      "loss": 0.6425,
+      "step": 6159
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.550033517112699e-07,
+      "loss": 0.6507,
+      "step": 6160
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.53150393444615e-07,
+      "loss": 0.6608,
+      "step": 6161
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.512993557458249e-07,
+      "loss": 0.594,
+      "step": 6162
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.494502390035642e-07,
+      "loss": 0.6606,
+      "step": 6163
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.476030436060956e-07,
+      "loss": 0.6859,
+      "step": 6164
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.457577699412711e-07,
+      "loss": 0.694,
+      "step": 6165
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.439144183965476e-07,
+      "loss": 0.6335,
+      "step": 6166
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.42072989358973e-07,
+      "loss": 0.7091,
+      "step": 6167
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.402334832151903e-07,
+      "loss": 0.6629,
+      "step": 6168
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.383959003514442e-07,
+      "loss": 0.6581,
+      "step": 6169
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.365602411535734e-07,
+      "loss": 0.7021,
+      "step": 6170
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.347265060070098e-07,
+      "loss": 0.6153,
+      "step": 6171
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.328946952967853e-07,
+      "loss": 0.6399,
+      "step": 6172
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.310648094075258e-07,
+      "loss": 0.551,
+      "step": 6173
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.292368487234503e-07,
+      "loss": 0.6075,
+      "step": 6174
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.27410813628381e-07,
+      "loss": 0.6745,
+      "step": 6175
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.255867045057265e-07,
+      "loss": 0.6628,
+      "step": 6176
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.237645217384982e-07,
+      "loss": 0.4972,
+      "step": 6177
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.219442657093036e-07,
+      "loss": 0.67,
+      "step": 6178
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.201259368003367e-07,
+      "loss": 0.622,
+      "step": 6179
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.183095353933978e-07,
+      "loss": 0.6109,
+      "step": 6180
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.164950618698764e-07,
+      "loss": 0.6292,
+      "step": 6181
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.146825166107552e-07,
+      "loss": 0.6571,
+      "step": 6182
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.128718999966179e-07,
+      "loss": 0.6656,
+      "step": 6183
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.11063212407639e-07,
+      "loss": 0.6079,
+      "step": 6184
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.092564542235904e-07,
+      "loss": 0.6838,
+      "step": 6185
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.074516258238385e-07,
+      "loss": 0.629,
+      "step": 6186
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.056487275873414e-07,
+      "loss": 0.6489,
+      "step": 6187
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.038477598926575e-07,
+      "loss": 0.6484,
+      "step": 6188
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.020487231179364e-07,
+      "loss": 0.6797,
+      "step": 6189
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.002516176409181e-07,
+      "loss": 0.6635,
+      "step": 6190
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.984564438389464e-07,
+      "loss": 0.6691,
+      "step": 6191
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.966632020889542e-07,
+      "loss": 0.6824,
+      "step": 6192
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.94871892767467e-07,
+      "loss": 0.6993,
+      "step": 6193
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.930825162506095e-07,
+      "loss": 0.6831,
+      "step": 6194
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.912950729140956e-07,
+      "loss": 0.6179,
+      "step": 6195
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.895095631332361e-07,
+      "loss": 0.6652,
+      "step": 6196
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.877259872829357e-07,
+      "loss": 0.6956,
+      "step": 6197
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.859443457376914e-07,
+      "loss": 0.6635,
+      "step": 6198
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.841646388715963e-07,
+      "loss": 0.6677,
+      "step": 6199
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.823868670583379e-07,
+      "loss": 0.6797,
+      "step": 6200
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.806110306711911e-07,
+      "loss": 0.6788,
+      "step": 6201
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.788371300830344e-07,
+      "loss": 0.6314,
+      "step": 6202
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.770651656663319e-07,
+      "loss": 0.6391,
+      "step": 6203
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.752951377931417e-07,
+      "loss": 0.7106,
+      "step": 6204
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.735270468351186e-07,
+      "loss": 0.6495,
+      "step": 6205
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.717608931635123e-07,
+      "loss": 0.6492,
+      "step": 6206
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.699966771491596e-07,
+      "loss": 0.6718,
+      "step": 6207
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.682343991624952e-07,
+      "loss": 0.6348,
+      "step": 6208
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.664740595735454e-07,
+      "loss": 0.6595,
+      "step": 6209
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.647156587519266e-07,
+      "loss": 0.6814,
+      "step": 6210
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.629591970668559e-07,
+      "loss": 0.6293,
+      "step": 6211
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.612046748871327e-07,
+      "loss": 0.6978,
+      "step": 6212
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.594520925811588e-07,
+      "loss": 0.5837,
+      "step": 6213
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.577014505169244e-07,
+      "loss": 0.6241,
+      "step": 6214
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.5595274906201e-07,
+      "loss": 0.6373,
+      "step": 6215
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.54205988583595e-07,
+      "loss": 0.6475,
+      "step": 6216
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.524611694484451e-07,
+      "loss": 0.6488,
+      "step": 6217
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.507182920229195e-07,
+      "loss": 0.665,
+      "step": 6218
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.48977356672973e-07,
+      "loss": 0.6591,
+      "step": 6219
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.472383637641489e-07,
+      "loss": 0.6587,
+      "step": 6220
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.455013136615851e-07,
+      "loss": 0.6616,
+      "step": 6221
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.437662067300122e-07,
+      "loss": 0.6862,
+      "step": 6222
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.420330433337508e-07,
+      "loss": 0.7024,
+      "step": 6223
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.40301823836711e-07,
+      "loss": 0.5407,
+      "step": 6224
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.385725486024009e-07,
+      "loss": 0.667,
+      "step": 6225
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.368452179939157e-07,
+      "loss": 0.6433,
+      "step": 6226
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.351198323739428e-07,
+      "loss": 0.6653,
+      "step": 6227
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.33396392104766e-07,
+      "loss": 0.6533,
+      "step": 6228
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.316748975482513e-07,
+      "loss": 0.6007,
+      "step": 6229
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.299553490658673e-07,
+      "loss": 0.6042,
+      "step": 6230
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.282377470186652e-07,
+      "loss": 0.6378,
+      "step": 6231
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.265220917672899e-07,
+      "loss": 0.563,
+      "step": 6232
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.24808383671981e-07,
+      "loss": 0.6412,
+      "step": 6233
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.230966230925628e-07,
+      "loss": 0.6231,
+      "step": 6234
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.213868103884558e-07,
+      "loss": 0.6649,
+      "step": 6235
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.196789459186738e-07,
+      "loss": 0.5547,
+      "step": 6236
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.179730300418141e-07,
+      "loss": 0.625,
+      "step": 6237
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.162690631160674e-07,
+      "loss": 0.6695,
+      "step": 6238
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.145670454992204e-07,
+      "loss": 0.6862,
+      "step": 6239
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.128669775486418e-07,
+      "loss": 0.6794,
+      "step": 6240
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.111688596212984e-07,
+      "loss": 0.6147,
+      "step": 6241
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.094726920737471e-07,
+      "loss": 0.6979,
+      "step": 6242
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.077784752621276e-07,
+      "loss": 0.6475,
+      "step": 6243
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.060862095421794e-07,
+      "loss": 0.6291,
+      "step": 6244
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.043958952692265e-07,
+      "loss": 0.6732,
+      "step": 6245
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.027075327981847e-07,
+      "loss": 0.5701,
+      "step": 6246
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.010211224835617e-07,
+      "loss": 0.6182,
+      "step": 6247
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.993366646794508e-07,
+      "loss": 0.6787,
+      "step": 6248
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.976541597395403e-07,
+      "loss": 0.6932,
+      "step": 6249
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.959736080171087e-07,
+      "loss": 0.6628,
+      "step": 6250
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.94295009865017e-07,
+      "loss": 0.6182,
+      "step": 6251
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.926183656357266e-07,
+      "loss": 0.629,
+      "step": 6252
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.909436756812804e-07,
+      "loss": 0.6284,
+      "step": 6253
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.892709403533126e-07,
+      "loss": 0.6356,
+      "step": 6254
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.876001600030513e-07,
+      "loss": 0.68,
+      "step": 6255
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.859313349813069e-07,
+      "loss": 0.677,
+      "step": 6256
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.842644656384856e-07,
+      "loss": 0.695,
+      "step": 6257
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.825995523245832e-07,
+      "loss": 0.656,
+      "step": 6258
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.809365953891789e-07,
+      "loss": 0.7144,
+      "step": 6259
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.792755951814433e-07,
+      "loss": 0.681,
+      "step": 6260
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.77616552050141e-07,
+      "loss": 0.6758,
+      "step": 6261
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.759594663436198e-07,
+      "loss": 0.6,
+      "step": 6262
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.743043384098181e-07,
+      "loss": 0.7203,
+      "step": 6263
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.726511685962667e-07,
+      "loss": 0.557,
+      "step": 6264
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.709999572500791e-07,
+      "loss": 0.6584,
+      "step": 6265
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.693507047179637e-07,
+      "loss": 0.6459,
+      "step": 6266
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.677034113462122e-07,
+      "loss": 0.542,
+      "step": 6267
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.66058077480708e-07,
+      "loss": 0.6447,
+      "step": 6268
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.644147034669246e-07,
+      "loss": 0.639,
+      "step": 6269
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.627732896499195e-07,
+      "loss": 0.632,
+      "step": 6270
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.611338363743414e-07,
+      "loss": 0.6674,
+      "step": 6271
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.594963439844293e-07,
+      "loss": 0.6581,
+      "step": 6272
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.578608128240061e-07,
+      "loss": 0.6274,
+      "step": 6273
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.562272432364847e-07,
+      "loss": 0.6696,
+      "step": 6274
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.545956355648675e-07,
+      "loss": 0.6951,
+      "step": 6275
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.529659901517427e-07,
+      "loss": 0.5986,
+      "step": 6276
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.513383073392876e-07,
+      "loss": 0.5877,
+      "step": 6277
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.4971258746927e-07,
+      "loss": 0.6186,
+      "step": 6278
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.480888308830402e-07,
+      "loss": 0.6792,
+      "step": 6279
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.46467037921541e-07,
+      "loss": 0.6934,
+      "step": 6280
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.448472089252999e-07,
+      "loss": 0.6906,
+      "step": 6281
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.432293442344317e-07,
+      "loss": 0.6362,
+      "step": 6282
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.416134441886434e-07,
+      "loss": 0.6506,
+      "step": 6283
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.399995091272227e-07,
+      "loss": 0.5885,
+      "step": 6284
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.383875393890493e-07,
+      "loss": 0.5738,
+      "step": 6285
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.367775353125916e-07,
+      "loss": 0.64,
+      "step": 6286
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.351694972359013e-07,
+      "loss": 0.6547,
+      "step": 6287
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.335634254966172e-07,
+      "loss": 0.5471,
+      "step": 6288
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.319593204319696e-07,
+      "loss": 0.6478,
+      "step": 6289
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.303571823787702e-07,
+      "loss": 0.6582,
+      "step": 6290
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.28757011673421e-07,
+      "loss": 0.6296,
+      "step": 6291
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.271588086519142e-07,
+      "loss": 0.6753,
+      "step": 6292
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.255625736498216e-07,
+      "loss": 0.6009,
+      "step": 6293
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.239683070023073e-07,
+      "loss": 0.5699,
+      "step": 6294
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.223760090441188e-07,
+      "loss": 0.6441,
+      "step": 6295
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.207856801095902e-07,
+      "loss": 0.5978,
+      "step": 6296
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.191973205326474e-07,
+      "loss": 0.6633,
+      "step": 6297
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.176109306467959e-07,
+      "loss": 0.6662,
+      "step": 6298
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.160265107851304e-07,
+      "loss": 0.6602,
+      "step": 6299
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.144440612803348e-07,
+      "loss": 0.6652,
+      "step": 6300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.128635824646745e-07,
+      "loss": 0.6907,
+      "step": 6301
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.112850746700061e-07,
+      "loss": 0.6367,
+      "step": 6302
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.097085382277668e-07,
+      "loss": 0.6694,
+      "step": 6303
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.081339734689818e-07,
+      "loss": 0.5575,
+      "step": 6304
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.065613807242654e-07,
+      "loss": 0.6879,
+      "step": 6305
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.049907603238147e-07,
+      "loss": 0.653,
+      "step": 6306
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.034221125974126e-07,
+      "loss": 0.6532,
+      "step": 6307
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.018554378744324e-07,
+      "loss": 0.6583,
+      "step": 6308
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.002907364838262e-07,
+      "loss": 0.6468,
+      "step": 6309
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.987280087541336e-07,
+      "loss": 0.6221,
+      "step": 6310
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.971672550134855e-07,
+      "loss": 0.6291,
+      "step": 6311
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.956084755895897e-07,
+      "loss": 0.6774,
+      "step": 6312
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.940516708097455e-07,
+      "loss": 0.6389,
+      "step": 6313
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.924968410008391e-07,
+      "loss": 0.6766,
+      "step": 6314
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.90943986489333e-07,
+      "loss": 0.6646,
+      "step": 6315
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.89393107601286e-07,
+      "loss": 0.6556,
+      "step": 6316
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.878442046623333e-07,
+      "loss": 0.7051,
+      "step": 6317
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.862972779976994e-07,
+      "loss": 0.6608,
+      "step": 6318
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.847523279321942e-07,
+      "loss": 0.6615,
+      "step": 6319
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.832093547902096e-07,
+      "loss": 0.6865,
+      "step": 6320
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.816683588957262e-07,
+      "loss": 0.6209,
+      "step": 6321
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.801293405723074e-07,
+      "loss": 0.6642,
+      "step": 6322
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.785923001431015e-07,
+      "loss": 0.6869,
+      "step": 6323
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.770572379308393e-07,
+      "loss": 0.6766,
+      "step": 6324
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.755241542578426e-07,
+      "loss": 0.6497,
+      "step": 6325
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.739930494460088e-07,
+      "loss": 0.6755,
+      "step": 6326
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.724639238168272e-07,
+      "loss": 0.642,
+      "step": 6327
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.709367776913699e-07,
+      "loss": 0.6785,
+      "step": 6328
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.694116113902903e-07,
+      "loss": 0.7074,
+      "step": 6329
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.678884252338312e-07,
+      "loss": 0.6599,
+      "step": 6330
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.663672195418147e-07,
+      "loss": 0.6793,
+      "step": 6331
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.648479946336471e-07,
+      "loss": 0.6395,
+      "step": 6332
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.633307508283237e-07,
+      "loss": 0.6361,
+      "step": 6333
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.618154884444182e-07,
+      "loss": 0.6555,
+      "step": 6334
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.60302207800093e-07,
+      "loss": 0.6452,
+      "step": 6335
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.587909092130927e-07,
+      "loss": 0.6582,
+      "step": 6336
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.572815930007447e-07,
+      "loss": 0.6833,
+      "step": 6337
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.557742594799587e-07,
+      "loss": 0.6085,
+      "step": 6338
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.542689089672326e-07,
+      "loss": 0.5524,
+      "step": 6339
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.527655417786437e-07,
+      "loss": 0.6267,
+      "step": 6340
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.512641582298572e-07,
+      "loss": 0.6216,
+      "step": 6341
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.497647586361165e-07,
+      "loss": 0.6565,
+      "step": 6342
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.48267343312252e-07,
+      "loss": 0.7034,
+      "step": 6343
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.467719125726778e-07,
+      "loss": 0.6399,
+      "step": 6344
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.452784667313904e-07,
+      "loss": 0.6551,
+      "step": 6345
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.437870061019657e-07,
+      "loss": 0.6529,
+      "step": 6346
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.422975309975698e-07,
+      "loss": 0.6841,
+      "step": 6347
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.40810041730947e-07,
+      "loss": 0.6294,
+      "step": 6348
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.393245386144252e-07,
+      "loss": 0.6984,
+      "step": 6349
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.378410219599195e-07,
+      "loss": 0.6964,
+      "step": 6350
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.363594920789206e-07,
+      "loss": 0.5306,
+      "step": 6351
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.348799492825096e-07,
+      "loss": 0.6685,
+      "step": 6352
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.334023938813448e-07,
+      "loss": 0.6357,
+      "step": 6353
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.319268261856669e-07,
+      "loss": 0.577,
+      "step": 6354
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.304532465053058e-07,
+      "loss": 0.6164,
+      "step": 6355
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.289816551496663e-07,
+      "loss": 0.6179,
+      "step": 6356
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.275120524277388e-07,
+      "loss": 0.6492,
+      "step": 6357
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.260444386481012e-07,
+      "loss": 0.6024,
+      "step": 6358
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.245788141189045e-07,
+      "loss": 0.5654,
+      "step": 6359
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.231151791478862e-07,
+      "loss": 0.6025,
+      "step": 6360
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.216535340423701e-07,
+      "loss": 0.5833,
+      "step": 6361
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.201938791092543e-07,
+      "loss": 0.6769,
+      "step": 6362
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.187362146550257e-07,
+      "loss": 0.681,
+      "step": 6363
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.172805409857506e-07,
+      "loss": 0.6805,
+      "step": 6364
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.158268584070758e-07,
+      "loss": 0.671,
+      "step": 6365
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.143751672242337e-07,
+      "loss": 0.6708,
+      "step": 6366
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.129254677420359e-07,
+      "loss": 0.6947,
+      "step": 6367
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.114777602648746e-07,
+      "loss": 0.7131,
+      "step": 6368
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.100320450967277e-07,
+      "loss": 0.6148,
+      "step": 6369
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.085883225411514e-07,
+      "loss": 0.6264,
+      "step": 6370
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.071465929012853e-07,
+      "loss": 0.6914,
+      "step": 6371
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.057068564798506e-07,
+      "loss": 0.6609,
+      "step": 6372
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.042691135791489e-07,
+      "loss": 0.6405,
+      "step": 6373
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.028333645010631e-07,
+      "loss": 0.6255,
+      "step": 6374
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.013996095470586e-07,
+      "loss": 0.5888,
+      "step": 6375
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.999678490181803e-07,
+      "loss": 0.6749,
+      "step": 6376
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.985380832150577e-07,
+      "loss": 0.7025,
+      "step": 6377
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.971103124378973e-07,
+      "loss": 0.6582,
+      "step": 6378
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.956845369864893e-07,
+      "loss": 0.6122,
+      "step": 6379
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.942607571602065e-07,
+      "loss": 0.6275,
+      "step": 6380
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.928389732579986e-07,
+      "loss": 0.6633,
+      "step": 6381
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.914191855783968e-07,
+      "loss": 0.6474,
+      "step": 6382
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.900013944195181e-07,
+      "loss": 0.674,
+      "step": 6383
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.885856000790524e-07,
+      "loss": 0.6427,
+      "step": 6384
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.871718028542772e-07,
+      "loss": 0.6386,
+      "step": 6385
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.857600030420495e-07,
+      "loss": 0.6579,
+      "step": 6386
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.843502009388035e-07,
+      "loss": 0.6671,
+      "step": 6387
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.829423968405544e-07,
+      "loss": 0.615,
+      "step": 6388
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.815365910429037e-07,
+      "loss": 0.6785,
+      "step": 6389
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.801327838410241e-07,
+      "loss": 0.6538,
+      "step": 6390
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.787309755296787e-07,
+      "loss": 0.6117,
+      "step": 6391
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.77331166403201e-07,
+      "loss": 0.5792,
+      "step": 6392
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7593335675551135e-07,
+      "loss": 0.6679,
+      "step": 6393
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7453754688011054e-07,
+      "loss": 0.6698,
+      "step": 6394
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7314373707007643e-07,
+      "loss": 0.6869,
+      "step": 6395
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.717519276180649e-07,
+      "loss": 0.6325,
+      "step": 6396
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7036211881631767e-07,
+      "loss": 0.6567,
+      "step": 6397
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6897431095665247e-07,
+      "loss": 0.5981,
+      "step": 6398
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.675885043304673e-07,
+      "loss": 0.5807,
+      "step": 6399
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6620469922874276e-07,
+      "loss": 0.6299,
+      "step": 6400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.648228959420342e-07,
+      "loss": 0.672,
+      "step": 6401
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.634430947604829e-07,
+      "loss": 0.6401,
+      "step": 6402
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.620652959738037e-07,
+      "loss": 0.6348,
+      "step": 6403
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.60689499871293e-07,
+      "loss": 0.6502,
+      "step": 6404
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5931570674182966e-07,
+      "loss": 0.653,
+      "step": 6405
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.579439168738664e-07,
+      "loss": 0.5741,
+      "step": 6406
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5657413055544166e-07,
+      "loss": 0.6431,
+      "step": 6407
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5520634807416975e-07,
+      "loss": 0.6806,
+      "step": 6408
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.538405697172432e-07,
+      "loss": 0.6787,
+      "step": 6409
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.524767957714349e-07,
+      "loss": 0.6625,
+      "step": 6410
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.511150265230979e-07,
+      "loss": 0.605,
+      "step": 6411
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4975526225816356e-07,
+      "loss": 0.6906,
+      "step": 6412
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4839750326214125e-07,
+      "loss": 0.6754,
+      "step": 6413
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.470417498201207e-07,
+      "loss": 0.625,
+      "step": 6414
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.456880022167698e-07,
+      "loss": 0.6472,
+      "step": 6415
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.443362607363377e-07,
+      "loss": 0.6408,
+      "step": 6416
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4298652566264755e-07,
+      "loss": 0.6827,
+      "step": 6417
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4163879727910255e-07,
+      "loss": 0.6538,
+      "step": 6418
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4029307586869075e-07,
+      "loss": 0.6293,
+      "step": 6419
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.389493617139684e-07,
+      "loss": 0.6738,
+      "step": 6420
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.376076550970776e-07,
+      "loss": 0.6555,
+      "step": 6421
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3626795629973963e-07,
+      "loss": 0.6233,
+      "step": 6422
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3493026560324945e-07,
+      "loss": 0.6775,
+      "step": 6423
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.335945832884814e-07,
+      "loss": 0.7108,
+      "step": 6424
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.322609096358921e-07,
+      "loss": 0.6933,
+      "step": 6425
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3092924492551094e-07,
+      "loss": 0.6752,
+      "step": 6426
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.295995894369498e-07,
+      "loss": 0.6311,
+      "step": 6427
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2827194344939536e-07,
+      "loss": 0.5605,
+      "step": 6428
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2694630724161466e-07,
+      "loss": 0.6284,
+      "step": 6429
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.25622681091955e-07,
+      "loss": 0.6478,
+      "step": 6430
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2430106527833413e-07,
+      "loss": 0.6576,
+      "step": 6431
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2298146007825446e-07,
+      "loss": 0.6755,
+      "step": 6432
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.216638657687944e-07,
+      "loss": 0.6846,
+      "step": 6433
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2034828262660807e-07,
+      "loss": 0.6461,
+      "step": 6434
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1903471092793137e-07,
+      "loss": 0.6421,
+      "step": 6435
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.177231509485746e-07,
+      "loss": 0.6434,
+      "step": 6436
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.164136029639265e-07,
+      "loss": 0.5821,
+      "step": 6437
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1510606724895266e-07,
+      "loss": 0.5629,
+      "step": 6438
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.138005440781978e-07,
+      "loss": 0.6906,
+      "step": 6439
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.124970337257828e-07,
+      "loss": 0.6588,
+      "step": 6440
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.111955364654063e-07,
+      "loss": 0.6413,
+      "step": 6441
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0989605257034435e-07,
+      "loss": 0.5634,
+      "step": 6442
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.085985823134486e-07,
+      "loss": 0.6022,
+      "step": 6443
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.073031259671534e-07,
+      "loss": 0.6762,
+      "step": 6444
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.060096838034633e-07,
+      "loss": 0.6723,
+      "step": 6445
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.047182560939611e-07,
+      "loss": 0.6513,
+      "step": 6446
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.034288431098132e-07,
+      "loss": 0.6995,
+      "step": 6447
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0214144512175424e-07,
+      "loss": 0.6179,
+      "step": 6448
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.008560624001012e-07,
+      "loss": 0.6719,
+      "step": 6449
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9957269521474716e-07,
+      "loss": 0.6514,
+      "step": 6450
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9829134383515877e-07,
+      "loss": 0.5988,
+      "step": 6451
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9701200853038637e-07,
+      "loss": 0.6835,
+      "step": 6452
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9573468956904835e-07,
+      "loss": 0.6354,
+      "step": 6453
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9445938721934474e-07,
+      "loss": 0.644,
+      "step": 6454
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9318610174905345e-07,
+      "loss": 0.6474,
+      "step": 6455
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9191483342552296e-07,
+      "loss": 0.6773,
+      "step": 6456
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.906455825156852e-07,
+      "loss": 0.6461,
+      "step": 6457
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.893783492860459e-07,
+      "loss": 0.634,
+      "step": 6458
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8811313400268444e-07,
+      "loss": 0.6529,
+      "step": 6459
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.868499369312584e-07,
+      "loss": 0.6672,
+      "step": 6460
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.855887583370033e-07,
+      "loss": 0.5583,
+      "step": 6461
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.843295984847273e-07,
+      "loss": 0.6492,
+      "step": 6462
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8307245763882004e-07,
+      "loss": 0.6767,
+      "step": 6463
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8181733606324047e-07,
+      "loss": 0.5814,
+      "step": 6464
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8056423402152874e-07,
+      "loss": 0.661,
+      "step": 6465
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7931315177680007e-07,
+      "loss": 0.5616,
+      "step": 6466
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7806408959174314e-07,
+      "loss": 0.6302,
+      "step": 6467
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.768170477286237e-07,
+      "loss": 0.6616,
+      "step": 6468
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.755720264492857e-07,
+      "loss": 0.5651,
+      "step": 6469
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.743290260151444e-07,
+      "loss": 0.6485,
+      "step": 6470
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7308804668719446e-07,
+      "loss": 0.6181,
+      "step": 6471
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7184908872600513e-07,
+      "loss": 0.6758,
+      "step": 6472
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7061215239172164e-07,
+      "loss": 0.5598,
+      "step": 6473
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6937723794406076e-07,
+      "loss": 0.559,
+      "step": 6474
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.681443456423217e-07,
+      "loss": 0.6449,
+      "step": 6475
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6691347574537295e-07,
+      "loss": 0.6487,
+      "step": 6476
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.656846285116622e-07,
+      "loss": 0.6412,
+      "step": 6477
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.644578041992097e-07,
+      "loss": 0.641,
+      "step": 6478
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.632330030656128e-07,
+      "loss": 0.5867,
+      "step": 6479
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6201022536804573e-07,
+      "loss": 0.5646,
+      "step": 6480
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6078947136325207e-07,
+      "loss": 0.6524,
+      "step": 6481
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.595707413075555e-07,
+      "loss": 0.6658,
+      "step": 6482
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.583540354568538e-07,
+      "loss": 0.6895,
+      "step": 6483
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5713935406661905e-07,
+      "loss": 0.6384,
+      "step": 6484
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5592669739189734e-07,
+      "loss": 0.6738,
+      "step": 6485
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5471606568731166e-07,
+      "loss": 0.6571,
+      "step": 6486
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5350745920705973e-07,
+      "loss": 0.6501,
+      "step": 6487
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.52300878204912e-07,
+      "loss": 0.6892,
+      "step": 6488
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5109632293421457e-07,
+      "loss": 0.6573,
+      "step": 6489
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.498937936478886e-07,
+      "loss": 0.6597,
+      "step": 6490
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4869329059843085e-07,
+      "loss": 0.6482,
+      "step": 6491
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4749481403790863e-07,
+      "loss": 0.681,
+      "step": 6492
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.462983642179696e-07,
+      "loss": 0.6679,
+      "step": 6493
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.451039413898316e-07,
+      "loss": 0.6645,
+      "step": 6494
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4391154580428855e-07,
+      "loss": 0.6925,
+      "step": 6495
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4272117771170675e-07,
+      "loss": 0.6168,
+      "step": 6496
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.415328373620308e-07,
+      "loss": 0.6666,
+      "step": 6497
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.403465250047744e-07,
+      "loss": 0.7198,
+      "step": 6498
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.391622408890305e-07,
+      "loss": 0.6704,
+      "step": 6499
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.379799852634613e-07,
+      "loss": 0.5502,
+      "step": 6500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3679975837630605e-07,
+      "loss": 0.6437,
+      "step": 6501
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3562156047537986e-07,
+      "loss": 0.6025,
+      "step": 6502
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.344453918080681e-07,
+      "loss": 0.6331,
+      "step": 6503
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3327125262132996e-07,
+      "loss": 0.6568,
+      "step": 6504
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3209914316170153e-07,
+      "loss": 0.6776,
+      "step": 6505
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3092906367528934e-07,
+      "loss": 0.6649,
+      "step": 6506
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2976101440777784e-07,
+      "loss": 0.679,
+      "step": 6507
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.285949956044221e-07,
+      "loss": 0.6543,
+      "step": 6508
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2743100751005064e-07,
+      "loss": 0.552,
+      "step": 6509
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.262690503690669e-07,
+      "loss": 0.6504,
+      "step": 6510
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2510912442544896e-07,
+      "loss": 0.6787,
+      "step": 6511
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.239512299227432e-07,
+      "loss": 0.6679,
+      "step": 6512
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.227953671040784e-07,
+      "loss": 0.5761,
+      "step": 6513
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.216415362121472e-07,
+      "loss": 0.6738,
+      "step": 6514
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.204897374892213e-07,
+      "loss": 0.6268,
+      "step": 6515
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.193399711771472e-07,
+      "loss": 0.6637,
+      "step": 6516
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.181922375173374e-07,
+      "loss": 0.566,
+      "step": 6517
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1704653675078357e-07,
+      "loss": 0.6188,
+      "step": 6518
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1590286911805103e-07,
+      "loss": 0.6401,
+      "step": 6519
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1476123485927323e-07,
+      "loss": 0.6647,
+      "step": 6520
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.136216342141607e-07,
+      "loss": 0.687,
+      "step": 6521
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1248406742199646e-07,
+      "loss": 0.6654,
+      "step": 6522
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1134853472163604e-07,
+      "loss": 0.7024,
+      "step": 6523
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1021503635150663e-07,
+      "loss": 0.6637,
+      "step": 6524
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0908357254961e-07,
+      "loss": 0.6317,
+      "step": 6525
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.079541435535194e-07,
+      "loss": 0.6321,
+      "step": 6526
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0682674960038295e-07,
+      "loss": 0.6605,
+      "step": 6527
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0570139092691795e-07,
+      "loss": 0.6265,
+      "step": 6528
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.045780677694177e-07,
+      "loss": 0.5396,
+      "step": 6529
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0345678036374894e-07,
+      "loss": 0.6551,
+      "step": 6530
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0233752894534454e-07,
+      "loss": 0.649,
+      "step": 6531
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0122031374921643e-07,
+      "loss": 0.6515,
+      "step": 6532
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.001051350099482e-07,
+      "loss": 0.6182,
+      "step": 6533
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9899199296169133e-07,
+      "loss": 0.6786,
+      "step": 6534
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9788088783817447e-07,
+      "loss": 0.6661,
+      "step": 6535
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.967718198726965e-07,
+      "loss": 0.5802,
+      "step": 6536
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9566478929813013e-07,
+      "loss": 0.6633,
+      "step": 6537
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9455979634691713e-07,
+      "loss": 0.5883,
+      "step": 6538
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.934568412510741e-07,
+      "loss": 0.6547,
+      "step": 6539
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9235592424218916e-07,
+      "loss": 0.6589,
+      "step": 6540
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.912570455514219e-07,
+      "loss": 0.6891,
+      "step": 6541
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.901602054095043e-07,
+      "loss": 0.6819,
+      "step": 6542
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.890654040467411e-07,
+      "loss": 0.5491,
+      "step": 6543
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.879726416930084e-07,
+      "loss": 0.6389,
+      "step": 6544
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.868819185777527e-07,
+      "loss": 0.5773,
+      "step": 6545
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.857932349299941e-07,
+      "loss": 0.65,
+      "step": 6546
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.847065909783242e-07,
+      "loss": 0.6801,
+      "step": 6547
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.83621986950906e-07,
+      "loss": 0.6692,
+      "step": 6548
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.825394230754741e-07,
+      "loss": 0.6817,
+      "step": 6549
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.814588995793355e-07,
+      "loss": 0.6147,
+      "step": 6550
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8038041668936645e-07,
+      "loss": 0.6344,
+      "step": 6551
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.793039746320203e-07,
+      "loss": 0.6549,
+      "step": 6552
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7822957363331294e-07,
+      "loss": 0.659,
+      "step": 6553
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7715721391883945e-07,
+      "loss": 0.5924,
+      "step": 6554
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.760868957137652e-07,
+      "loss": 0.5271,
+      "step": 6555
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.750186192428217e-07,
+      "loss": 0.6574,
+      "step": 6556
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7395238473031827e-07,
+      "loss": 0.6593,
+      "step": 6557
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7288819240013144e-07,
+      "loss": 0.5505,
+      "step": 6558
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7182604247571134e-07,
+      "loss": 0.6541,
+      "step": 6559
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.707659351800762e-07,
+      "loss": 0.5712,
+      "step": 6560
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6970787073581916e-07,
+      "loss": 0.6779,
+      "step": 6561
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6865184936510026e-07,
+      "loss": 0.5595,
+      "step": 6562
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.675978712896543e-07,
+      "loss": 0.6451,
+      "step": 6563
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6654593673078545e-07,
+      "loss": 0.6583,
+      "step": 6564
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6549604590936694e-07,
+      "loss": 0.6502,
+      "step": 6565
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.644481990458492e-07,
+      "loss": 0.678,
+      "step": 6566
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.634023963602439e-07,
+      "loss": 0.6917,
+      "step": 6567
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.62358638072141e-07,
+      "loss": 0.6607,
+      "step": 6568
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6131692440070077e-07,
+      "loss": 0.6569,
+      "step": 6569
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.602772555646482e-07,
+      "loss": 0.6382,
+      "step": 6570
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.592396317822854e-07,
+      "loss": 0.6294,
+      "step": 6571
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5820405327148354e-07,
+      "loss": 0.6427,
+      "step": 6572
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5717052024968324e-07,
+      "loss": 0.6957,
+      "step": 6573
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.561390329338931e-07,
+      "loss": 0.6392,
+      "step": 6574
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5510959154069756e-07,
+      "loss": 0.5748,
+      "step": 6575
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.540821962862472e-07,
+      "loss": 0.6515,
+      "step": 6576
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.530568473862671e-07,
+      "loss": 0.5241,
+      "step": 6577
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.520335450560474e-07,
+      "loss": 0.6722,
+      "step": 6578
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5101228951045277e-07,
+      "loss": 0.5339,
+      "step": 6579
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.499930809639184e-07,
+      "loss": 0.6399,
+      "step": 6580
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.489759196304442e-07,
+      "loss": 0.6216,
+      "step": 6581
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.479608057236049e-07,
+      "loss": 0.6872,
+      "step": 6582
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.469477394565467e-07,
+      "loss": 0.6045,
+      "step": 6583
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.459367210419816e-07,
+      "loss": 0.6425,
+      "step": 6584
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4492775069219543e-07,
+      "loss": 0.6733,
+      "step": 6585
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4392082861903864e-07,
+      "loss": 0.5801,
+      "step": 6586
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.429159550339388e-07,
+      "loss": 0.5433,
+      "step": 6587
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.419131301478872e-07,
+      "loss": 0.6831,
+      "step": 6588
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4091235417144976e-07,
+      "loss": 0.6579,
+      "step": 6589
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.399136273147562e-07,
+      "loss": 0.6309,
+      "step": 6590
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3891694978751323e-07,
+      "loss": 0.6794,
+      "step": 6591
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3792232179899123e-07,
+      "loss": 0.6358,
+      "step": 6592
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.369297435580342e-07,
+      "loss": 0.6602,
+      "step": 6593
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.359392152730544e-07,
+      "loss": 0.6187,
+      "step": 6594
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.349507371520321e-07,
+      "loss": 0.5834,
+      "step": 6595
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3396430940251903e-07,
+      "loss": 0.6908,
+      "step": 6596
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3297993223163617e-07,
+      "loss": 0.6806,
+      "step": 6597
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3199760584607157e-07,
+      "loss": 0.6724,
+      "step": 6598
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3101733045208908e-07,
+      "loss": 0.641,
+      "step": 6599
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3003910625551297e-07,
+      "loss": 0.5917,
+      "step": 6600
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2906293346174335e-07,
+      "loss": 0.6746,
+      "step": 6601
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2808881227574965e-07,
+      "loss": 0.5527,
+      "step": 6602
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2711674290206486e-07,
+      "loss": 0.6619,
+      "step": 6603
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2614672554479576e-07,
+      "loss": 0.6685,
+      "step": 6604
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2517876040761942e-07,
+      "loss": 0.5289,
+      "step": 6605
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.242128476937777e-07,
+      "loss": 0.6473,
+      "step": 6606
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2324898760608394e-07,
+      "loss": 0.6462,
+      "step": 6607
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2228718034692175e-07,
+      "loss": 0.6479,
+      "step": 6608
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2132742611824076e-07,
+      "loss": 0.665,
+      "step": 6609
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2036972512156195e-07,
+      "loss": 0.6663,
+      "step": 6610
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.194140775579734e-07,
+      "loss": 0.6615,
+      "step": 6611
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1846048362813233e-07,
+      "loss": 0.6231,
+      "step": 6612
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1750894353226637e-07,
+      "loss": 0.6408,
+      "step": 6613
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1655945747017017e-07,
+      "loss": 0.66,
+      "step": 6614
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1561202564120865e-07,
+      "loss": 0.6539,
+      "step": 6615
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1466664824431495e-07,
+      "loss": 0.6771,
+      "step": 6616
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1372332547798691e-07,
+      "loss": 0.6804,
+      "step": 6617
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1278205754029725e-07,
+      "loss": 0.6361,
+      "step": 6618
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.118428446288845e-07,
+      "loss": 0.6617,
+      "step": 6619
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1090568694095426e-07,
+      "loss": 0.6171,
+      "step": 6620
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0997058467328356e-07,
+      "loss": 0.6116,
+      "step": 6621
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0903753802221427e-07,
+      "loss": 0.5875,
+      "step": 6622
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0810654718366075e-07,
+      "loss": 0.6561,
+      "step": 6623
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.071776123531011e-07,
+      "loss": 0.5372,
+      "step": 6624
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0625073372558592e-07,
+      "loss": 0.6121,
+      "step": 6625
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0532591149573177e-07,
+      "loss": 0.6523,
+      "step": 6626
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0440314585772447e-07,
+      "loss": 0.6604,
+      "step": 6627
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.034824370053168e-07,
+      "loss": 0.6711,
+      "step": 6628
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0256378513182961e-07,
+      "loss": 0.565,
+      "step": 6629
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0164719043015645e-07,
+      "loss": 0.6214,
+      "step": 6630
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0073265309275004e-07,
+      "loss": 0.665,
+      "step": 6631
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9982017331163784e-07,
+      "loss": 0.6778,
+      "step": 6632
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.989097512784166e-07,
+      "loss": 0.6484,
+      "step": 6633
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9800138718424343e-07,
+      "loss": 0.6728,
+      "step": 6634
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9709508121985133e-07,
+      "loss": 0.6987,
+      "step": 6635
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9619083357553582e-07,
+      "loss": 0.6352,
+      "step": 6636
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.952886444411639e-07,
+      "loss": 0.6662,
+      "step": 6637
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9438851400616742e-07,
+      "loss": 0.6736,
+      "step": 6638
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.934904424595463e-07,
+      "loss": 0.6718,
+      "step": 6639
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9259442998986965e-07,
+      "loss": 0.6785,
+      "step": 6640
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.917004767852748e-07,
+      "loss": 0.6569,
+      "step": 6641
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9080858303346494e-07,
+      "loss": 0.5435,
+      "step": 6642
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8991874892171024e-07,
+      "loss": 0.6766,
+      "step": 6643
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8903097463685128e-07,
+      "loss": 0.6512,
+      "step": 6644
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8814526036529335e-07,
+      "loss": 0.6721,
+      "step": 6645
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.872616062930088e-07,
+      "loss": 0.6187,
+      "step": 6646
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.863800126055426e-07,
+      "loss": 0.6373,
+      "step": 6647
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8550047948799998e-07,
+      "loss": 0.6638,
+      "step": 6648
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8462300712505877e-07,
+      "loss": 0.6433,
+      "step": 6649
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8374759570096045e-07,
+      "loss": 0.6583,
+      "step": 6650
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8287424539951692e-07,
+      "loss": 0.6585,
+      "step": 6651
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8200295640410702e-07,
+      "loss": 0.6652,
+      "step": 6652
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8113372889767223e-07,
+      "loss": 0.6224,
+      "step": 6653
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8026656306272762e-07,
+      "loss": 0.6918,
+      "step": 6654
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7940145908135087e-07,
+      "loss": 0.5893,
+      "step": 6655
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7853841713518893e-07,
+      "loss": 0.6377,
+      "step": 6656
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7767743740545573e-07,
+      "loss": 0.6864,
+      "step": 6657
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7681852007292887e-07,
+      "loss": 0.5651,
+      "step": 6658
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7596166531795854e-07,
+      "loss": 0.6821,
+      "step": 6659
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.751068733204575e-07,
+      "loss": 0.6635,
+      "step": 6660
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7425414425990773e-07,
+      "loss": 0.5943,
+      "step": 6661
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7340347831535597e-07,
+      "loss": 0.6551,
+      "step": 6662
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7255487566541717e-07,
+      "loss": 0.5918,
+      "step": 6663
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7170833648827322e-07,
+      "loss": 0.646,
+      "step": 6664
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7086386096167085e-07,
+      "loss": 0.5818,
+      "step": 6665
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7002144926292818e-07,
+      "loss": 0.6401,
+      "step": 6666
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6918110156892377e-07,
+      "loss": 0.6648,
+      "step": 6667
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6834281805610642e-07,
+      "loss": 0.696,
+      "step": 6668
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6750659890049205e-07,
+      "loss": 0.6739,
+      "step": 6669
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6667244427766017e-07,
+      "loss": 0.7022,
+      "step": 6670
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.658403543627596e-07,
+      "loss": 0.6066,
+      "step": 6671
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6501032933050497e-07,
+      "loss": 0.621,
+      "step": 6672
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6418236935517695e-07,
+      "loss": 0.6099,
+      "step": 6673
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6335647461062197e-07,
+      "loss": 0.6573,
+      "step": 6674
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6253264527025248e-07,
+      "loss": 0.6654,
+      "step": 6675
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6171088150705005e-07,
+      "loss": 0.5723,
+      "step": 6676
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6089118349356113e-07,
+      "loss": 0.6526,
+      "step": 6677
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6007355140189585e-07,
+      "loss": 0.6458,
+      "step": 6678
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.592579854037346e-07,
+      "loss": 0.6699,
+      "step": 6679
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5844448567032266e-07,
+      "loss": 0.6484,
+      "step": 6680
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5763305237246894e-07,
+      "loss": 0.6864,
+      "step": 6681
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.568236856805505e-07,
+      "loss": 0.6684,
+      "step": 6682
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.560163857645136e-07,
+      "loss": 0.5913,
+      "step": 6683
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5521115279386267e-07,
+      "loss": 0.5793,
+      "step": 6684
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.544079869376769e-07,
+      "loss": 0.6194,
+      "step": 6685
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5360688836459358e-07,
+      "loss": 0.667,
+      "step": 6686
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5280785724282376e-07,
+      "loss": 0.6556,
+      "step": 6687
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5201089374013766e-07,
+      "loss": 0.6622,
+      "step": 6688
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5121599802387478e-07,
+      "loss": 0.6003,
+      "step": 6689
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5042317026093822e-07,
+      "loss": 0.6855,
+      "step": 6690
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4963241061780154e-07,
+      "loss": 0.6545,
+      "step": 6691
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4884371926049857e-07,
+      "loss": 0.6589,
+      "step": 6692
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4805709635463127e-07,
+      "loss": 0.6336,
+      "step": 6693
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4727254206536866e-07,
+      "loss": 0.6691,
+      "step": 6694
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.464900565574423e-07,
+      "loss": 0.6449,
+      "step": 6695
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4570963999515075e-07,
+      "loss": 0.584,
+      "step": 6696
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4493129254236072e-07,
+      "loss": 0.6609,
+      "step": 6697
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4415501436249924e-07,
+      "loss": 0.614,
+      "step": 6698
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4338080561856483e-07,
+      "loss": 0.634,
+      "step": 6699
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.426086664731152e-07,
+      "loss": 0.605,
+      "step": 6700
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.418385970882785e-07,
+      "loss": 0.6666,
+      "step": 6701
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4107059762574871e-07,
+      "loss": 0.6231,
+      "step": 6702
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.403046682467779e-07,
+      "loss": 0.6654,
+      "step": 6703
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.395408091121919e-07,
+      "loss": 0.705,
+      "step": 6704
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3877902038237912e-07,
+      "loss": 0.694,
+      "step": 6705
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3801930221729042e-07,
+      "loss": 0.6141,
+      "step": 6706
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3726165477644603e-07,
+      "loss": 0.6807,
+      "step": 6707
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3650607821892759e-07,
+      "loss": 0.678,
+      "step": 6708
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3575257270338592e-07,
+      "loss": 0.6411,
+      "step": 6709
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3500113838803453e-07,
+      "loss": 0.6887,
+      "step": 6710
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3425177543065048e-07,
+      "loss": 0.6904,
+      "step": 6711
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3350448398857907e-07,
+      "loss": 0.6745,
+      "step": 6712
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.327592642187303e-07,
+      "loss": 0.6273,
+      "step": 6713
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3201611627757683e-07,
+      "loss": 0.6659,
+      "step": 6714
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3127504032115824e-07,
+      "loss": 0.6747,
+      "step": 6715
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3053603650508008e-07,
+      "loss": 0.6279,
+      "step": 6716
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.297991049845082e-07,
+      "loss": 0.5273,
+      "step": 6717
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2906424591417888e-07,
+      "loss": 0.6617,
+      "step": 6718
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2833145944838977e-07,
+      "loss": 0.6365,
+      "step": 6719
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2760074574100444e-07,
+      "loss": 0.6546,
+      "step": 6720
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2687210494545244e-07,
+      "loss": 0.6669,
+      "step": 6721
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2614553721472355e-07,
+      "loss": 0.6587,
+      "step": 6722
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.254210427013791e-07,
+      "loss": 0.6427,
+      "step": 6723
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2469862155754075e-07,
+      "loss": 0.6455,
+      "step": 6724
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.239782739348938e-07,
+      "loss": 0.579,
+      "step": 6725
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2325999998469063e-07,
+      "loss": 0.5997,
+      "step": 6726
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2254379985774834e-07,
+      "loss": 0.645,
+      "step": 6727
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2182967370444665e-07,
+      "loss": 0.6206,
+      "step": 6728
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.211176216747323e-07,
+      "loss": 0.6552,
+      "step": 6729
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2040764391811566e-07,
+      "loss": 0.6656,
+      "step": 6730
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1969974058366862e-07,
+      "loss": 0.6686,
+      "step": 6731
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1899391182003006e-07,
+      "loss": 0.563,
+      "step": 6732
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1829015777540587e-07,
+      "loss": 0.6177,
+      "step": 6733
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1758847859756117e-07,
+      "loss": 0.6526,
+      "step": 6734
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1688887443382923e-07,
+      "loss": 0.6956,
+      "step": 6735
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1619134543110477e-07,
+      "loss": 0.6622,
+      "step": 6736
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1549589173585063e-07,
+      "loss": 0.615,
+      "step": 6737
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1480251349408888e-07,
+      "loss": 0.6411,
+      "step": 6738
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1411121085140975e-07,
+      "loss": 0.6411,
+      "step": 6739
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.13421983952966e-07,
+      "loss": 0.6691,
+      "step": 6740
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.127348329434752e-07,
+      "loss": 0.6674,
+      "step": 6741
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1204975796721862e-07,
+      "loss": 0.6626,
+      "step": 6742
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1136675916804118e-07,
+      "loss": 0.6105,
+      "step": 6743
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1068583668935373e-07,
+      "loss": 0.6847,
+      "step": 6744
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1000699067412857e-07,
+      "loss": 0.5996,
+      "step": 6745
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0933022126490388e-07,
+      "loss": 0.5515,
+      "step": 6746
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0865552860378159e-07,
+      "loss": 0.6895,
+      "step": 6747
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0798291283242612e-07,
+      "loss": 0.6336,
+      "step": 6748
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0731237409206784e-07,
+      "loss": 0.5521,
+      "step": 6749
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0664391252349971e-07,
+      "loss": 0.6784,
+      "step": 6750
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.059775282670783e-07,
+      "loss": 0.6405,
+      "step": 6751
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0531322146272615e-07,
+      "loss": 0.6609,
+      "step": 6752
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0465099224992614e-07,
+      "loss": 0.62,
+      "step": 6753
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0399084076772813e-07,
+      "loss": 0.688,
+      "step": 6754
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0333276715474461e-07,
+      "loss": 0.5993,
+      "step": 6755
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0267677154915056e-07,
+      "loss": 0.6506,
+      "step": 6756
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0202285408868584e-07,
+      "loss": 0.6929,
+      "step": 6757
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0137101491065394e-07,
+      "loss": 0.6376,
+      "step": 6758
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0072125415192202e-07,
+      "loss": 0.6438,
+      "step": 6759
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0007357194892098e-07,
+      "loss": 0.5723,
+      "step": 6760
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.942796843764424e-08,
+      "loss": 0.6714,
+      "step": 6761
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.878444375364893e-08,
+      "loss": 0.6634,
+      "step": 6762
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.814299803205807e-08,
+      "loss": 0.6076,
+      "step": 6763
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.750363140755392e-08,
+      "loss": 0.6598,
+      "step": 6764
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.686634401438578e-08,
+      "loss": 0.6269,
+      "step": 6765
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.623113598636658e-08,
+      "loss": 0.6871,
+      "step": 6766
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.559800745686853e-08,
+      "loss": 0.6738,
+      "step": 6767
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.496695855883087e-08,
+      "loss": 0.6559,
+      "step": 6768
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.433798942475647e-08,
+      "loss": 0.6505,
+      "step": 6769
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.371110018670859e-08,
+      "loss": 0.5533,
+      "step": 6770
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.308629097631639e-08,
+      "loss": 0.6668,
+      "step": 6771
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.246356192477158e-08,
+      "loss": 0.6665,
+      "step": 6772
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.184291316282734e-08,
+      "loss": 0.6383,
+      "step": 6773
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.122434482080277e-08,
+      "loss": 0.6662,
+      "step": 6774
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.060785702857844e-08,
+      "loss": 0.6471,
+      "step": 6775
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.999344991559855e-08,
+      "loss": 0.6222,
+      "step": 6776
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.938112361087215e-08,
+      "loss": 0.6624,
+      "step": 6777
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.877087824296637e-08,
+      "loss": 0.6969,
+      "step": 6778
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.816271394001763e-08,
+      "loss": 0.5806,
+      "step": 6779
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.755663082972043e-08,
+      "loss": 0.6271,
+      "step": 6780
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.69526290393352e-08,
+      "loss": 0.5713,
+      "step": 6781
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.635070869568385e-08,
+      "loss": 0.5814,
+      "step": 6782
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.575086992515413e-08,
+      "loss": 0.6773,
+      "step": 6783
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.515311285369088e-08,
+      "loss": 0.6549,
+      "step": 6784
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.455743760680812e-08,
+      "loss": 0.6595,
+      "step": 6785
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.396384430957916e-08,
+      "loss": 0.6599,
+      "step": 6786
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.337233308664094e-08,
+      "loss": 0.5791,
+      "step": 6787
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.278290406219303e-08,
+      "loss": 0.6482,
+      "step": 6788
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.219555735999862e-08,
+      "loss": 0.6623,
+      "step": 6789
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.161029310338241e-08,
+      "loss": 0.6803,
+      "step": 6790
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.1027111415235e-08,
+      "loss": 0.512,
+      "step": 6791
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.044601241800287e-08,
+      "loss": 0.5712,
+      "step": 6792
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.986699623370397e-08,
+      "loss": 0.6951,
+      "step": 6793
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.929006298391218e-08,
+      "loss": 0.6908,
+      "step": 6794
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.871521278976724e-08,
+      "loss": 0.5494,
+      "step": 6795
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.814244577197038e-08,
+      "loss": 0.6655,
+      "step": 6796
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.757176205078542e-08,
+      "loss": 0.6434,
+      "step": 6797
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.700316174603984e-08,
+      "loss": 0.565,
+      "step": 6798
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.643664497712255e-08,
+      "loss": 0.7085,
+      "step": 6799
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.58722118629851e-08,
+      "loss": 0.6346,
+      "step": 6800
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.530986252214157e-08,
+      "loss": 0.6377,
+      "step": 6801
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.474959707267082e-08,
+      "loss": 0.6432,
+      "step": 6802
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.419141563220766e-08,
+      "loss": 0.704,
+      "step": 6803
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.363531831795723e-08,
+      "loss": 0.5806,
+      "step": 6804
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.308130524668389e-08,
+      "loss": 0.6395,
+      "step": 6805
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.252937653471125e-08,
+      "loss": 0.5343,
+      "step": 6806
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.197953229792998e-08,
+      "loss": 0.6409,
+      "step": 6807
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.143177265178991e-08,
+      "loss": 0.6773,
+      "step": 6808
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.088609771130572e-08,
+      "loss": 0.6514,
+      "step": 6809
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.034250759105244e-08,
+      "loss": 0.5617,
+      "step": 6810
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.980100240516652e-08,
+      "loss": 0.6248,
+      "step": 6811
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.926158226735035e-08,
+      "loss": 0.6859,
+      "step": 6812
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.872424729086447e-08,
+      "loss": 0.6035,
+      "step": 6813
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.818899758853414e-08,
+      "loss": 0.5264,
+      "step": 6814
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.765583327274617e-08,
+      "loss": 0.6549,
+      "step": 6815
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.712475445544875e-08,
+      "loss": 0.6628,
+      "step": 6816
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.65957612481527e-08,
+      "loss": 0.6653,
+      "step": 6817
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.606885376193139e-08,
+      "loss": 0.6883,
+      "step": 6818
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.554403210741966e-08,
+      "loss": 0.6439,
+      "step": 6819
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.502129639481492e-08,
+      "loss": 0.592,
+      "step": 6820
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.450064673387491e-08,
+      "loss": 0.6541,
+      "step": 6821
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.398208323392218e-08,
+      "loss": 0.6589,
+      "step": 6822
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.346560600383855e-08,
+      "loss": 0.6394,
+      "step": 6823
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.295121515206948e-08,
+      "loss": 0.6603,
+      "step": 6824
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.24389107866219e-08,
+      "loss": 0.6749,
+      "step": 6825
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.192869301506421e-08,
+      "loss": 0.6439,
+      "step": 6826
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.142056194452739e-08,
+      "loss": 0.6562,
+      "step": 6827
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.091451768170386e-08,
+      "loss": 0.6652,
+      "step": 6828
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.041056033284864e-08,
+      "loss": 0.6587,
+      "step": 6829
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.990869000377708e-08,
+      "loss": 0.6619,
+      "step": 6830
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9408906799867105e-08,
+      "loss": 0.6746,
+      "step": 6831
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8911210826058107e-08,
+      "loss": 0.6796,
+      "step": 6832
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.841560218685205e-08,
+      "loss": 0.6431,
+      "step": 6833
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.792208098631125e-08,
+      "loss": 0.6451,
+      "step": 6834
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7430647328061694e-08,
+      "loss": 0.7024,
+      "step": 6835
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6941301315289744e-08,
+      "loss": 0.5985,
+      "step": 6836
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6454043050743204e-08,
+      "loss": 0.5788,
+      "step": 6837
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.596887263673245e-08,
+      "loss": 0.6067,
+      "step": 6838
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.548579017512823e-08,
+      "loss": 0.637,
+      "step": 6839
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.500479576736384e-08,
+      "loss": 0.6829,
+      "step": 6840
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.452588951443405e-08,
+      "loss": 0.541,
+      "step": 6841
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4049071516893984e-08,
+      "loss": 0.5643,
+      "step": 6842
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3574341874863545e-08,
+      "loss": 0.6515,
+      "step": 6843
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3101700688018564e-08,
+      "loss": 0.6633,
+      "step": 6844
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2631148055602986e-08,
+      "loss": 0.6848,
+      "step": 6845
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.216268407641778e-08,
+      "loss": 0.5537,
+      "step": 6846
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.169630884882648e-08,
+      "loss": 0.6694,
+      "step": 6847
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1232022470752985e-08,
+      "loss": 0.648,
+      "step": 6848
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.076982503968597e-08,
+      "loss": 0.6432,
+      "step": 6849
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0309716652671147e-08,
+      "loss": 0.6404,
+      "step": 6850
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.98516974063179e-08,
+      "loss": 0.6392,
+      "step": 6851
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.93957673967993e-08,
+      "loss": 0.634,
+      "step": 6852
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.894192671984321e-08,
+      "loss": 0.6894,
+      "step": 6853
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.849017547074564e-08,
+      "loss": 0.6877,
+      "step": 6854
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8040513744359584e-08,
+      "loss": 0.6545,
+      "step": 6855
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.759294163510175e-08,
+      "loss": 0.6191,
+      "step": 6856
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.714745923694808e-08,
+      "loss": 0.6701,
+      "step": 6857
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.670406664343707e-08,
+      "loss": 0.6449,
+      "step": 6858
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.626276394766871e-08,
+      "loss": 0.6631,
+      "step": 6859
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.58235512423022e-08,
+      "loss": 0.6691,
+      "step": 6860
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.538642861956044e-08,
+      "loss": 0.6715,
+      "step": 6861
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.495139617122446e-08,
+      "loss": 0.6677,
+      "step": 6862
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4518453988641183e-08,
+      "loss": 0.6349,
+      "step": 6863
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.408760216271235e-08,
+      "loss": 0.6519,
+      "step": 6864
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3658840783906695e-08,
+      "loss": 0.671,
+      "step": 6865
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.323216994224999e-08,
+      "loss": 0.6853,
+      "step": 6866
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.280758972733168e-08,
+      "loss": 0.6515,
+      "step": 6867
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.238510022829934e-08,
+      "loss": 0.6615,
+      "step": 6868
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1964701533864226e-08,
+      "loss": 0.6819,
+      "step": 6869
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1546393732297964e-08,
+      "loss": 0.6793,
+      "step": 6870
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1130176911432506e-08,
+      "loss": 0.5931,
+      "step": 6871
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0716051158660175e-08,
+      "loss": 0.656,
+      "step": 6872
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.030401656093697e-08,
+      "loss": 0.7101,
+      "step": 6873
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9894073204777004e-08,
+      "loss": 0.6914,
+      "step": 6874
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.948622117625589e-08,
+      "loss": 0.6682,
+      "step": 6875
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.908046056101067e-08,
+      "loss": 0.661,
+      "step": 6876
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.867679144423986e-08,
+      "loss": 0.5562,
+      "step": 6877
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.827521391070121e-08,
+      "loss": 0.6584,
+      "step": 6878
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7875728044713957e-08,
+      "loss": 0.6655,
+      "step": 6879
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.747833393015987e-08,
+      "loss": 0.6424,
+      "step": 6880
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7083031650477776e-08,
+      "loss": 0.6627,
+      "step": 6881
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.668982128867016e-08,
+      "loss": 0.6626,
+      "step": 6882
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6298702927300976e-08,
+      "loss": 0.6747,
+      "step": 6883
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.590967664849121e-08,
+      "loss": 0.6178,
+      "step": 6884
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5522742533926626e-08,
+      "loss": 0.664,
+      "step": 6885
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5137900664851124e-08,
+      "loss": 0.5436,
+      "step": 6886
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.475515112207117e-08,
+      "loss": 0.6747,
+      "step": 6887
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.437449398595138e-08,
+      "loss": 0.6705,
+      "step": 6888
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.399592933641893e-08,
+      "loss": 0.6675,
+      "step": 6889
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.361945725296134e-08,
+      "loss": 0.7133,
+      "step": 6890
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.32450778146276e-08,
+      "loss": 0.6312,
+      "step": 6891
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.287279110002484e-08,
+      "loss": 0.6694,
+      "step": 6892
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2502597187322736e-08,
+      "loss": 0.6486,
+      "step": 6893
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.213449615425135e-08,
+      "loss": 0.6324,
+      "step": 6894
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.176848807810107e-08,
+      "loss": 0.6452,
+      "step": 6895
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1404573035722644e-08,
+      "loss": 0.6497,
+      "step": 6896
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.104275110352717e-08,
+      "loss": 0.6885,
+      "step": 6897
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0683022357486104e-08,
+      "loss": 0.6873,
+      "step": 6898
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0325386873133466e-08,
+      "loss": 0.6255,
+      "step": 6899
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.99698447255603e-08,
+      "loss": 0.6255,
+      "step": 6900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9616395989421342e-08,
+      "loss": 0.6573,
+      "step": 6901
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.926504073893055e-08,
+      "loss": 0.5715,
+      "step": 6902
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8915779047860026e-08,
+      "loss": 0.6572,
+      "step": 6903
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8568610989545553e-08,
+      "loss": 0.5567,
+      "step": 6904
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8223536636883265e-08,
+      "loss": 0.541,
+      "step": 6905
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7880556062327425e-08,
+      "loss": 0.7004,
+      "step": 6906
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.753966933789376e-08,
+      "loss": 0.6836,
+      "step": 6907
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.720087653515946e-08,
+      "loss": 0.5091,
+      "step": 6908
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6864177725259843e-08,
+      "loss": 0.6469,
+      "step": 6909
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6529572978891694e-08,
+      "loss": 0.6488,
+      "step": 6910
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6197062366313254e-08,
+      "loss": 0.6678,
+      "step": 6911
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5866645957339787e-08,
+      "loss": 0.6469,
+      "step": 6912
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5538323821351352e-08,
+      "loss": 0.6724,
+      "step": 6913
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5212096027285028e-08,
+      "loss": 0.7082,
+      "step": 6914
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4887962643638243e-08,
+      "loss": 0.5807,
+      "step": 6915
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4565923738471e-08,
+      "loss": 0.6216,
+      "step": 6916
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.424597937940032e-08,
+      "loss": 0.6498,
+      "step": 6917
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.39281296336058e-08,
+      "loss": 0.6582,
+      "step": 6918
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.361237456782628e-08,
+      "loss": 0.5569,
+      "step": 6919
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3298714248360943e-08,
+      "loss": 0.6341,
+      "step": 6920
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.298714874106933e-08,
+      "loss": 0.701,
+      "step": 6921
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2677678111370227e-08,
+      "loss": 0.661,
+      "step": 6922
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2370302424246095e-08,
+      "loss": 0.6949,
+      "step": 6923
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2065021744233084e-08,
+      "loss": 0.6546,
+      "step": 6924
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1761836135433257e-08,
+      "loss": 0.5639,
+      "step": 6925
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1460745661505687e-08,
+      "loss": 0.665,
+      "step": 6926
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.116175038567092e-08,
+      "loss": 0.6771,
+      "step": 6927
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0864850370709843e-08,
+      "loss": 0.657,
+      "step": 6928
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.057004567896148e-08,
+      "loss": 0.729,
+      "step": 6929
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0277336372326317e-08,
+      "loss": 0.675,
+      "step": 6930
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9986722512265187e-08,
+      "loss": 0.5637,
+      "step": 6931
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9698204159798173e-08,
+      "loss": 0.559,
+      "step": 6932
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.941178137550459e-08,
+      "loss": 0.6776,
+      "step": 6933
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9127454219526332e-08,
+      "loss": 0.6312,
+      "step": 6934
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.884522275156342e-08,
+      "loss": 0.6458,
+      "step": 6935
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8565087030875118e-08,
+      "loss": 0.6581,
+      "step": 6936
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8287047116282154e-08,
+      "loss": 0.6828,
+      "step": 6937
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8011103066165603e-08,
+      "loss": 0.659,
+      "step": 6938
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7737254938463566e-08,
+      "loss": 0.5643,
+      "step": 6939
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7465502790677823e-08,
+      "loss": 0.6859,
+      "step": 6940
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.719584667986829e-08,
+      "loss": 0.665,
+      "step": 6941
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6928286662653005e-08,
+      "loss": 0.6433,
+      "step": 6942
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6662822795213696e-08,
+      "loss": 0.6623,
+      "step": 6943
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6399455133288e-08,
+      "loss": 0.6428,
+      "step": 6944
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6138183732177236e-08,
+      "loss": 0.6565,
+      "step": 6945
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5879008646739747e-08,
+      "loss": 0.6311,
+      "step": 6946
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.562192993139422e-08,
+      "loss": 0.663,
+      "step": 6947
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5366947640120812e-08,
+      "loss": 0.6651,
+      "step": 6948
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5114061826456696e-08,
+      "loss": 0.663,
+      "step": 6949
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4863272543501617e-08,
+      "loss": 0.6681,
+      "step": 6950
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4614579843913458e-08,
+      "loss": 0.6462,
+      "step": 6951
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4367983779910445e-08,
+      "loss": 0.6722,
+      "step": 6952
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4123484403270049e-08,
+      "loss": 0.6722,
+      "step": 6953
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3881081765331206e-08,
+      "loss": 0.6818,
+      "step": 6954
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3640775916989867e-08,
+      "loss": 0.5337,
+      "step": 6955
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.340256690870234e-08,
+      "loss": 0.6599,
+      "step": 6956
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3166454790488614e-08,
+      "loss": 0.6194,
+      "step": 6957
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2932439611922365e-08,
+      "loss": 0.575,
+      "step": 6958
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2700521422140955e-08,
+      "loss": 0.6415,
+      "step": 6959
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2470700269840985e-08,
+      "loss": 0.6525,
+      "step": 6960
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.224297620327608e-08,
+      "loss": 0.6501,
+      "step": 6961
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2017349270263545e-08,
+      "loss": 0.6635,
+      "step": 6962
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1793819518178818e-08,
+      "loss": 0.6392,
+      "step": 6963
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1572386993954354e-08,
+      "loss": 0.7053,
+      "step": 6964
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1353051744085185e-08,
+      "loss": 0.6509,
+      "step": 6965
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1135813814625585e-08,
+      "loss": 0.5775,
+      "step": 6966
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0920673251190172e-08,
+      "loss": 0.6106,
+      "step": 6967
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0707630098949484e-08,
+      "loss": 0.6745,
+      "step": 6968
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0496684402638847e-08,
+      "loss": 0.5323,
+      "step": 6969
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0287836206548385e-08,
+      "loss": 0.6286,
+      "step": 6970
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.008108555453191e-08,
+      "loss": 0.6591,
+      "step": 6971
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.876432490000254e-09,
+      "loss": 0.6826,
+      "step": 6972
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.673877055924596e-09,
+      "loss": 0.6295,
+      "step": 6973
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.473419294835362e-09,
+      "loss": 0.6668,
+      "step": 6974
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.275059248823326e-09,
+      "loss": 0.6505,
+      "step": 6975
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.078796959536284e-09,
+      "loss": 0.6426,
+      "step": 6976
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.884632468186827e-09,
+      "loss": 0.6222,
+      "step": 6977
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.69256581554012e-09,
+      "loss": 0.5443,
+      "step": 6978
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.502597041927242e-09,
+      "loss": 0.5698,
+      "step": 6979
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.314726187235166e-09,
+      "loss": 0.6848,
+      "step": 6980
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.128953290911235e-09,
+      "loss": 0.6746,
+      "step": 6981
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.945278391962018e-09,
+      "loss": 0.6399,
+      "step": 6982
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.763701528953338e-09,
+      "loss": 0.6797,
+      "step": 6983
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.58422274001136e-09,
+      "loss": 0.6239,
+      "step": 6984
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.4068420628226055e-09,
+      "loss": 0.6461,
+      "step": 6985
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.231559534629506e-09,
+      "loss": 0.6621,
+      "step": 6986
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.058375192238176e-09,
+      "loss": 0.6831,
+      "step": 6987
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.8872890720117495e-09,
+      "loss": 0.6624,
+      "step": 6988
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.718301209872602e-09,
+      "loss": 0.6139,
+      "step": 6989
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.551411641303462e-09,
+      "loss": 0.634,
+      "step": 6990
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.386620401345189e-09,
+      "loss": 0.6746,
+      "step": 6991
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.223927524601214e-09,
+      "loss": 0.6477,
+      "step": 6992
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.063333045230879e-09,
+      "loss": 0.7105,
+      "step": 6993
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.9048369969538775e-09,
+      "loss": 0.638,
+      "step": 6994
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.7484394130513656e-09,
+      "loss": 0.6644,
+      "step": 6995
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.5941403263615194e-09,
+      "loss": 0.6455,
+      "step": 6996
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.441939769280646e-09,
+      "loss": 0.6625,
+      "step": 6997
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.291837773768738e-09,
+      "loss": 0.6327,
+      "step": 6998
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.143834371342804e-09,
+      "loss": 0.6937,
+      "step": 6999
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.997929593076878e-09,
+      "loss": 0.6927,
+      "step": 7000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.854123469608673e-09,
+      "loss": 0.6872,
+      "step": 7001
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.712416031132927e-09,
+      "loss": 0.6732,
+      "step": 7002
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.572807307404725e-09,
+      "loss": 0.6203,
+      "step": 7003
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.435297327735067e-09,
+      "loss": 0.6498,
+      "step": 7004
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.2998861209997415e-09,
+      "loss": 0.6443,
+      "step": 7005
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.166573715629341e-09,
+      "loss": 0.5736,
+      "step": 7006
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.035360139617028e-09,
+      "loss": 0.668,
+      "step": 7007
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.906245420512988e-09,
+      "loss": 0.5444,
+      "step": 7008
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7792295854277575e-09,
+      "loss": 0.6682,
+      "step": 7009
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.654312661030002e-09,
+      "loss": 0.6777,
+      "step": 7010
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.531494673548741e-09,
+      "loss": 0.5609,
+      "step": 7011
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4107756487733457e-09,
+      "loss": 0.5879,
+      "step": 7012
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2921556120513178e-09,
+      "loss": 0.6545,
+      "step": 7013
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1756345882871795e-09,
+      "loss": 0.6322,
+      "step": 7014
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0612126019491374e-09,
+      "loss": 0.6615,
+      "step": 7015
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9488896770613064e-09,
+      "loss": 0.6424,
+      "step": 7016
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8386658372081543e-09,
+      "loss": 0.6622,
+      "step": 7017
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7305411055333907e-09,
+      "loss": 0.6681,
+      "step": 7018
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6245155047410764e-09,
+      "loss": 0.6681,
+      "step": 7019
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.520589057092293e-09,
+      "loss": 0.6502,
+      "step": 7020
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4187617844095846e-09,
+      "loss": 0.6665,
+      "step": 7021
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3190337080725155e-09,
+      "loss": 0.6033,
+      "step": 7022
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.221404849022113e-09,
+      "loss": 0.6451,
+      "step": 7023
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.125875227756424e-09,
+      "loss": 0.6546,
+      "step": 7024
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.032444864333849e-09,
+      "loss": 0.6536,
+      "step": 7025
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.941113778373138e-09,
+      "loss": 0.6883,
+      "step": 7026
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8518819890511742e-09,
+      "loss": 0.6768,
+      "step": 7027
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7647495151018602e-09,
+      "loss": 0.6692,
+      "step": 7028
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6797163748227819e-09,
+      "loss": 0.6277,
+      "step": 7029
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5967825860674358e-09,
+      "loss": 0.6406,
+      "step": 7030
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.51594816624967e-09,
+      "loss": 0.6928,
+      "step": 7031
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4372131323425742e-09,
+      "loss": 0.6478,
+      "step": 7032
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.360577500877369e-09,
+      "loss": 0.6579,
+      "step": 7033
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2860412879456275e-09,
+      "loss": 0.6633,
+      "step": 7034
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2136045091970527e-09,
+      "loss": 0.6794,
+      "step": 7035
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1432671798428108e-09,
+      "loss": 0.6406,
+      "step": 7036
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.075029314651088e-09,
+      "loss": 0.6724,
+      "step": 7037
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.008890927949313e-09,
+      "loss": 0.6531,
+      "step": 7038
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.448520336252653e-10,
+      "loss": 0.6768,
+      "step": 7039
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.829126451237457e-10,
+      "loss": 0.6185,
+      "step": 7040
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.230727754521273e-10,
+      "loss": 0.6708,
+      "step": 7041
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.653324371736936e-10,
+      "loss": 0.6577,
+      "step": 7042
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.096916424131906e-10,
+      "loss": 0.6367,
+      "step": 7043
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.561504028523847e-10,
+      "loss": 0.643,
+      "step": 7044
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.047087297345045e-10,
+      "loss": 0.6373,
+      "step": 7045
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.553666338609099e-10,
+      "loss": 0.6014,
+      "step": 7046
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.081241255899816e-10,
+      "loss": 0.5489,
+      "step": 7047
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6298121484267265e-10,
+      "loss": 0.6902,
+      "step": 7048
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1993791109806727e-10,
+      "loss": 0.5945,
+      "step": 7049
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7899422339338075e-10,
+      "loss": 0.633,
+      "step": 7050
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.401501603261803e-10,
+      "loss": 0.5827,
+      "step": 7051
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.03405730051054e-10,
+      "loss": 0.6593,
+      "step": 7052
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.687609402851621e-10,
+      "loss": 0.6924,
+      "step": 7053
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.362157983015756e-10,
+      "loss": 0.6722,
+      "step": 7054
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0577031093371725e-10,
+      "loss": 0.6379,
+      "step": 7055
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7742448457536142e-10,
+      "loss": 0.6551,
+      "step": 7056
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.511783251784138e-10,
+      "loss": 0.6304,
+      "step": 7057
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2703183825180098e-10,
+      "loss": 0.659,
+      "step": 7058
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0498502886702178e-10,
+      "loss": 0.5354,
+      "step": 7059
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.503790165370618e-11,
+      "loss": 0.6526,
+      "step": 7060
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.719046079961545e-11,
+      "loss": 0.6345,
+      "step": 7061
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.144271005064205e-11,
+      "loss": 0.5502,
+      "step": 7062
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7794652716360844e-11,
+      "loss": 0.6178,
+      "step": 7063
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6246291660037003e-11,
+      "loss": 0.7018,
+      "step": 7064
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6797629307507835e-11,
+      "loss": 0.6356,
+      "step": 7065
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.448667642741883e-12,
+      "loss": 0.6535,
+      "step": 7066
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1994082089491515e-12,
+      "loss": 0.6424,
+      "step": 7067
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0498521074708835e-12,
+      "loss": 0.6384,
+      "step": 7068
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.5967,
+      "step": 7069
+    },
+    {
+      "epoch": 1.0,
+      "step": 7069,
+      "total_flos": 9.040982324654218e+19,
+      "train_loss": 0.08859374577463862,
+      "train_runtime": 8443.7343,
+      "train_samples_per_second": 214.311,
+      "train_steps_per_second": 0.837
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 7069,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 9.040982324654218e+19,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}