diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,10405 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 10.0,
+  "global_step": 1730,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3094688221709008e-07,
+      "loss": 3.8008,
+      "step": 1
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6189376443418015e-07,
+      "loss": 3.7969,
+      "step": 2
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.928406466512703e-07,
+      "loss": 3.8027,
+      "step": 3
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.237875288683603e-07,
+      "loss": 3.8242,
+      "step": 4
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1547344110854505e-06,
+      "loss": 3.8086,
+      "step": 5
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3856812933025406e-06,
+      "loss": 3.8516,
+      "step": 6
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6166281755196307e-06,
+      "loss": 3.7969,
+      "step": 7
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8475750577367206e-06,
+      "loss": 3.7891,
+      "step": 8
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.0785219399538105e-06,
+      "loss": 3.8359,
+      "step": 9
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.309468822170901e-06,
+      "loss": 3.8262,
+      "step": 10
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.5404157043879907e-06,
+      "loss": 3.8164,
+      "step": 11
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7713625866050812e-06,
+      "loss": 3.793,
+      "step": 12
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.002309468822171e-06,
+      "loss": 3.8223,
+      "step": 13
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.2332563510392614e-06,
+      "loss": 3.8672,
+      "step": 14
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.464203233256351e-06,
+      "loss": 3.8574,
+      "step": 15
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.6951501154734412e-06,
+      "loss": 3.8164,
+      "step": 16
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.926096997690531e-06,
+      "loss": 3.7773,
+      "step": 17
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.157043879907621e-06,
+      "loss": 3.7617,
+      "step": 18
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.3879907621247115e-06,
+      "loss": 3.8438,
+      "step": 19
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.618937644341802e-06,
+      "loss": 3.8301,
+      "step": 20
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.849884526558892e-06,
+      "loss": 3.7949,
+      "step": 21
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.080831408775981e-06,
+      "loss": 3.8086,
+      "step": 22
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.311778290993072e-06,
+      "loss": 3.832,
+      "step": 23
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.5427251732101625e-06,
+      "loss": 3.8027,
+      "step": 24
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.773672055427252e-06,
+      "loss": 3.8105,
+      "step": 25
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.004618937644342e-06,
+      "loss": 3.8242,
+      "step": 26
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.235565819861432e-06,
+      "loss": 3.7793,
+      "step": 27
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.466512702078523e-06,
+      "loss": 3.8281,
+      "step": 28
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.6974595842956126e-06,
+      "loss": 3.8145,
+      "step": 29
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.928406466512702e-06,
+      "loss": 3.7852,
+      "step": 30
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.159353348729793e-06,
+      "loss": 3.8086,
+      "step": 31
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.3903002309468824e-06,
+      "loss": 3.8203,
+      "step": 32
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.621247113163972e-06,
+      "loss": 3.8242,
+      "step": 33
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.852193995381063e-06,
+      "loss": 3.7637,
+      "step": 34
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.083140877598153e-06,
+      "loss": 3.8281,
+      "step": 35
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.314087759815242e-06,
+      "loss": 3.7949,
+      "step": 36
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.545034642032333e-06,
+      "loss": 3.7734,
+      "step": 37
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.775981524249423e-06,
+      "loss": 3.7812,
+      "step": 38
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.006928406466514e-06,
+      "loss": 3.8105,
+      "step": 39
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.237875288683604e-06,
+      "loss": 3.7695,
+      "step": 40
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.468822170900693e-06,
+      "loss": 3.8242,
+      "step": 41
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.699769053117783e-06,
+      "loss": 3.7773,
+      "step": 42
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.930715935334874e-06,
+      "loss": 3.8379,
+      "step": 43
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.0161662817551963e-05,
+      "loss": 3.7148,
+      "step": 44
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.0392609699769053e-05,
+      "loss": 3.8242,
+      "step": 45
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0623556581986144e-05,
+      "loss": 3.7852,
+      "step": 46
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0854503464203233e-05,
+      "loss": 3.7461,
+      "step": 47
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1085450346420325e-05,
+      "loss": 3.7734,
+      "step": 48
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.1316397228637414e-05,
+      "loss": 3.6523,
+      "step": 49
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1547344110854504e-05,
+      "loss": 3.7852,
+      "step": 50
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1778290993071595e-05,
+      "loss": 3.7637,
+      "step": 51
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.2009237875288684e-05,
+      "loss": 3.6641,
+      "step": 52
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2240184757505774e-05,
+      "loss": 3.873,
+      "step": 53
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2471131639722865e-05,
+      "loss": 3.8047,
+      "step": 54
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2702078521939955e-05,
+      "loss": 3.7246,
+      "step": 55
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.2933025404157046e-05,
+      "loss": 3.7305,
+      "step": 56
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3163972286374135e-05,
+      "loss": 3.8516,
+      "step": 57
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3394919168591225e-05,
+      "loss": 3.8809,
+      "step": 58
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3625866050808314e-05,
+      "loss": 3.7988,
+      "step": 59
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3856812933025404e-05,
+      "loss": 3.7871,
+      "step": 60
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4087759815242497e-05,
+      "loss": 3.6758,
+      "step": 61
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4318706697459586e-05,
+      "loss": 3.8594,
+      "step": 62
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4549653579676676e-05,
+      "loss": 3.6445,
+      "step": 63
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4780600461893765e-05,
+      "loss": 3.6523,
+      "step": 64
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5011547344110854e-05,
+      "loss": 3.6465,
+      "step": 65
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.5242494226327944e-05,
+      "loss": 3.7402,
+      "step": 66
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5473441108545038e-05,
+      "loss": 3.8711,
+      "step": 67
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5704387990762125e-05,
+      "loss": 3.7754,
+      "step": 68
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5935334872979216e-05,
+      "loss": 3.8516,
+      "step": 69
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6166281755196306e-05,
+      "loss": 3.7109,
+      "step": 70
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6397228637413393e-05,
+      "loss": 3.7285,
+      "step": 71
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6628175519630484e-05,
+      "loss": 3.6641,
+      "step": 72
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6859122401847578e-05,
+      "loss": 3.6992,
+      "step": 73
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7090069284064665e-05,
+      "loss": 3.5723,
+      "step": 74
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7321016166281756e-05,
+      "loss": 3.7676,
+      "step": 75
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7551963048498846e-05,
+      "loss": 3.8027,
+      "step": 76
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7782909930715937e-05,
+      "loss": 3.8203,
+      "step": 77
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8013856812933027e-05,
+      "loss": 3.752,
+      "step": 78
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8244803695150118e-05,
+      "loss": 3.793,
+      "step": 79
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8475750577367208e-05,
+      "loss": 3.5781,
+      "step": 80
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8706697459584295e-05,
+      "loss": 3.5977,
+      "step": 81
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8937644341801386e-05,
+      "loss": 3.8633,
+      "step": 82
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9168591224018476e-05,
+      "loss": 3.8477,
+      "step": 83
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9399538106235567e-05,
+      "loss": 3.7578,
+      "step": 84
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9630484988452657e-05,
+      "loss": 3.7715,
+      "step": 85
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9861431870669748e-05,
+      "loss": 3.8262,
+      "step": 86
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0092378752886835e-05,
+      "loss": 3.7246,
+      "step": 87
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0323325635103926e-05,
+      "loss": 3.8477,
+      "step": 88
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0554272517321016e-05,
+      "loss": 3.6406,
+      "step": 89
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.0785219399538107e-05,
+      "loss": 3.7617,
+      "step": 90
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.1016166281755197e-05,
+      "loss": 3.7969,
+      "step": 91
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.1247113163972288e-05,
+      "loss": 3.6699,
+      "step": 92
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1478060046189378e-05,
+      "loss": 3.5469,
+      "step": 93
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1709006928406465e-05,
+      "loss": 3.5859,
+      "step": 94
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1939953810623556e-05,
+      "loss": 3.7695,
+      "step": 95
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.217090069284065e-05,
+      "loss": 3.8125,
+      "step": 96
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2401847575057737e-05,
+      "loss": 3.7012,
+      "step": 97
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.2632794457274828e-05,
+      "loss": 3.7285,
+      "step": 98
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.2863741339491918e-05,
+      "loss": 3.8184,
+      "step": 99
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.309468822170901e-05,
+      "loss": 3.6641,
+      "step": 100
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.3325635103926096e-05,
+      "loss": 3.7344,
+      "step": 101
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.355658198614319e-05,
+      "loss": 3.6973,
+      "step": 102
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.378752886836028e-05,
+      "loss": 3.791,
+      "step": 103
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4018475750577367e-05,
+      "loss": 3.5508,
+      "step": 104
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4249422632794458e-05,
+      "loss": 3.7598,
+      "step": 105
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.448036951501155e-05,
+      "loss": 3.6504,
+      "step": 106
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.471131639722864e-05,
+      "loss": 3.6836,
+      "step": 107
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.494226327944573e-05,
+      "loss": 3.6875,
+      "step": 108
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.517321016166282e-05,
+      "loss": 3.457,
+      "step": 109
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.540415704387991e-05,
+      "loss": 3.6855,
+      "step": 110
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.5635103926096998e-05,
+      "loss": 3.7305,
+      "step": 111
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.586605080831409e-05,
+      "loss": 3.9258,
+      "step": 112
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.609699769053118e-05,
+      "loss": 3.6875,
+      "step": 113
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.632794457274827e-05,
+      "loss": 3.6211,
+      "step": 114
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6558891454965356e-05,
+      "loss": 3.7227,
+      "step": 115
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.678983833718245e-05,
+      "loss": 3.6934,
+      "step": 116
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.702078521939954e-05,
+      "loss": 3.6309,
+      "step": 117
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.7251732101616628e-05,
+      "loss": 3.8477,
+      "step": 118
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.7482678983833722e-05,
+      "loss": 3.7109,
+      "step": 119
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.771362586605081e-05,
+      "loss": 3.6484,
+      "step": 120
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.79445727482679e-05,
+      "loss": 3.8125,
+      "step": 121
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.8175519630484993e-05,
+      "loss": 3.6719,
+      "step": 122
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.840646651270208e-05,
+      "loss": 3.5547,
+      "step": 123
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.863741339491917e-05,
+      "loss": 3.3711,
+      "step": 124
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8868360277136258e-05,
+      "loss": 3.5195,
+      "step": 125
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9099307159353352e-05,
+      "loss": 3.4961,
+      "step": 126
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9330254041570436e-05,
+      "loss": 3.5449,
+      "step": 127
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.956120092378753e-05,
+      "loss": 3.4102,
+      "step": 128
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9792147806004624e-05,
+      "loss": 3.7871,
+      "step": 129
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0023094688221707e-05,
+      "loss": 3.4609,
+      "step": 130
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.02540415704388e-05,
+      "loss": 3.7773,
+      "step": 131
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.048498845265589e-05,
+      "loss": 3.7578,
+      "step": 132
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.071593533487298e-05,
+      "loss": 3.8164,
+      "step": 133
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0946882217090076e-05,
+      "loss": 3.6836,
+      "step": 134
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.117782909930716e-05,
+      "loss": 3.5625,
+      "step": 135
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.140877598152425e-05,
+      "loss": 3.5312,
+      "step": 136
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.163972286374134e-05,
+      "loss": 3.6367,
+      "step": 137
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.187066974595843e-05,
+      "loss": 3.5488,
+      "step": 138
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.2101616628175526e-05,
+      "loss": 3.5352,
+      "step": 139
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.233256351039261e-05,
+      "loss": 3.5371,
+      "step": 140
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.25635103926097e-05,
+      "loss": 3.334,
+      "step": 141
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.279445727482679e-05,
+      "loss": 3.7559,
+      "step": 142
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.302540415704388e-05,
+      "loss": 3.709,
+      "step": 143
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.325635103926097e-05,
+      "loss": 3.5684,
+      "step": 144
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.348729792147806e-05,
+      "loss": 3.9023,
+      "step": 145
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3718244803695156e-05,
+      "loss": 3.7285,
+      "step": 146
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.394919168591224e-05,
+      "loss": 3.5957,
+      "step": 147
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.418013856812933e-05,
+      "loss": 3.75,
+      "step": 148
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.441108545034642e-05,
+      "loss": 3.4902,
+      "step": 149
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.464203233256351e-05,
+      "loss": 3.457,
+      "step": 150
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.4872979214780605e-05,
+      "loss": 3.8281,
+      "step": 151
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.510392609699769e-05,
+      "loss": 3.4297,
+      "step": 152
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5334872979214786e-05,
+      "loss": 3.8262,
+      "step": 153
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.556581986143187e-05,
+      "loss": 3.2344,
+      "step": 154
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.579676674364896e-05,
+      "loss": 3.7422,
+      "step": 155
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.6027713625866054e-05,
+      "loss": 3.5215,
+      "step": 156
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.625866050808314e-05,
+      "loss": 3.4316,
+      "step": 157
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6489607390300235e-05,
+      "loss": 3.3672,
+      "step": 158
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.672055427251732e-05,
+      "loss": 3.7969,
+      "step": 159
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6951501154734416e-05,
+      "loss": 3.334,
+      "step": 160
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.7182448036951504e-05,
+      "loss": 3.4941,
+      "step": 161
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.741339491916859e-05,
+      "loss": 3.7852,
+      "step": 162
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.7644341801385685e-05,
+      "loss": 3.9492,
+      "step": 163
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.787528868360277e-05,
+      "loss": 3.9648,
+      "step": 164
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8106235565819866e-05,
+      "loss": 3.7441,
+      "step": 165
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.833718244803695e-05,
+      "loss": 3.707,
+      "step": 166
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.856812933025405e-05,
+      "loss": 3.2266,
+      "step": 167
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8799076212471134e-05,
+      "loss": 3.1875,
+      "step": 168
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.903002309468822e-05,
+      "loss": 3.5156,
+      "step": 169
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.9260969976905315e-05,
+      "loss": 3.8594,
+      "step": 170
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.94919168591224e-05,
+      "loss": 3.2793,
+      "step": 171
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.9722863741339496e-05,
+      "loss": 3.4844,
+      "step": 172
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.995381062355658e-05,
+      "loss": 3.5742,
+      "step": 173
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.018475750577367e-05,
+      "loss": 3.7383,
+      "step": 174
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0415704387990764e-05,
+      "loss": 3.3828,
+      "step": 175
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.064665127020785e-05,
+      "loss": 3.4316,
+      "step": 176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0877598152424945e-05,
+      "loss": 3.2461,
+      "step": 177
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.110854503464203e-05,
+      "loss": 3.1387,
+      "step": 178
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.1339491916859126e-05,
+      "loss": 3.5859,
+      "step": 179
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.1570438799076213e-05,
+      "loss": 3.3711,
+      "step": 180
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.18013856812933e-05,
+      "loss": 3.7637,
+      "step": 181
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.2032332563510394e-05,
+      "loss": 3.5469,
+      "step": 182
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.226327944572748e-05,
+      "loss": 3.1426,
+      "step": 183
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.2494226327944576e-05,
+      "loss": 3.4453,
+      "step": 184
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.272517321016167e-05,
+      "loss": 3.2539,
+      "step": 185
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.2956120092378757e-05,
+      "loss": 3.7383,
+      "step": 186
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.3187066974595844e-05,
+      "loss": 3.2812,
+      "step": 187
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.341801385681293e-05,
+      "loss": 3.2734,
+      "step": 188
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.3648960739030025e-05,
+      "loss": 3.3828,
+      "step": 189
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.387990762124711e-05,
+      "loss": 3.6465,
+      "step": 190
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.4110854503464206e-05,
+      "loss": 3.5195,
+      "step": 191
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.43418013856813e-05,
+      "loss": 3.2266,
+      "step": 192
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.457274826789839e-05,
+      "loss": 3.502,
+      "step": 193
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4803695150115474e-05,
+      "loss": 3.4277,
+      "step": 194
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.503464203233256e-05,
+      "loss": 3.7207,
+      "step": 195
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5265588914549655e-05,
+      "loss": 3.4785,
+      "step": 196
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.549653579676675e-05,
+      "loss": 3.6777,
+      "step": 197
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.5727482678983836e-05,
+      "loss": 3.3223,
+      "step": 198
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.595842956120093e-05,
+      "loss": 3.7832,
+      "step": 199
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.618937644341802e-05,
+      "loss": 3.5566,
+      "step": 200
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.6420323325635104e-05,
+      "loss": 4.2383,
+      "step": 201
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.665127020785219e-05,
+      "loss": 3.4414,
+      "step": 202
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.6882217090069285e-05,
+      "loss": 3.5781,
+      "step": 203
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.711316397228638e-05,
+      "loss": 3.207,
+      "step": 204
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.7344110854503466e-05,
+      "loss": 2.7812,
+      "step": 205
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.757505773672056e-05,
+      "loss": 3.5117,
+      "step": 206
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.780600461893765e-05,
+      "loss": 2.4727,
+      "step": 207
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.8036951501154735e-05,
+      "loss": 2.6152,
+      "step": 208
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.826789838337183e-05,
+      "loss": 3.2148,
+      "step": 209
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.8498845265588916e-05,
+      "loss": 3.1699,
+      "step": 210
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.872979214780601e-05,
+      "loss": 3.4258,
+      "step": 211
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.89607390300231e-05,
+      "loss": 3.9863,
+      "step": 212
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.919168591224019e-05,
+      "loss": 2.9453,
+      "step": 213
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.942263279445728e-05,
+      "loss": 3.916,
+      "step": 214
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.9653579676674365e-05,
+      "loss": 3.9316,
+      "step": 215
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.988452655889146e-05,
+      "loss": 3.0117,
+      "step": 216
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.011547344110855e-05,
+      "loss": 3.125,
+      "step": 217
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.034642032332564e-05,
+      "loss": 3.1367,
+      "step": 218
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.057736720554273e-05,
+      "loss": 3.4453,
+      "step": 219
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.080831408775982e-05,
+      "loss": 3.4219,
+      "step": 220
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.103926096997691e-05,
+      "loss": 3.8633,
+      "step": 221
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.1270207852193995e-05,
+      "loss": 3.6387,
+      "step": 222
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.150115473441108e-05,
+      "loss": 3.8379,
+      "step": 223
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.173210161662818e-05,
+      "loss": 3.0547,
+      "step": 224
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.196304849884527e-05,
+      "loss": 3.3066,
+      "step": 225
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.219399538106236e-05,
+      "loss": 3.5723,
+      "step": 226
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.242494226327945e-05,
+      "loss": 2.8691,
+      "step": 227
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.265588914549654e-05,
+      "loss": 2.9277,
+      "step": 228
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.2886836027713626e-05,
+      "loss": 3.1719,
+      "step": 229
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.311778290993071e-05,
+      "loss": 2.6406,
+      "step": 230
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.334872979214781e-05,
+      "loss": 3.3691,
+      "step": 231
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.35796766743649e-05,
+      "loss": 3.2832,
+      "step": 232
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.381062355658199e-05,
+      "loss": 3.375,
+      "step": 233
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.404157043879908e-05,
+      "loss": 2.7109,
+      "step": 234
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.427251732101617e-05,
+      "loss": 3.1094,
+      "step": 235
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.4503464203233256e-05,
+      "loss": 3.3516,
+      "step": 236
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.4734411085450356e-05,
+      "loss": 3.9355,
+      "step": 237
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.4965357967667444e-05,
+      "loss": 3.5781,
+      "step": 238
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.519630484988453e-05,
+      "loss": 2.8398,
+      "step": 239
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.542725173210162e-05,
+      "loss": 3.2891,
+      "step": 240
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.565819861431871e-05,
+      "loss": 3.1465,
+      "step": 241
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.58891454965358e-05,
+      "loss": 3.7188,
+      "step": 242
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.6120092378752886e-05,
+      "loss": 3.375,
+      "step": 243
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.635103926096999e-05,
+      "loss": 3.1953,
+      "step": 244
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.6581986143187074e-05,
+      "loss": 3.0703,
+      "step": 245
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.681293302540416e-05,
+      "loss": 3.5586,
+      "step": 246
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.704387990762125e-05,
+      "loss": 3.1484,
+      "step": 247
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.727482678983834e-05,
+      "loss": 3.6094,
+      "step": 248
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.750577367205543e-05,
+      "loss": 3.3008,
+      "step": 249
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.7736720554272516e-05,
+      "loss": 3.4297,
+      "step": 250
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.796766743648962e-05,
+      "loss": 3.4297,
+      "step": 251
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.8198614318706704e-05,
+      "loss": 3.4805,
+      "step": 252
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.842956120092379e-05,
+      "loss": 2.5977,
+      "step": 253
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.866050808314087e-05,
+      "loss": 3.8203,
+      "step": 254
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.889145496535797e-05,
+      "loss": 2.3711,
+      "step": 255
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.912240184757506e-05,
+      "loss": 3.1172,
+      "step": 256
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.935334872979215e-05,
+      "loss": 3.502,
+      "step": 257
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.958429561200925e-05,
+      "loss": 3.5996,
+      "step": 258
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.9815242494226335e-05,
+      "loss": 3.3027,
+      "step": 259
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.0046189376443415e-05,
+      "loss": 3.5977,
+      "step": 260
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.0277136258660516e-05,
+      "loss": 3.4629,
+      "step": 261
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.05080831408776e-05,
+      "loss": 3.1367,
+      "step": 262
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.073903002309469e-05,
+      "loss": 3.332,
+      "step": 263
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.096997690531178e-05,
+      "loss": 3.1211,
+      "step": 264
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.120092378752888e-05,
+      "loss": 3.7441,
+      "step": 265
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.143187066974596e-05,
+      "loss": 3.0762,
+      "step": 266
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.166281755196305e-05,
+      "loss": 2.9766,
+      "step": 267
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.189376443418015e-05,
+      "loss": 3.0977,
+      "step": 268
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.212471131639723e-05,
+      "loss": 2.9492,
+      "step": 269
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.235565819861431e-05,
+      "loss": 2.4551,
+      "step": 270
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.25866050808314e-05,
+      "loss": 3.0449,
+      "step": 271
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.28175519630485e-05,
+      "loss": 4.1133,
+      "step": 272
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.304849884526559e-05,
+      "loss": 3.2207,
+      "step": 273
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.327944572748268e-05,
+      "loss": 3.0898,
+      "step": 274
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.351039260969978e-05,
+      "loss": 2.7402,
+      "step": 275
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.374133949191686e-05,
+      "loss": 2.7773,
+      "step": 276
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.397228637413395e-05,
+      "loss": 3.2539,
+      "step": 277
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.420323325635105e-05,
+      "loss": 3.1309,
+      "step": 278
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.443418013856814e-05,
+      "loss": 2.1621,
+      "step": 279
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 6.466512702078523e-05,
+      "loss": 2.7812,
+      "step": 280
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 6.489607390300231e-05,
+      "loss": 3.4258,
+      "step": 281
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 6.51270207852194e-05,
+      "loss": 1.3887,
+      "step": 282
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 6.535796766743649e-05,
+      "loss": 2.0488,
+      "step": 283
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 6.558891454965357e-05,
+      "loss": 3.1367,
+      "step": 284
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 6.581986143187067e-05,
+      "loss": 3.3926,
+      "step": 285
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 6.605080831408776e-05,
+      "loss": 2.5742,
+      "step": 286
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.628175519630485e-05,
+      "loss": 2.582,
+      "step": 287
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.651270207852194e-05,
+      "loss": 3.1777,
+      "step": 288
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.674364896073904e-05,
+      "loss": 2.8047,
+      "step": 289
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.697459584295612e-05,
+      "loss": 2.3086,
+      "step": 290
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.720554272517321e-05,
+      "loss": 3.1875,
+      "step": 291
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.743648960739031e-05,
+      "loss": 2.6152,
+      "step": 292
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.76674364896074e-05,
+      "loss": 2.7637,
+      "step": 293
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 6.789838337182449e-05,
+      "loss": 3.2246,
+      "step": 294
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 6.812933025404157e-05,
+      "loss": 3.3066,
+      "step": 295
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 6.836027713625866e-05,
+      "loss": 3.1035,
+      "step": 296
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 6.859122401847575e-05,
+      "loss": 3.3789,
+      "step": 297
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 6.882217090069283e-05,
+      "loss": 2.0879,
+      "step": 298
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 6.905311778290994e-05,
+      "loss": 3.9219,
+      "step": 299
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 6.928406466512702e-05,
+      "loss": 3.0605,
+      "step": 300
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.951501154734411e-05,
+      "loss": 2.3438,
+      "step": 301
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.974595842956121e-05,
+      "loss": 2.6484,
+      "step": 302
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.99769053117783e-05,
+      "loss": 3.1992,
+      "step": 303
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.020785219399538e-05,
+      "loss": 2.1816,
+      "step": 304
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.043879907621247e-05,
+      "loss": 2.4961,
+      "step": 305
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.066974595842957e-05,
+      "loss": 2.6426,
+      "step": 306
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.090069284064666e-05,
+      "loss": 1.793,
+      "step": 307
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.113163972286375e-05,
+      "loss": 3.0547,
+      "step": 308
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.136258660508083e-05,
+      "loss": 3.2754,
+      "step": 309
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.159353348729792e-05,
+      "loss": 2.834,
+      "step": 310
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.182448036951501e-05,
+      "loss": 3.2539,
+      "step": 311
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.205542725173211e-05,
+      "loss": 2.7656,
+      "step": 312
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.22863741339492e-05,
+      "loss": 2.3691,
+      "step": 313
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.251732101616628e-05,
+      "loss": 3.1152,
+      "step": 314
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.274826789838337e-05,
+      "loss": 3.0645,
+      "step": 315
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.297921478060047e-05,
+      "loss": 3.207,
+      "step": 316
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.321016166281756e-05,
+      "loss": 3.0527,
+      "step": 317
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.344110854503465e-05,
+      "loss": 2.6504,
+      "step": 318
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.367205542725175e-05,
+      "loss": 3.1523,
+      "step": 319
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.390300230946883e-05,
+      "loss": 1.8408,
+      "step": 320
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.413394919168592e-05,
+      "loss": 2.4883,
+      "step": 321
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.436489607390301e-05,
+      "loss": 2.416,
+      "step": 322
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.45958429561201e-05,
+      "loss": 2.5293,
+      "step": 323
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.482678983833718e-05,
+      "loss": 2.8281,
+      "step": 324
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.505773672055427e-05,
+      "loss": 2.4473,
+      "step": 325
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.528868360277137e-05,
+      "loss": 3.7188,
+      "step": 326
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.551963048498846e-05,
+      "loss": 2.6953,
+      "step": 327
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.575057736720554e-05,
+      "loss": 3.4844,
+      "step": 328
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.598152424942263e-05,
+      "loss": 2.5547,
+      "step": 329
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.621247113163973e-05,
+      "loss": 3.0996,
+      "step": 330
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.644341801385682e-05,
+      "loss": 2.9355,
+      "step": 331
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.66743648960739e-05,
+      "loss": 2.6309,
+      "step": 332
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.6905311778291e-05,
+      "loss": 2.9004,
+      "step": 333
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.71362586605081e-05,
+      "loss": 1.8682,
+      "step": 334
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.736720554272518e-05,
+      "loss": 3.6973,
+      "step": 335
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.759815242494227e-05,
+      "loss": 2.8965,
+      "step": 336
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.782909930715935e-05,
+      "loss": 2.6777,
+      "step": 337
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.806004618937644e-05,
+      "loss": 3.1895,
+      "step": 338
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.829099307159353e-05,
+      "loss": 2.7539,
+      "step": 339
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.852193995381063e-05,
+      "loss": 2.625,
+      "step": 340
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.875288683602772e-05,
+      "loss": 2.4941,
+      "step": 341
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.89838337182448e-05,
+      "loss": 3.1797,
+      "step": 342
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.92147806004619e-05,
+      "loss": 2.1387,
+      "step": 343
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.944572748267899e-05,
+      "loss": 2.209,
+      "step": 344
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.967667436489608e-05,
+      "loss": 2.625,
+      "step": 345
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.990762124711317e-05,
+      "loss": 2.5625,
+      "step": 346
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 8.013856812933027e-05,
+      "loss": 2.9004,
+      "step": 347
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 8.036951501154734e-05,
+      "loss": 1.9824,
+      "step": 348
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 8.060046189376443e-05,
+      "loss": 1.8652,
+      "step": 349
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 8.083140877598153e-05,
+      "loss": 2.5781,
+      "step": 350
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 8.106235565819862e-05,
+      "loss": 1.8047,
+      "step": 351
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 8.12933025404157e-05,
+      "loss": 2.3906,
+      "step": 352
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 8.15242494226328e-05,
+      "loss": 2.3848,
+      "step": 353
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 8.175519630484989e-05,
+      "loss": 2.7031,
+      "step": 354
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 8.198614318706698e-05,
+      "loss": 2.0957,
+      "step": 355
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 8.221709006928406e-05,
+      "loss": 2.748,
+      "step": 356
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 8.244803695150117e-05,
+      "loss": 1.8008,
+      "step": 357
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 8.267898383371825e-05,
+      "loss": 2.4648,
+      "step": 358
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 8.290993071593534e-05,
+      "loss": 2.4277,
+      "step": 359
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 8.314087759815243e-05,
+      "loss": 2.4395,
+      "step": 360
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.337182448036951e-05,
+      "loss": 1.8379,
+      "step": 361
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.36027713625866e-05,
+      "loss": 2.5293,
+      "step": 362
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.383371824480369e-05,
+      "loss": 2.3145,
+      "step": 363
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.406466512702079e-05,
+      "loss": 2.4453,
+      "step": 364
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.429561200923788e-05,
+      "loss": 1.624,
+      "step": 365
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.452655889145496e-05,
+      "loss": 3.0742,
+      "step": 366
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.475750577367206e-05,
+      "loss": 1.3809,
+      "step": 367
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.498845265588915e-05,
+      "loss": 1.375,
+      "step": 368
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.521939953810624e-05,
+      "loss": 2.2578,
+      "step": 369
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.545034642032334e-05,
+      "loss": 2.5898,
+      "step": 370
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.568129330254043e-05,
+      "loss": 4.1172,
+      "step": 371
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.591224018475751e-05,
+      "loss": 2.6992,
+      "step": 372
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.61431870669746e-05,
+      "loss": 2.0293,
+      "step": 373
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.637413394919169e-05,
+      "loss": 2.9648,
+      "step": 374
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.660508083140877e-05,
+      "loss": 2.0176,
+      "step": 375
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.683602771362586e-05,
+      "loss": 2.2832,
+      "step": 376
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.706697459584296e-05,
+      "loss": 2.2461,
+      "step": 377
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.729792147806005e-05,
+      "loss": 1.7695,
+      "step": 378
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.752886836027714e-05,
+      "loss": 2.6406,
+      "step": 379
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.775981524249422e-05,
+      "loss": 1.7969,
+      "step": 380
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.799076212471132e-05,
+      "loss": 3.2305,
+      "step": 381
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.822170900692841e-05,
+      "loss": 2.6758,
+      "step": 382
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.84526558891455e-05,
+      "loss": 2.9102,
+      "step": 383
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.86836027713626e-05,
+      "loss": 2.0742,
+      "step": 384
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.891454965357969e-05,
+      "loss": 2.2832,
+      "step": 385
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.914549653579677e-05,
+      "loss": 1.7305,
+      "step": 386
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.937644341801386e-05,
+      "loss": 3.7031,
+      "step": 387
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.960739030023095e-05,
+      "loss": 2.791,
+      "step": 388
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.983833718244804e-05,
+      "loss": 1.9805,
+      "step": 389
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.006928406466512e-05,
+      "loss": 2.2988,
+      "step": 390
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.030023094688222e-05,
+      "loss": 3.1289,
+      "step": 391
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.053117782909931e-05,
+      "loss": 2.0098,
+      "step": 392
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.07621247113164e-05,
+      "loss": 2.3164,
+      "step": 393
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.09930715935335e-05,
+      "loss": 2.3535,
+      "step": 394
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.122401847575059e-05,
+      "loss": 2.7168,
+      "step": 395
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.145496535796767e-05,
+      "loss": 2.0332,
+      "step": 396
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.168591224018476e-05,
+      "loss": 1.4395,
+      "step": 397
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.191685912240186e-05,
+      "loss": 2.0762,
+      "step": 398
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.214780600461895e-05,
+      "loss": 2.3984,
+      "step": 399
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.237875288683603e-05,
+      "loss": 1.4941,
+      "step": 400
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.260969976905312e-05,
+      "loss": 2.9414,
+      "step": 401
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.284064665127021e-05,
+      "loss": 2.2402,
+      "step": 402
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.30715935334873e-05,
+      "loss": 1.833,
+      "step": 403
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.330254041570438e-05,
+      "loss": 2.084,
+      "step": 404
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.353348729792148e-05,
+      "loss": 2.9375,
+      "step": 405
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.376443418013857e-05,
+      "loss": 1.5479,
+      "step": 406
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.399538106235566e-05,
+      "loss": 1.8301,
+      "step": 407
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.422632794457276e-05,
+      "loss": 1.9629,
+      "step": 408
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.445727482678985e-05,
+      "loss": 3.8145,
+      "step": 409
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.468822170900693e-05,
+      "loss": 1.4844,
+      "step": 410
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.491916859122403e-05,
+      "loss": 2.6211,
+      "step": 411
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.515011547344112e-05,
+      "loss": 2.1719,
+      "step": 412
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.538106235565821e-05,
+      "loss": 1.7764,
+      "step": 413
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.56120092378753e-05,
+      "loss": 2.2168,
+      "step": 414
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.584295612009238e-05,
+      "loss": 1.5762,
+      "step": 415
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.607390300230947e-05,
+      "loss": 1.8867,
+      "step": 416
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.630484988452656e-05,
+      "loss": 2.0352,
+      "step": 417
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.653579676674366e-05,
+      "loss": 1.96,
+      "step": 418
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.676674364896074e-05,
+      "loss": 1.9961,
+      "step": 419
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.699769053117783e-05,
+      "loss": 2.3223,
+      "step": 420
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.722863741339492e-05,
+      "loss": 2.1055,
+      "step": 421
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.745958429561202e-05,
+      "loss": 2.4883,
+      "step": 422
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.76905311778291e-05,
+      "loss": 1.8203,
+      "step": 423
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.79214780600462e-05,
+      "loss": 1.3896,
+      "step": 424
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.81524249422633e-05,
+      "loss": 1.7969,
+      "step": 425
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.838337182448038e-05,
+      "loss": 3.3906,
+      "step": 426
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.861431870669747e-05,
+      "loss": 2.2832,
+      "step": 427
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.884526558891456e-05,
+      "loss": 1.7637,
+      "step": 428
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.907621247113164e-05,
+      "loss": 2.5312,
+      "step": 429
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.930715935334873e-05,
+      "loss": 1.3271,
+      "step": 430
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.953810623556582e-05,
+      "loss": 2.0215,
+      "step": 431
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.976905311778292e-05,
+      "loss": 2.1445,
+      "step": 432
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001,
+      "loss": 1.3916,
+      "step": 433
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.992289899768697e-05,
+      "loss": 2.3984,
+      "step": 434
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.984579799537394e-05,
+      "loss": 2.0586,
+      "step": 435
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.976869699306092e-05,
+      "loss": 2.4766,
+      "step": 436
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.969159599074789e-05,
+      "loss": 1.7764,
+      "step": 437
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.961449498843486e-05,
+      "loss": 1.5039,
+      "step": 438
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.953739398612182e-05,
+      "loss": 1.5869,
+      "step": 439
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.946029298380879e-05,
+      "loss": 1.7188,
+      "step": 440
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.938319198149576e-05,
+      "loss": 1.9922,
+      "step": 441
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.930609097918273e-05,
+      "loss": 2.832,
+      "step": 442
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.92289899768697e-05,
+      "loss": 1.4023,
+      "step": 443
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.915188897455667e-05,
+      "loss": 2.0449,
+      "step": 444
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.907478797224364e-05,
+      "loss": 2.1758,
+      "step": 445
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.899768696993061e-05,
+      "loss": 1.8535,
+      "step": 446
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.892058596761759e-05,
+      "loss": 2.8789,
+      "step": 447
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.884348496530456e-05,
+      "loss": 3.6465,
+      "step": 448
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.876638396299152e-05,
+      "loss": 1.6875,
+      "step": 449
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.868928296067849e-05,
+      "loss": 2.9707,
+      "step": 450
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.861218195836547e-05,
+      "loss": 1.3398,
+      "step": 451
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.853508095605243e-05,
+      "loss": 2.1992,
+      "step": 452
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.84579799537394e-05,
+      "loss": 1.2314,
+      "step": 453
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.838087895142638e-05,
+      "loss": 0.6992,
+      "step": 454
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.830377794911334e-05,
+      "loss": 1.0684,
+      "step": 455
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.822667694680031e-05,
+      "loss": 1.668,
+      "step": 456
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.814957594448728e-05,
+      "loss": 2.5898,
+      "step": 457
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.807247494217426e-05,
+      "loss": 1.1094,
+      "step": 458
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.799537393986123e-05,
+      "loss": 1.6758,
+      "step": 459
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.791827293754819e-05,
+      "loss": 2.3438,
+      "step": 460
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.784117193523516e-05,
+      "loss": 3.0195,
+      "step": 461
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.776407093292214e-05,
+      "loss": 1.4111,
+      "step": 462
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.768696993060911e-05,
+      "loss": 2.3145,
+      "step": 463
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.760986892829608e-05,
+      "loss": 0.9761,
+      "step": 464
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.753276792598304e-05,
+      "loss": 1.623,
+      "step": 465
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.745566692367001e-05,
+      "loss": 2.123,
+      "step": 466
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.737856592135698e-05,
+      "loss": 1.8115,
+      "step": 467
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.730146491904395e-05,
+      "loss": 1.0479,
+      "step": 468
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.722436391673093e-05,
+      "loss": 1.2959,
+      "step": 469
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.71472629144179e-05,
+      "loss": 1.54,
+      "step": 470
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.707016191210486e-05,
+      "loss": 1.4951,
+      "step": 471
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.699306090979183e-05,
+      "loss": 1.6113,
+      "step": 472
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.691595990747881e-05,
+      "loss": 1.6719,
+      "step": 473
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.683885890516578e-05,
+      "loss": 1.5332,
+      "step": 474
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.676175790285275e-05,
+      "loss": 1.8408,
+      "step": 475
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.668465690053971e-05,
+      "loss": 1.4004,
+      "step": 476
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.66075558982267e-05,
+      "loss": 1.9053,
+      "step": 477
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.653045489591365e-05,
+      "loss": 1.542,
+      "step": 478
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.645335389360061e-05,
+      "loss": 2.2539,
+      "step": 479
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.637625289128758e-05,
+      "loss": 2.4375,
+      "step": 480
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.629915188897456e-05,
+      "loss": 2.8359,
+      "step": 481
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.622205088666153e-05,
+      "loss": 1.3203,
+      "step": 482
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.61449498843485e-05,
+      "loss": 2.4961,
+      "step": 483
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.606784888203547e-05,
+      "loss": 1.6943,
+      "step": 484
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.599074787972245e-05,
+      "loss": 0.5859,
+      "step": 485
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 9.591364687740941e-05,
+      "loss": 1.4434,
+      "step": 486
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.583654587509638e-05,
+      "loss": 1.7051,
+      "step": 487
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.575944487278335e-05,
+      "loss": 2.1992,
+      "step": 488
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.568234387047033e-05,
+      "loss": 1.4863,
+      "step": 489
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.560524286815728e-05,
+      "loss": 1.668,
+      "step": 490
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 9.552814186584425e-05,
+      "loss": 2.1152,
+      "step": 491
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 9.545104086353123e-05,
+      "loss": 2.5723,
+      "step": 492
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.53739398612182e-05,
+      "loss": 1.7539,
+      "step": 493
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.529683885890517e-05,
+      "loss": 2.6797,
+      "step": 494
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.521973785659213e-05,
+      "loss": 1.1611,
+      "step": 495
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.514263685427912e-05,
+      "loss": 0.9307,
+      "step": 496
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.506553585196608e-05,
+      "loss": 1.0537,
+      "step": 497
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.498843484965305e-05,
+      "loss": 1.4473,
+      "step": 498
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.491133384734002e-05,
+      "loss": 1.1006,
+      "step": 499
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 9.4834232845027e-05,
+      "loss": 1.6826,
+      "step": 500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.475713184271397e-05,
+      "loss": 2.0234,
+      "step": 501
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.468003084040093e-05,
+      "loss": 0.8071,
+      "step": 502
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.46029298380879e-05,
+      "loss": 2.9375,
+      "step": 503
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.452582883577487e-05,
+      "loss": 1.1768,
+      "step": 504
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.444872783346184e-05,
+      "loss": 1.4414,
+      "step": 505
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.43716268311488e-05,
+      "loss": 3.2715,
+      "step": 506
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 9.429452582883578e-05,
+      "loss": 1.127,
+      "step": 507
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.421742482652275e-05,
+      "loss": 2.7754,
+      "step": 508
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.414032382420972e-05,
+      "loss": 1.4629,
+      "step": 509
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.406322282189669e-05,
+      "loss": 1.2441,
+      "step": 510
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.398612181958367e-05,
+      "loss": 2.3047,
+      "step": 511
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.390902081727064e-05,
+      "loss": 2.0977,
+      "step": 512
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 9.38319198149576e-05,
+      "loss": 0.9102,
+      "step": 513
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 9.375481881264457e-05,
+      "loss": 2.1777,
+      "step": 514
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.367771781033154e-05,
+      "loss": 1.1982,
+      "step": 515
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.36006168080185e-05,
+      "loss": 1.4639,
+      "step": 516
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.352351580570547e-05,
+      "loss": 1.4434,
+      "step": 517
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.344641480339244e-05,
+      "loss": 1.3828,
+      "step": 518
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.336931380107942e-05,
+      "loss": 1.9004,
+      "step": 519
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 9.329221279876639e-05,
+      "loss": 1.7412,
+      "step": 520
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 9.321511179645336e-05,
+      "loss": 1.5088,
+      "step": 521
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 9.313801079414032e-05,
+      "loss": 1.5225,
+      "step": 522
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 9.30609097918273e-05,
+      "loss": 1.1455,
+      "step": 523
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 9.298380878951427e-05,
+      "loss": 1.6006,
+      "step": 524
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 9.290670778720124e-05,
+      "loss": 0.7188,
+      "step": 525
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 9.28296067848882e-05,
+      "loss": 1.9199,
+      "step": 526
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 9.275250578257519e-05,
+      "loss": 0.8188,
+      "step": 527
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 9.267540478026214e-05,
+      "loss": 1.2637,
+      "step": 528
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.259830377794911e-05,
+      "loss": 1.29,
+      "step": 529
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.252120277563609e-05,
+      "loss": 0.8755,
+      "step": 530
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.244410177332306e-05,
+      "loss": 1.1064,
+      "step": 531
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.236700077101002e-05,
+      "loss": 2.0508,
+      "step": 532
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.228989976869699e-05,
+      "loss": 1.2686,
+      "step": 533
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.221279876638397e-05,
+      "loss": 0.8242,
+      "step": 534
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.213569776407094e-05,
+      "loss": 1.1836,
+      "step": 535
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.205859676175791e-05,
+      "loss": 1.7852,
+      "step": 536
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.198149575944487e-05,
+      "loss": 2.123,
+      "step": 537
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.190439475713186e-05,
+      "loss": 1.1445,
+      "step": 538
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.182729375481882e-05,
+      "loss": 0.9185,
+      "step": 539
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.175019275250579e-05,
+      "loss": 1.9336,
+      "step": 540
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.167309175019276e-05,
+      "loss": 0.6821,
+      "step": 541
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.159599074787973e-05,
+      "loss": 2.0195,
+      "step": 542
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.151888974556669e-05,
+      "loss": 2.8398,
+      "step": 543
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.144178874325366e-05,
+      "loss": 1.293,
+      "step": 544
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.136468774094064e-05,
+      "loss": 0.9961,
+      "step": 545
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.128758673862761e-05,
+      "loss": 0.7358,
+      "step": 546
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.121048573631458e-05,
+      "loss": 1.4102,
+      "step": 547
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.113338473400154e-05,
+      "loss": 2.127,
+      "step": 548
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.105628373168852e-05,
+      "loss": 0.5854,
+      "step": 549
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 9.097918272937549e-05,
+      "loss": 1.1992,
+      "step": 550
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 9.090208172706246e-05,
+      "loss": 0.5967,
+      "step": 551
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 9.082498072474943e-05,
+      "loss": 0.8628,
+      "step": 552
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.07478797224364e-05,
+      "loss": 0.7686,
+      "step": 553
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.067077872012336e-05,
+      "loss": 1.2471,
+      "step": 554
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.059367771781033e-05,
+      "loss": 1.6406,
+      "step": 555
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.05165767154973e-05,
+      "loss": 1.3574,
+      "step": 556
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.043947571318428e-05,
+      "loss": 1.0684,
+      "step": 557
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 9.036237471087125e-05,
+      "loss": 3.0469,
+      "step": 558
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 9.028527370855821e-05,
+      "loss": 0.9067,
+      "step": 559
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.020817270624518e-05,
+      "loss": 1.8164,
+      "step": 560
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.013107170393216e-05,
+      "loss": 1.4961,
+      "step": 561
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 9.005397070161913e-05,
+      "loss": 0.8892,
+      "step": 562
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.99768696993061e-05,
+      "loss": 2.0098,
+      "step": 563
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.989976869699306e-05,
+      "loss": 0.9702,
+      "step": 564
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.982266769468004e-05,
+      "loss": 0.8608,
+      "step": 565
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.9745566692367e-05,
+      "loss": 0.7856,
+      "step": 566
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.966846569005397e-05,
+      "loss": 0.8389,
+      "step": 567
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.959136468774095e-05,
+      "loss": 0.8032,
+      "step": 568
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.951426368542791e-05,
+      "loss": 1.5244,
+      "step": 569
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.943716268311488e-05,
+      "loss": 0.8242,
+      "step": 570
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.936006168080185e-05,
+      "loss": 1.582,
+      "step": 571
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.928296067848883e-05,
+      "loss": 1.9609,
+      "step": 572
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.92058596761758e-05,
+      "loss": 1.917,
+      "step": 573
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.912875867386276e-05,
+      "loss": 0.8511,
+      "step": 574
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.905165767154973e-05,
+      "loss": 0.3486,
+      "step": 575
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.897455666923671e-05,
+      "loss": 1.373,
+      "step": 576
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.889745566692368e-05,
+      "loss": 0.6963,
+      "step": 577
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.882035466461063e-05,
+      "loss": 0.9614,
+      "step": 578
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.874325366229762e-05,
+      "loss": 0.5103,
+      "step": 579
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.866615265998458e-05,
+      "loss": 0.8765,
+      "step": 580
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.858905165767155e-05,
+      "loss": 1.583,
+      "step": 581
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.851195065535852e-05,
+      "loss": 0.8325,
+      "step": 582
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.84348496530455e-05,
+      "loss": 1.7217,
+      "step": 583
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.835774865073247e-05,
+      "loss": 0.6855,
+      "step": 584
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.828064764841943e-05,
+      "loss": 1.5576,
+      "step": 585
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.82035466461064e-05,
+      "loss": 0.8994,
+      "step": 586
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.812644564379338e-05,
+      "loss": 2.1074,
+      "step": 587
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.804934464148035e-05,
+      "loss": 1.1191,
+      "step": 588
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.797224363916732e-05,
+      "loss": 1.665,
+      "step": 589
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.789514263685428e-05,
+      "loss": 3.0938,
+      "step": 590
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.781804163454125e-05,
+      "loss": 0.8892,
+      "step": 591
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.774094063222822e-05,
+      "loss": 1.5146,
+      "step": 592
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.766383962991519e-05,
+      "loss": 1.1338,
+      "step": 593
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.758673862760215e-05,
+      "loss": 1.2861,
+      "step": 594
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.750963762528913e-05,
+      "loss": 1.9541,
+      "step": 595
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.74325366229761e-05,
+      "loss": 0.6392,
+      "step": 596
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.735543562066307e-05,
+      "loss": 0.6748,
+      "step": 597
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.727833461835004e-05,
+      "loss": 0.6836,
+      "step": 598
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.720123361603702e-05,
+      "loss": 1.8193,
+      "step": 599
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 8.712413261372399e-05,
+      "loss": 2.3066,
+      "step": 600
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 8.704703161141095e-05,
+      "loss": 1.4277,
+      "step": 601
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.696993060909792e-05,
+      "loss": 1.457,
+      "step": 602
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.689282960678489e-05,
+      "loss": 0.3665,
+      "step": 603
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.681572860447186e-05,
+      "loss": 0.5049,
+      "step": 604
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.673862760215882e-05,
+      "loss": 0.7598,
+      "step": 605
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.66615265998458e-05,
+      "loss": 0.9512,
+      "step": 606
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.658442559753277e-05,
+      "loss": 0.7305,
+      "step": 607
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.650732459521974e-05,
+      "loss": 0.9956,
+      "step": 608
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.64302235929067e-05,
+      "loss": 1.8623,
+      "step": 609
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 8.635312259059369e-05,
+      "loss": 0.7773,
+      "step": 610
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 8.627602158828065e-05,
+      "loss": 2.2266,
+      "step": 611
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 8.619892058596762e-05,
+      "loss": 1.3184,
+      "step": 612
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 8.612181958365459e-05,
+      "loss": 1.4219,
+      "step": 613
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 8.604471858134157e-05,
+      "loss": 2.2949,
+      "step": 614
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 8.596761757902854e-05,
+      "loss": 3.0977,
+      "step": 615
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 8.589051657671549e-05,
+      "loss": 0.9102,
+      "step": 616
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 8.581341557440247e-05,
+      "loss": 1.0938,
+      "step": 617
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 8.573631457208944e-05,
+      "loss": 2.2324,
+      "step": 618
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 8.565921356977641e-05,
+      "loss": 1.6631,
+      "step": 619
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 8.558211256746337e-05,
+      "loss": 1.6328,
+      "step": 620
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 8.550501156515036e-05,
+      "loss": 1.2207,
+      "step": 621
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 8.542791056283732e-05,
+      "loss": 0.4226,
+      "step": 622
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 8.535080956052429e-05,
+      "loss": 0.8247,
+      "step": 623
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 8.527370855821126e-05,
+      "loss": 2.9258,
+      "step": 624
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 8.519660755589824e-05,
+      "loss": 1.2832,
+      "step": 625
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 8.51195065535852e-05,
+      "loss": 1.3564,
+      "step": 626
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 8.504240555127217e-05,
+      "loss": 0.8281,
+      "step": 627
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 8.496530454895914e-05,
+      "loss": 0.7993,
+      "step": 628
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 8.488820354664611e-05,
+      "loss": 0.8208,
+      "step": 629
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 8.481110254433308e-05,
+      "loss": 0.9502,
+      "step": 630
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.473400154202004e-05,
+      "loss": 0.7656,
+      "step": 631
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.465690053970702e-05,
+      "loss": 0.2705,
+      "step": 632
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.457979953739399e-05,
+      "loss": 1.084,
+      "step": 633
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.450269853508096e-05,
+      "loss": 2.8242,
+      "step": 634
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.442559753276793e-05,
+      "loss": 0.688,
+      "step": 635
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.434849653045491e-05,
+      "loss": 0.873,
+      "step": 636
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.427139552814188e-05,
+      "loss": 1.4326,
+      "step": 637
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.419429452582884e-05,
+      "loss": 0.4094,
+      "step": 638
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.411719352351581e-05,
+      "loss": 2.0078,
+      "step": 639
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.404009252120279e-05,
+      "loss": 0.6719,
+      "step": 640
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.396299151888975e-05,
+      "loss": 1.6416,
+      "step": 641
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.388589051657671e-05,
+      "loss": 0.7173,
+      "step": 642
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.380878951426368e-05,
+      "loss": 1.1016,
+      "step": 643
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.373168851195066e-05,
+      "loss": 2.1602,
+      "step": 644
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.365458750963763e-05,
+      "loss": 0.1714,
+      "step": 645
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.35774865073246e-05,
+      "loss": 0.7021,
+      "step": 646
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.350038550501156e-05,
+      "loss": 0.6519,
+      "step": 647
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.342328450269854e-05,
+      "loss": 1.0566,
+      "step": 648
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.334618350038551e-05,
+      "loss": 0.4314,
+      "step": 649
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.326908249807248e-05,
+      "loss": 0.354,
+      "step": 650
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.319198149575945e-05,
+      "loss": 1.0508,
+      "step": 651
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.311488049344643e-05,
+      "loss": 1.3643,
+      "step": 652
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.30377794911334e-05,
+      "loss": 2.6953,
+      "step": 653
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.296067848882035e-05,
+      "loss": 0.7573,
+      "step": 654
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.288357748650733e-05,
+      "loss": 1.9834,
+      "step": 655
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.28064764841943e-05,
+      "loss": 1.7373,
+      "step": 656
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.272937548188126e-05,
+      "loss": 2.1973,
+      "step": 657
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.265227447956823e-05,
+      "loss": 0.8623,
+      "step": 658
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.257517347725521e-05,
+      "loss": 0.4915,
+      "step": 659
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.249807247494218e-05,
+      "loss": 2.1816,
+      "step": 660
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.242097147262915e-05,
+      "loss": 1.334,
+      "step": 661
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.234387047031612e-05,
+      "loss": 0.2893,
+      "step": 662
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.22667694680031e-05,
+      "loss": 0.6621,
+      "step": 663
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.218966846569006e-05,
+      "loss": 2.2773,
+      "step": 664
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.211256746337703e-05,
+      "loss": 1.1758,
+      "step": 665
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.2035466461064e-05,
+      "loss": 0.1781,
+      "step": 666
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.195836545875097e-05,
+      "loss": 1.2354,
+      "step": 667
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.188126445643793e-05,
+      "loss": 0.2585,
+      "step": 668
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.18041634541249e-05,
+      "loss": 0.7886,
+      "step": 669
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.172706245181188e-05,
+      "loss": 1.6709,
+      "step": 670
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.164996144949885e-05,
+      "loss": 1.1006,
+      "step": 671
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.157286044718582e-05,
+      "loss": 0.6152,
+      "step": 672
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 8.149575944487278e-05,
+      "loss": 1.0596,
+      "step": 673
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.141865844255977e-05,
+      "loss": 0.9277,
+      "step": 674
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.134155744024673e-05,
+      "loss": 1.417,
+      "step": 675
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 8.12644564379337e-05,
+      "loss": 1.4395,
+      "step": 676
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 8.118735543562067e-05,
+      "loss": 1.4736,
+      "step": 677
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 8.111025443330765e-05,
+      "loss": 0.9976,
+      "step": 678
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 8.10331534309946e-05,
+      "loss": 1.0713,
+      "step": 679
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 8.095605242868157e-05,
+      "loss": 3.9688,
+      "step": 680
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.087895142636854e-05,
+      "loss": 0.5947,
+      "step": 681
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.080185042405552e-05,
+      "loss": 1.3623,
+      "step": 682
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.072474942174249e-05,
+      "loss": 0.5635,
+      "step": 683
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.064764841942945e-05,
+      "loss": 1.2412,
+      "step": 684
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 8.057054741711642e-05,
+      "loss": 0.4321,
+      "step": 685
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 8.04934464148034e-05,
+      "loss": 0.4873,
+      "step": 686
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 8.041634541249037e-05,
+      "loss": 0.5288,
+      "step": 687
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 8.033924441017734e-05,
+      "loss": 0.9893,
+      "step": 688
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 8.02621434078643e-05,
+      "loss": 1.0264,
+      "step": 689
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 8.018504240555128e-05,
+      "loss": 1.2354,
+      "step": 690
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 8.010794140323825e-05,
+      "loss": 0.4053,
+      "step": 691
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 8.00308404009252e-05,
+      "loss": 0.9316,
+      "step": 692
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 7.995373939861219e-05,
+      "loss": 0.1957,
+      "step": 693
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 7.987663839629915e-05,
+      "loss": 1.5771,
+      "step": 694
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 7.979953739398612e-05,
+      "loss": 1.3877,
+      "step": 695
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 7.972243639167309e-05,
+      "loss": 0.9761,
+      "step": 696
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 7.964533538936007e-05,
+      "loss": 1.0869,
+      "step": 697
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 7.956823438704704e-05,
+      "loss": 0.957,
+      "step": 698
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.9491133384734e-05,
+      "loss": 0.3342,
+      "step": 699
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 7.941403238242097e-05,
+      "loss": 0.4351,
+      "step": 700
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 7.933693138010795e-05,
+      "loss": 0.7803,
+      "step": 701
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.925983037779492e-05,
+      "loss": 0.2742,
+      "step": 702
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.918272937548189e-05,
+      "loss": 0.5776,
+      "step": 703
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.910562837316886e-05,
+      "loss": 0.5078,
+      "step": 704
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 7.902852737085582e-05,
+      "loss": 1.124,
+      "step": 705
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 7.895142636854279e-05,
+      "loss": 1.2441,
+      "step": 706
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 7.887432536622976e-05,
+      "loss": 0.6992,
+      "step": 707
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 7.879722436391674e-05,
+      "loss": 0.3477,
+      "step": 708
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 7.87201233616037e-05,
+      "loss": 0.6162,
+      "step": 709
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 7.864302235929067e-05,
+      "loss": 0.5444,
+      "step": 710
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 7.856592135697764e-05,
+      "loss": 0.5259,
+      "step": 711
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 7.848882035466462e-05,
+      "loss": 1.1787,
+      "step": 712
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 7.841171935235159e-05,
+      "loss": 0.6543,
+      "step": 713
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 7.833461835003856e-05,
+      "loss": 0.2932,
+      "step": 714
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 7.825751734772552e-05,
+      "loss": 1.1064,
+      "step": 715
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 7.81804163454125e-05,
+      "loss": 0.6733,
+      "step": 716
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 7.810331534309946e-05,
+      "loss": 0.5444,
+      "step": 717
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 7.802621434078643e-05,
+      "loss": 1.4482,
+      "step": 718
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 7.79491133384734e-05,
+      "loss": 0.3245,
+      "step": 719
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 7.787201233616038e-05,
+      "loss": 0.9424,
+      "step": 720
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 7.779491133384734e-05,
+      "loss": 0.7539,
+      "step": 721
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 7.771781033153431e-05,
+      "loss": 0.6431,
+      "step": 722
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 7.764070932922128e-05,
+      "loss": 0.8262,
+      "step": 723
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 7.756360832690826e-05,
+      "loss": 1.4004,
+      "step": 724
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 7.748650732459523e-05,
+      "loss": 0.5713,
+      "step": 725
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 7.74094063222822e-05,
+      "loss": 0.5854,
+      "step": 726
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 7.733230531996916e-05,
+      "loss": 1.3105,
+      "step": 727
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 7.725520431765614e-05,
+      "loss": 0.7627,
+      "step": 728
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 7.71781033153431e-05,
+      "loss": 0.325,
+      "step": 729
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 7.710100231303006e-05,
+      "loss": 1.3301,
+      "step": 730
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 7.702390131071704e-05,
+      "loss": 0.5825,
+      "step": 731
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 7.694680030840401e-05,
+      "loss": 1.3789,
+      "step": 732
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 7.686969930609098e-05,
+      "loss": 0.3752,
+      "step": 733
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 7.679259830377795e-05,
+      "loss": 0.5303,
+      "step": 734
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 7.671549730146493e-05,
+      "loss": 0.855,
+      "step": 735
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 7.66383962991519e-05,
+      "loss": 0.6064,
+      "step": 736
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 7.656129529683886e-05,
+      "loss": 1.7305,
+      "step": 737
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 7.648419429452583e-05,
+      "loss": 1.0615,
+      "step": 738
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 7.640709329221281e-05,
+      "loss": 0.2074,
+      "step": 739
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 7.632999228989978e-05,
+      "loss": 0.7061,
+      "step": 740
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 7.625289128758675e-05,
+      "loss": 0.2883,
+      "step": 741
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 7.617579028527371e-05,
+      "loss": 2.0273,
+      "step": 742
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 7.609868928296068e-05,
+      "loss": 0.9233,
+      "step": 743
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 7.602158828064765e-05,
+      "loss": 0.1758,
+      "step": 744
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 7.594448727833462e-05,
+      "loss": 1.3965,
+      "step": 745
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 7.58673862760216e-05,
+      "loss": 0.1292,
+      "step": 746
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 7.579028527370856e-05,
+      "loss": 0.3279,
+      "step": 747
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 7.571318427139553e-05,
+      "loss": 0.3599,
+      "step": 748
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 7.56360832690825e-05,
+      "loss": 0.939,
+      "step": 749
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 7.555898226676948e-05,
+      "loss": 0.707,
+      "step": 750
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 7.548188126445645e-05,
+      "loss": 0.4722,
+      "step": 751
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 7.540478026214341e-05,
+      "loss": 0.5312,
+      "step": 752
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 7.532767925983038e-05,
+      "loss": 0.3713,
+      "step": 753
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 7.525057825751736e-05,
+      "loss": 0.3372,
+      "step": 754
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 7.517347725520432e-05,
+      "loss": 0.4668,
+      "step": 755
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 7.509637625289128e-05,
+      "loss": 0.1705,
+      "step": 756
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.501927525057825e-05,
+      "loss": 1.5625,
+      "step": 757
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.494217424826523e-05,
+      "loss": 0.416,
+      "step": 758
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.48650732459522e-05,
+      "loss": 1.7422,
+      "step": 759
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.478797224363917e-05,
+      "loss": 0.8501,
+      "step": 760
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.471087124132613e-05,
+      "loss": 1.5078,
+      "step": 761
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.463377023901312e-05,
+      "loss": 0.4238,
+      "step": 762
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.455666923670008e-05,
+      "loss": 0.8032,
+      "step": 763
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.447956823438705e-05,
+      "loss": 0.6787,
+      "step": 764
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.440246723207402e-05,
+      "loss": 1.0381,
+      "step": 765
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 7.4325366229761e-05,
+      "loss": 0.4087,
+      "step": 766
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 7.424826522744795e-05,
+      "loss": 0.8135,
+      "step": 767
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 7.417116422513492e-05,
+      "loss": 0.2104,
+      "step": 768
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 7.40940632228219e-05,
+      "loss": 0.354,
+      "step": 769
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 7.401696222050887e-05,
+      "loss": 1.0752,
+      "step": 770
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 7.393986121819584e-05,
+      "loss": 1.2139,
+      "step": 771
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 7.38627602158828e-05,
+      "loss": 0.3684,
+      "step": 772
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 7.378565921356978e-05,
+      "loss": 0.1077,
+      "step": 773
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 7.370855821125675e-05,
+      "loss": 1.1064,
+      "step": 774
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.363145720894372e-05,
+      "loss": 0.2053,
+      "step": 775
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.355435620663069e-05,
+      "loss": 0.6655,
+      "step": 776
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.347725520431767e-05,
+      "loss": 0.5415,
+      "step": 777
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.340015420200464e-05,
+      "loss": 0.564,
+      "step": 778
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.33230531996916e-05,
+      "loss": 0.6162,
+      "step": 779
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.324595219737857e-05,
+      "loss": 0.8589,
+      "step": 780
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.316885119506554e-05,
+      "loss": 0.5537,
+      "step": 781
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 7.30917501927525e-05,
+      "loss": 0.8804,
+      "step": 782
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 7.301464919043947e-05,
+      "loss": 0.7988,
+      "step": 783
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 7.293754818812645e-05,
+      "loss": 1.4355,
+      "step": 784
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 7.286044718581342e-05,
+      "loss": 0.584,
+      "step": 785
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 7.278334618350039e-05,
+      "loss": 0.2227,
+      "step": 786
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 7.270624518118736e-05,
+      "loss": 2.2656,
+      "step": 787
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 7.262914417887434e-05,
+      "loss": 0.2554,
+      "step": 788
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 7.25520431765613e-05,
+      "loss": 0.7319,
+      "step": 789
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 7.247494217424827e-05,
+      "loss": 0.1246,
+      "step": 790
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 7.239784117193524e-05,
+      "loss": 0.3296,
+      "step": 791
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 7.23207401696222e-05,
+      "loss": 0.2874,
+      "step": 792
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 7.224363916730917e-05,
+      "loss": 1.0684,
+      "step": 793
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 7.216653816499614e-05,
+      "loss": 0.0903,
+      "step": 794
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 7.208943716268311e-05,
+      "loss": 0.2715,
+      "step": 795
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 7.201233616037009e-05,
+      "loss": 0.394,
+      "step": 796
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.193523515805706e-05,
+      "loss": 0.2566,
+      "step": 797
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.185813415574402e-05,
+      "loss": 0.9497,
+      "step": 798
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.178103315343099e-05,
+      "loss": 1.3574,
+      "step": 799
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.170393215111797e-05,
+      "loss": 0.4441,
+      "step": 800
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.162683114880494e-05,
+      "loss": 0.5986,
+      "step": 801
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 7.154973014649191e-05,
+      "loss": 0.4897,
+      "step": 802
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 7.147262914417888e-05,
+      "loss": 1.3643,
+      "step": 803
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 7.139552814186586e-05,
+      "loss": 0.2257,
+      "step": 804
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 7.131842713955281e-05,
+      "loss": 0.079,
+      "step": 805
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 7.124132613723978e-05,
+      "loss": 0.0884,
+      "step": 806
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 7.116422513492676e-05,
+      "loss": 0.8276,
+      "step": 807
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 7.108712413261373e-05,
+      "loss": 0.3525,
+      "step": 808
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 7.10100231303007e-05,
+      "loss": 1.6689,
+      "step": 809
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 7.093292212798766e-05,
+      "loss": 0.3567,
+      "step": 810
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 7.085582112567464e-05,
+      "loss": 0.7935,
+      "step": 811
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 7.077872012336161e-05,
+      "loss": 0.8022,
+      "step": 812
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 7.070161912104858e-05,
+      "loss": 2.9707,
+      "step": 813
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 7.062451811873554e-05,
+      "loss": 0.3547,
+      "step": 814
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 7.054741711642253e-05,
+      "loss": 0.1075,
+      "step": 815
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 7.047031611410949e-05,
+      "loss": 0.0768,
+      "step": 816
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 7.039321511179646e-05,
+      "loss": 0.875,
+      "step": 817
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 7.031611410948343e-05,
+      "loss": 0.4331,
+      "step": 818
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 7.02390131071704e-05,
+      "loss": 0.1364,
+      "step": 819
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 7.016191210485736e-05,
+      "loss": 0.5728,
+      "step": 820
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 7.008481110254433e-05,
+      "loss": 0.5396,
+      "step": 821
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 7.000771010023131e-05,
+      "loss": 0.2108,
+      "step": 822
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 6.993060909791828e-05,
+      "loss": 0.1407,
+      "step": 823
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 6.985350809560525e-05,
+      "loss": 1.0039,
+      "step": 824
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 6.977640709329221e-05,
+      "loss": 0.5405,
+      "step": 825
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 6.96993060909792e-05,
+      "loss": 0.2605,
+      "step": 826
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 6.962220508866616e-05,
+      "loss": 0.2532,
+      "step": 827
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 6.954510408635313e-05,
+      "loss": 0.6343,
+      "step": 828
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 6.94680030840401e-05,
+      "loss": 0.1754,
+      "step": 829
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 6.939090208172706e-05,
+      "loss": 0.3975,
+      "step": 830
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 6.931380107941403e-05,
+      "loss": 0.6416,
+      "step": 831
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 6.9236700077101e-05,
+      "loss": 0.9487,
+      "step": 832
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.915959907478798e-05,
+      "loss": 1.2188,
+      "step": 833
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.908249807247495e-05,
+      "loss": 1.6152,
+      "step": 834
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.900539707016191e-05,
+      "loss": 1.2051,
+      "step": 835
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.892829606784888e-05,
+      "loss": 0.458,
+      "step": 836
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.885119506553586e-05,
+      "loss": 0.4106,
+      "step": 837
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.877409406322283e-05,
+      "loss": 0.5317,
+      "step": 838
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.86969930609098e-05,
+      "loss": 0.6328,
+      "step": 839
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.861989205859677e-05,
+      "loss": 0.1519,
+      "step": 840
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.854279105628375e-05,
+      "loss": 1.4766,
+      "step": 841
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 6.846569005397071e-05,
+      "loss": 0.3022,
+      "step": 842
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 6.838858905165767e-05,
+      "loss": 0.1764,
+      "step": 843
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 6.831148804934463e-05,
+      "loss": 0.5884,
+      "step": 844
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 6.823438704703162e-05,
+      "loss": 0.8398,
+      "step": 845
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.815728604471858e-05,
+      "loss": 0.3755,
+      "step": 846
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 6.808018504240555e-05,
+      "loss": 0.3857,
+      "step": 847
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 6.800308404009252e-05,
+      "loss": 0.304,
+      "step": 848
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 6.79259830377795e-05,
+      "loss": 1.4521,
+      "step": 849
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 6.784888203546647e-05,
+      "loss": 0.1824,
+      "step": 850
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.777178103315343e-05,
+      "loss": 0.0878,
+      "step": 851
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.76946800308404e-05,
+      "loss": 0.8818,
+      "step": 852
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 6.761757902852738e-05,
+      "loss": 0.9541,
+      "step": 853
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.754047802621435e-05,
+      "loss": 2.0996,
+      "step": 854
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.74633770239013e-05,
+      "loss": 0.3452,
+      "step": 855
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.738627602158828e-05,
+      "loss": 0.8452,
+      "step": 856
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.730917501927525e-05,
+      "loss": 0.0999,
+      "step": 857
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 6.723207401696222e-05,
+      "loss": 0.771,
+      "step": 858
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.715497301464919e-05,
+      "loss": 0.3694,
+      "step": 859
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.707787201233617e-05,
+      "loss": 0.3176,
+      "step": 860
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.700077101002314e-05,
+      "loss": 0.1249,
+      "step": 861
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.69236700077101e-05,
+      "loss": 0.7627,
+      "step": 862
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.684656900539707e-05,
+      "loss": 0.4744,
+      "step": 863
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.676946800308405e-05,
+      "loss": 0.4375,
+      "step": 864
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 6.669236700077102e-05,
+      "loss": 0.3735,
+      "step": 865
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 6.661526599845799e-05,
+      "loss": 0.9131,
+      "step": 866
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 6.653816499614495e-05,
+      "loss": 0.1492,
+      "step": 867
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 6.646106399383192e-05,
+      "loss": 0.4363,
+      "step": 868
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 6.638396299151889e-05,
+      "loss": 0.166,
+      "step": 869
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 6.630686198920586e-05,
+      "loss": 0.1625,
+      "step": 870
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 6.622976098689284e-05,
+      "loss": 0.668,
+      "step": 871
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 6.61526599845798e-05,
+      "loss": 1.0449,
+      "step": 872
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 6.607555898226677e-05,
+      "loss": 1.7256,
+      "step": 873
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 6.599845797995374e-05,
+      "loss": 0.3667,
+      "step": 874
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 6.592135697764072e-05,
+      "loss": 0.1605,
+      "step": 875
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 6.584425597532769e-05,
+      "loss": 1.1719,
+      "step": 876
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 6.576715497301465e-05,
+      "loss": 0.5776,
+      "step": 877
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 6.569005397070162e-05,
+      "loss": 0.4678,
+      "step": 878
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 6.56129529683886e-05,
+      "loss": 0.241,
+      "step": 879
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 6.553585196607557e-05,
+      "loss": 2.2949,
+      "step": 880
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 6.545875096376252e-05,
+      "loss": 0.1116,
+      "step": 881
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 6.538164996144949e-05,
+      "loss": 1.2998,
+      "step": 882
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 6.530454895913647e-05,
+      "loss": 0.3,
+      "step": 883
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 6.522744795682344e-05,
+      "loss": 0.0408,
+      "step": 884
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 6.515034695451041e-05,
+      "loss": 0.5635,
+      "step": 885
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 6.507324595219738e-05,
+      "loss": 0.26,
+      "step": 886
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 6.499614494988436e-05,
+      "loss": 0.0569,
+      "step": 887
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 6.491904394757132e-05,
+      "loss": 0.5938,
+      "step": 888
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 6.484194294525829e-05,
+      "loss": 0.4324,
+      "step": 889
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 6.476484194294526e-05,
+      "loss": 1.2383,
+      "step": 890
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 6.468774094063224e-05,
+      "loss": 0.2559,
+      "step": 891
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 6.461063993831921e-05,
+      "loss": 0.8726,
+      "step": 892
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 6.453353893600616e-05,
+      "loss": 1.251,
+      "step": 893
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 6.445643793369314e-05,
+      "loss": 1.2266,
+      "step": 894
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 6.437933693138011e-05,
+      "loss": 1.0605,
+      "step": 895
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 6.430223592906708e-05,
+      "loss": 0.2646,
+      "step": 896
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 6.422513492675404e-05,
+      "loss": 0.2389,
+      "step": 897
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 6.414803392444103e-05,
+      "loss": 0.1136,
+      "step": 898
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 6.407093292212799e-05,
+      "loss": 0.2194,
+      "step": 899
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 6.399383191981496e-05,
+      "loss": 0.2231,
+      "step": 900
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 6.391673091750193e-05,
+      "loss": 0.3203,
+      "step": 901
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 6.383962991518891e-05,
+      "loss": 0.2157,
+      "step": 902
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 6.376252891287588e-05,
+      "loss": 0.2891,
+      "step": 903
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 6.368542791056284e-05,
+      "loss": 1.0996,
+      "step": 904
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 6.360832690824981e-05,
+      "loss": 1.1914,
+      "step": 905
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 6.353122590593678e-05,
+      "loss": 0.6709,
+      "step": 906
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 6.345412490362375e-05,
+      "loss": 0.1008,
+      "step": 907
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 6.337702390131071e-05,
+      "loss": 0.4233,
+      "step": 908
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 6.32999228989977e-05,
+      "loss": 0.3667,
+      "step": 909
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 6.322282189668466e-05,
+      "loss": 0.6455,
+      "step": 910
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 6.314572089437163e-05,
+      "loss": 0.1208,
+      "step": 911
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 6.30686198920586e-05,
+      "loss": 1.9023,
+      "step": 912
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 6.299151888974558e-05,
+      "loss": 0.3062,
+      "step": 913
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 6.291441788743254e-05,
+      "loss": 0.0718,
+      "step": 914
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 6.283731688511951e-05,
+      "loss": 0.6821,
+      "step": 915
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 6.276021588280648e-05,
+      "loss": 0.1096,
+      "step": 916
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 6.268311488049346e-05,
+      "loss": 0.1227,
+      "step": 917
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.260601387818041e-05,
+      "loss": 0.2278,
+      "step": 918
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.252891287586738e-05,
+      "loss": 0.8188,
+      "step": 919
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.245181187355435e-05,
+      "loss": 0.2903,
+      "step": 920
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.237471087124133e-05,
+      "loss": 0.2057,
+      "step": 921
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.22976098689283e-05,
+      "loss": 0.3049,
+      "step": 922
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.222050886661527e-05,
+      "loss": 0.1025,
+      "step": 923
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.214340786430223e-05,
+      "loss": 0.321,
+      "step": 924
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.206630686198921e-05,
+      "loss": 0.8555,
+      "step": 925
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.198920585967618e-05,
+      "loss": 0.2546,
+      "step": 926
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 6.191210485736315e-05,
+      "loss": 0.068,
+      "step": 927
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 6.183500385505012e-05,
+      "loss": 0.6694,
+      "step": 928
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 6.17579028527371e-05,
+      "loss": 0.489,
+      "step": 929
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 6.168080185042406e-05,
+      "loss": 0.1044,
+      "step": 930
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 6.160370084811102e-05,
+      "loss": 0.8252,
+      "step": 931
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 6.1526599845798e-05,
+      "loss": 0.8516,
+      "step": 932
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 6.144949884348497e-05,
+      "loss": 0.3889,
+      "step": 933
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 6.137239784117193e-05,
+      "loss": 0.1619,
+      "step": 934
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 6.12952968388589e-05,
+      "loss": 0.1848,
+      "step": 935
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 6.121819583654588e-05,
+      "loss": 0.792,
+      "step": 936
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 6.114109483423285e-05,
+      "loss": 0.9644,
+      "step": 937
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 6.106399383191982e-05,
+      "loss": 0.0786,
+      "step": 938
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 6.098689282960679e-05,
+      "loss": 1.71,
+      "step": 939
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 6.090979182729376e-05,
+      "loss": 0.3508,
+      "step": 940
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 6.083269082498073e-05,
+      "loss": 0.1937,
+      "step": 941
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 6.07555898226677e-05,
+      "loss": 0.1797,
+      "step": 942
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 6.067848882035466e-05,
+      "loss": 0.5571,
+      "step": 943
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 6.0601387818041635e-05,
+      "loss": 0.2478,
+      "step": 944
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 6.05242868157286e-05,
+      "loss": 0.2256,
+      "step": 945
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 6.044718581341558e-05,
+      "loss": 0.1606,
+      "step": 946
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 6.0370084811102544e-05,
+      "loss": 1.8037,
+      "step": 947
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 6.029298380878952e-05,
+      "loss": 0.3386,
+      "step": 948
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 6.0215882806476486e-05,
+      "loss": 0.9424,
+      "step": 949
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 6.013878180416346e-05,
+      "loss": 0.4563,
+      "step": 950
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 6.006168080185043e-05,
+      "loss": 0.3098,
+      "step": 951
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 5.99845797995374e-05,
+      "loss": 1.0762,
+      "step": 952
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 5.990747879722437e-05,
+      "loss": 0.0321,
+      "step": 953
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 5.9830377794911343e-05,
+      "loss": 0.2556,
+      "step": 954
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 5.975327679259831e-05,
+      "loss": 0.2991,
+      "step": 955
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 5.967617579028527e-05,
+      "loss": 0.1608,
+      "step": 956
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 5.959907478797224e-05,
+      "loss": 0.4678,
+      "step": 957
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 5.952197378565921e-05,
+      "loss": 0.1064,
+      "step": 958
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 5.944487278334618e-05,
+      "loss": 0.3318,
+      "step": 959
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 5.9367771781033155e-05,
+      "loss": 0.2006,
+      "step": 960
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 5.929067077872012e-05,
+      "loss": 0.0407,
+      "step": 961
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 5.9213569776407096e-05,
+      "loss": 0.1545,
+      "step": 962
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 5.9136468774094064e-05,
+      "loss": 0.3525,
+      "step": 963
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 5.905936777178104e-05,
+      "loss": 0.293,
+      "step": 964
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 5.8982266769468005e-05,
+      "loss": 0.1218,
+      "step": 965
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 5.890516576715498e-05,
+      "loss": 0.0904,
+      "step": 966
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 5.882806476484195e-05,
+      "loss": 0.3354,
+      "step": 967
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 5.875096376252892e-05,
+      "loss": 0.3574,
+      "step": 968
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 5.867386276021588e-05,
+      "loss": 0.2659,
+      "step": 969
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 5.859676175790285e-05,
+      "loss": 0.1691,
+      "step": 970
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 5.8519660755589824e-05,
+      "loss": 0.3452,
+      "step": 971
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 5.844255975327679e-05,
+      "loss": 0.9146,
+      "step": 972
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 5.8365458750963765e-05,
+      "loss": 0.063,
+      "step": 973
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 5.828835774865073e-05,
+      "loss": 0.8843,
+      "step": 974
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 5.821125674633771e-05,
+      "loss": 0.2642,
+      "step": 975
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 5.8134155744024674e-05,
+      "loss": 0.6499,
+      "step": 976
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 5.805705474171165e-05,
+      "loss": 1.0293,
+      "step": 977
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 5.7979953739398616e-05,
+      "loss": 0.252,
+      "step": 978
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 5.790285273708559e-05,
+      "loss": 0.6362,
+      "step": 979
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 5.782575173477256e-05,
+      "loss": 0.1084,
+      "step": 980
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 5.774865073245952e-05,
+      "loss": 0.1119,
+      "step": 981
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 5.767154973014649e-05,
+      "loss": 0.1307,
+      "step": 982
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 5.759444872783346e-05,
+      "loss": 0.4487,
+      "step": 983
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 5.7517347725520434e-05,
+      "loss": 0.0582,
+      "step": 984
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 5.74402467232074e-05,
+      "loss": 0.1307,
+      "step": 985
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 5.7363145720894376e-05,
+      "loss": 0.1953,
+      "step": 986
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 5.728604471858134e-05,
+      "loss": 0.8022,
+      "step": 987
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 5.720894371626832e-05,
+      "loss": 0.292,
+      "step": 988
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 5.7131842713955285e-05,
+      "loss": 0.2849,
+      "step": 989
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 5.705474171164226e-05,
+      "loss": 0.73,
+      "step": 990
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 5.6977640709329226e-05,
+      "loss": 0.2283,
+      "step": 991
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 5.69005397070162e-05,
+      "loss": 2.5703,
+      "step": 992
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 5.682343870470317e-05,
+      "loss": 0.4678,
+      "step": 993
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 5.674633770239013e-05,
+      "loss": 0.8042,
+      "step": 994
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 5.66692367000771e-05,
+      "loss": 0.1105,
+      "step": 995
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 5.659213569776407e-05,
+      "loss": 0.1361,
+      "step": 996
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 5.6515034695451045e-05,
+      "loss": 0.4233,
+      "step": 997
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 5.643793369313801e-05,
+      "loss": 0.1292,
+      "step": 998
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 5.6360832690824986e-05,
+      "loss": 0.9385,
+      "step": 999
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 5.6283731688511954e-05,
+      "loss": 0.1225,
+      "step": 1000
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 5.620663068619893e-05,
+      "loss": 0.3562,
+      "step": 1001
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 5.6129529683885895e-05,
+      "loss": 0.3708,
+      "step": 1002
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 5.605242868157287e-05,
+      "loss": 0.0524,
+      "step": 1003
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 5.597532767925984e-05,
+      "loss": 0.0998,
+      "step": 1004
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 5.589822667694681e-05,
+      "loss": 0.2615,
+      "step": 1005
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 5.5821125674633765e-05,
+      "loss": 0.1592,
+      "step": 1006
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 5.574402467232074e-05,
+      "loss": 0.2939,
+      "step": 1007
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 5.5666923670007707e-05,
+      "loss": 0.2341,
+      "step": 1008
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 5.558982266769468e-05,
+      "loss": 0.1646,
+      "step": 1009
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 5.551272166538165e-05,
+      "loss": 0.1405,
+      "step": 1010
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 5.543562066306862e-05,
+      "loss": 0.0772,
+      "step": 1011
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 5.535851966075559e-05,
+      "loss": 0.1002,
+      "step": 1012
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 5.5281418658442564e-05,
+      "loss": 0.1469,
+      "step": 1013
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 5.520431765612953e-05,
+      "loss": 0.1213,
+      "step": 1014
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 5.5127216653816506e-05,
+      "loss": 1.0439,
+      "step": 1015
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 5.505011565150347e-05,
+      "loss": 0.2306,
+      "step": 1016
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 5.497301464919045e-05,
+      "loss": 0.0955,
+      "step": 1017
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 5.4895913646877415e-05,
+      "loss": 0.3665,
+      "step": 1018
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 5.4818812644564375e-05,
+      "loss": 0.7314,
+      "step": 1019
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 5.474171164225135e-05,
+      "loss": 0.364,
+      "step": 1020
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 5.466461063993832e-05,
+      "loss": 0.5737,
+      "step": 1021
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 5.458750963762529e-05,
+      "loss": 1.2441,
+      "step": 1022
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 5.451040863531226e-05,
+      "loss": 0.5034,
+      "step": 1023
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 5.443330763299923e-05,
+      "loss": 1.2051,
+      "step": 1024
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 5.43562066306862e-05,
+      "loss": 0.0754,
+      "step": 1025
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 5.4279105628373175e-05,
+      "loss": 0.709,
+      "step": 1026
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 5.420200462606014e-05,
+      "loss": 0.0859,
+      "step": 1027
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 5.4124903623747116e-05,
+      "loss": 0.324,
+      "step": 1028
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 5.4047802621434084e-05,
+      "loss": 0.0635,
+      "step": 1029
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 5.397070161912106e-05,
+      "loss": 0.6831,
+      "step": 1030
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 5.3893600616808025e-05,
+      "loss": 0.092,
+      "step": 1031
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 5.3816499614494986e-05,
+      "loss": 0.3835,
+      "step": 1032
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 5.373939861218196e-05,
+      "loss": 0.2937,
+      "step": 1033
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 5.366229760986893e-05,
+      "loss": 0.0645,
+      "step": 1034
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 5.35851966075559e-05,
+      "loss": 0.3757,
+      "step": 1035
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 5.350809560524287e-05,
+      "loss": 0.1304,
+      "step": 1036
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 5.343099460292984e-05,
+      "loss": 0.3657,
+      "step": 1037
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 5.335389360061681e-05,
+      "loss": 0.3699,
+      "step": 1038
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 5.3276792598303785e-05,
+      "loss": 0.2103,
+      "step": 1039
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 5.319969159599075e-05,
+      "loss": 0.1365,
+      "step": 1040
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 5.312259059367773e-05,
+      "loss": 0.1757,
+      "step": 1041
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 5.3045489591364694e-05,
+      "loss": 0.1317,
+      "step": 1042
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 5.296838858905167e-05,
+      "loss": 0.0259,
+      "step": 1043
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 5.289128758673862e-05,
+      "loss": 0.4578,
+      "step": 1044
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 5.2814186584425596e-05,
+      "loss": 0.2374,
+      "step": 1045
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 5.2737085582112564e-05,
+      "loss": 0.5176,
+      "step": 1046
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 5.265998457979954e-05,
+      "loss": 0.9272,
+      "step": 1047
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 5.2582883577486505e-05,
+      "loss": 0.6035,
+      "step": 1048
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 5.250578257517348e-05,
+      "loss": 0.2988,
+      "step": 1049
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 5.242868157286045e-05,
+      "loss": 0.2468,
+      "step": 1050
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 5.235158057054742e-05,
+      "loss": 0.083,
+      "step": 1051
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 5.227447956823439e-05,
+      "loss": 0.8149,
+      "step": 1052
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 5.219737856592136e-05,
+      "loss": 0.3259,
+      "step": 1053
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 5.212027756360833e-05,
+      "loss": 0.1233,
+      "step": 1054
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 5.2043176561295305e-05,
+      "loss": 0.6562,
+      "step": 1055
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 5.196607555898227e-05,
+      "loss": 0.0846,
+      "step": 1056
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 5.188897455666923e-05,
+      "loss": 0.0748,
+      "step": 1057
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 5.181187355435621e-05,
+      "loss": 0.1151,
+      "step": 1058
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 5.1734772552043174e-05,
+      "loss": 0.2742,
+      "step": 1059
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 5.165767154973015e-05,
+      "loss": 0.0782,
+      "step": 1060
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 5.1580570547417116e-05,
+      "loss": 0.1965,
+      "step": 1061
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 5.150346954510409e-05,
+      "loss": 0.3318,
+      "step": 1062
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 5.142636854279106e-05,
+      "loss": 0.0961,
+      "step": 1063
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 5.134926754047803e-05,
+      "loss": 0.116,
+      "step": 1064
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 5.1272166538165e-05,
+      "loss": 0.1616,
+      "step": 1065
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 5.119506553585197e-05,
+      "loss": 0.0618,
+      "step": 1066
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 5.111796453353894e-05,
+      "loss": 0.2443,
+      "step": 1067
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 5.1040863531225915e-05,
+      "loss": 0.0749,
+      "step": 1068
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 5.0963762528912876e-05,
+      "loss": 0.9692,
+      "step": 1069
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 5.088666152659984e-05,
+      "loss": 0.1871,
+      "step": 1070
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 5.080956052428682e-05,
+      "loss": 0.0464,
+      "step": 1071
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 5.0732459521973785e-05,
+      "loss": 0.1909,
+      "step": 1072
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 5.065535851966076e-05,
+      "loss": 0.1632,
+      "step": 1073
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 5.0578257517347726e-05,
+      "loss": 0.1764,
+      "step": 1074
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 5.05011565150347e-05,
+      "loss": 0.0398,
+      "step": 1075
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 5.042405551272167e-05,
+      "loss": 0.9648,
+      "step": 1076
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 5.034695451040864e-05,
+      "loss": 0.5747,
+      "step": 1077
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 5.026985350809561e-05,
+      "loss": 1.041,
+      "step": 1078
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 5.0192752505782584e-05,
+      "loss": 0.606,
+      "step": 1079
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 5.011565150346955e-05,
+      "loss": 0.1214,
+      "step": 1080
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 5.0038550501156525e-05,
+      "loss": 0.2505,
+      "step": 1081
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 4.9961449498843486e-05,
+      "loss": 0.1248,
+      "step": 1082
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 4.988434849653046e-05,
+      "loss": 0.1238,
+      "step": 1083
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 4.980724749421743e-05,
+      "loss": 1.8555,
+      "step": 1084
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 4.9730146491904395e-05,
+      "loss": 0.0548,
+      "step": 1085
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 4.965304548959136e-05,
+      "loss": 0.0336,
+      "step": 1086
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 4.957594448727834e-05,
+      "loss": 0.2184,
+      "step": 1087
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 4.9498843484965304e-05,
+      "loss": 0.0361,
+      "step": 1088
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 4.942174248265228e-05,
+      "loss": 0.2571,
+      "step": 1089
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 4.9344641480339246e-05,
+      "loss": 0.4995,
+      "step": 1090
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 4.926754047802621e-05,
+      "loss": 0.4607,
+      "step": 1091
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 4.919043947571319e-05,
+      "loss": 0.3596,
+      "step": 1092
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 4.9113338473400155e-05,
+      "loss": 0.5059,
+      "step": 1093
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 4.903623747108713e-05,
+      "loss": 0.3491,
+      "step": 1094
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 4.8959136468774097e-05,
+      "loss": 0.1655,
+      "step": 1095
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 4.888203546646107e-05,
+      "loss": 0.0481,
+      "step": 1096
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 4.880493446414804e-05,
+      "loss": 0.1058,
+      "step": 1097
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 4.8727833461835006e-05,
+      "loss": 0.8247,
+      "step": 1098
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 4.865073245952197e-05,
+      "loss": 0.1753,
+      "step": 1099
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 4.857363145720895e-05,
+      "loss": 0.0592,
+      "step": 1100
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 4.8496530454895915e-05,
+      "loss": 0.238,
+      "step": 1101
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 4.841942945258289e-05,
+      "loss": 1.0674,
+      "step": 1102
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 4.8342328450269856e-05,
+      "loss": 1.6104,
+      "step": 1103
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 4.8265227447956824e-05,
+      "loss": 0.037,
+      "step": 1104
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 4.818812644564379e-05,
+      "loss": 0.2379,
+      "step": 1105
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 4.8111025443330765e-05,
+      "loss": 0.3315,
+      "step": 1106
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 4.803392444101773e-05,
+      "loss": 1.2666,
+      "step": 1107
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 4.795682343870471e-05,
+      "loss": 0.1809,
+      "step": 1108
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 4.7879722436391674e-05,
+      "loss": 0.0294,
+      "step": 1109
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 4.780262143407864e-05,
+      "loss": 0.2444,
+      "step": 1110
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 4.7725520431765616e-05,
+      "loss": 0.0414,
+      "step": 1111
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 4.7648419429452584e-05,
+      "loss": 0.0914,
+      "step": 1112
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 4.757131842713956e-05,
+      "loss": 1.1953,
+      "step": 1113
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 4.7494217424826525e-05,
+      "loss": 0.1866,
+      "step": 1114
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 4.74171164225135e-05,
+      "loss": 0.4297,
+      "step": 1115
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 4.734001542020047e-05,
+      "loss": 0.2573,
+      "step": 1116
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 4.7262914417887434e-05,
+      "loss": 0.4158,
+      "step": 1117
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 4.71858134155744e-05,
+      "loss": 0.2666,
+      "step": 1118
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 4.7108712413261376e-05,
+      "loss": 0.0297,
+      "step": 1119
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 4.703161141094834e-05,
+      "loss": 0.4983,
+      "step": 1120
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 4.695451040863532e-05,
+      "loss": 0.0524,
+      "step": 1121
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 4.6877409406322285e-05,
+      "loss": 0.1164,
+      "step": 1122
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 4.680030840400925e-05,
+      "loss": 1.3066,
+      "step": 1123
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 4.672320740169622e-05,
+      "loss": 0.075,
+      "step": 1124
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 4.6646106399383194e-05,
+      "loss": 0.0744,
+      "step": 1125
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 4.656900539707016e-05,
+      "loss": 0.0635,
+      "step": 1126
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 4.6491904394757136e-05,
+      "loss": 0.061,
+      "step": 1127
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 4.64148033924441e-05,
+      "loss": 0.1294,
+      "step": 1128
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 4.633770239013107e-05,
+      "loss": 0.2637,
+      "step": 1129
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 4.6260601387818045e-05,
+      "loss": 0.0882,
+      "step": 1130
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 4.618350038550501e-05,
+      "loss": 0.353,
+      "step": 1131
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 4.6106399383191986e-05,
+      "loss": 0.1173,
+      "step": 1132
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 4.6029298380878954e-05,
+      "loss": 0.0536,
+      "step": 1133
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 4.595219737856593e-05,
+      "loss": 0.198,
+      "step": 1134
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 4.5875096376252895e-05,
+      "loss": 0.0435,
+      "step": 1135
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 4.579799537393986e-05,
+      "loss": 0.1599,
+      "step": 1136
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 4.572089437162683e-05,
+      "loss": 0.0831,
+      "step": 1137
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 4.5643793369313804e-05,
+      "loss": 0.1553,
+      "step": 1138
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 4.556669236700077e-05,
+      "loss": 0.0691,
+      "step": 1139
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 4.5489591364687746e-05,
+      "loss": 0.0417,
+      "step": 1140
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 4.5412490362374714e-05,
+      "loss": 0.1066,
+      "step": 1141
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 4.533538936006168e-05,
+      "loss": 0.1704,
+      "step": 1142
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 4.525828835774865e-05,
+      "loss": 0.0549,
+      "step": 1143
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 4.518118735543562e-05,
+      "loss": 0.1006,
+      "step": 1144
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 4.510408635312259e-05,
+      "loss": 0.304,
+      "step": 1145
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 4.5026985350809564e-05,
+      "loss": 0.1742,
+      "step": 1146
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 4.494988434849653e-05,
+      "loss": 0.4961,
+      "step": 1147
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 4.48727833461835e-05,
+      "loss": 0.1215,
+      "step": 1148
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 4.479568234387047e-05,
+      "loss": 0.5859,
+      "step": 1149
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 4.471858134155744e-05,
+      "loss": 0.0602,
+      "step": 1150
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 4.4641480339244415e-05,
+      "loss": 0.0396,
+      "step": 1151
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 4.456437933693138e-05,
+      "loss": 0.076,
+      "step": 1152
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 4.4487278334618357e-05,
+      "loss": 0.0558,
+      "step": 1153
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 4.441017733230532e-05,
+      "loss": 0.2476,
+      "step": 1154
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 4.433307632999229e-05,
+      "loss": 0.2421,
+      "step": 1155
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 4.425597532767926e-05,
+      "loss": 0.1367,
+      "step": 1156
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 4.417887432536623e-05,
+      "loss": 0.0757,
+      "step": 1157
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 4.41017733230532e-05,
+      "loss": 0.356,
+      "step": 1158
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 4.4024672320740175e-05,
+      "loss": 0.0928,
+      "step": 1159
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 4.394757131842714e-05,
+      "loss": 0.5005,
+      "step": 1160
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 4.387047031611411e-05,
+      "loss": 0.5923,
+      "step": 1161
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 4.379336931380108e-05,
+      "loss": 0.1904,
+      "step": 1162
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 4.371626831148805e-05,
+      "loss": 0.2102,
+      "step": 1163
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 4.363916730917502e-05,
+      "loss": 0.668,
+      "step": 1164
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 4.356206630686199e-05,
+      "loss": 0.3806,
+      "step": 1165
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 4.348496530454896e-05,
+      "loss": 0.0352,
+      "step": 1166
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 4.340786430223593e-05,
+      "loss": 0.854,
+      "step": 1167
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 4.33307632999229e-05,
+      "loss": 0.384,
+      "step": 1168
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 4.325366229760987e-05,
+      "loss": 0.4443,
+      "step": 1169
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 4.3176561295296844e-05,
+      "loss": 0.0317,
+      "step": 1170
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 4.309946029298381e-05,
+      "loss": 0.0955,
+      "step": 1171
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 4.3022359290670785e-05,
+      "loss": 0.1824,
+      "step": 1172
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 4.2945258288357746e-05,
+      "loss": 0.1453,
+      "step": 1173
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 4.286815728604472e-05,
+      "loss": 0.4683,
+      "step": 1174
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 4.279105628373169e-05,
+      "loss": 0.2732,
+      "step": 1175
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 4.271395528141866e-05,
+      "loss": 0.124,
+      "step": 1176
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 4.263685427910563e-05,
+      "loss": 0.2881,
+      "step": 1177
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 4.25597532767926e-05,
+      "loss": 0.0414,
+      "step": 1178
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 4.248265227447957e-05,
+      "loss": 0.063,
+      "step": 1179
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 4.240555127216654e-05,
+      "loss": 0.0378,
+      "step": 1180
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 4.232845026985351e-05,
+      "loss": 0.2646,
+      "step": 1181
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 4.225134926754048e-05,
+      "loss": 0.0699,
+      "step": 1182
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 4.2174248265227454e-05,
+      "loss": 0.4038,
+      "step": 1183
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 4.209714726291442e-05,
+      "loss": 0.0834,
+      "step": 1184
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 4.2020046260601396e-05,
+      "loss": 0.1848,
+      "step": 1185
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 4.1942945258288356e-05,
+      "loss": 0.1013,
+      "step": 1186
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 4.186584425597533e-05,
+      "loss": 0.0468,
+      "step": 1187
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 4.17887432536623e-05,
+      "loss": 0.035,
+      "step": 1188
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 4.171164225134927e-05,
+      "loss": 1.0605,
+      "step": 1189
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 4.163454124903624e-05,
+      "loss": 0.0646,
+      "step": 1190
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 4.1557440246723214e-05,
+      "loss": 0.0546,
+      "step": 1191
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 4.1480339244410174e-05,
+      "loss": 0.2279,
+      "step": 1192
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 4.140323824209715e-05,
+      "loss": 0.0521,
+      "step": 1193
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 4.1326137239784116e-05,
+      "loss": 1.3828,
+      "step": 1194
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 4.124903623747109e-05,
+      "loss": 0.1058,
+      "step": 1195
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 4.117193523515806e-05,
+      "loss": 0.0215,
+      "step": 1196
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 4.109483423284503e-05,
+      "loss": 0.0441,
+      "step": 1197
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 4.1017733230532e-05,
+      "loss": 0.0392,
+      "step": 1198
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 4.094063222821897e-05,
+      "loss": 0.046,
+      "step": 1199
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 4.086353122590594e-05,
+      "loss": 0.0862,
+      "step": 1200
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 4.078643022359291e-05,
+      "loss": 0.3079,
+      "step": 1201
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 4.070932922127988e-05,
+      "loss": 0.1536,
+      "step": 1202
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 4.063222821896685e-05,
+      "loss": 0.0878,
+      "step": 1203
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 4.0555127216653824e-05,
+      "loss": 0.2263,
+      "step": 1204
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 4.0478026214340785e-05,
+      "loss": 0.0508,
+      "step": 1205
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 4.040092521202776e-05,
+      "loss": 0.0187,
+      "step": 1206
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 4.0323824209714727e-05,
+      "loss": 0.1627,
+      "step": 1207
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 4.02467232074017e-05,
+      "loss": 0.0289,
+      "step": 1208
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 4.016962220508867e-05,
+      "loss": 0.3242,
+      "step": 1209
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 4.009252120277564e-05,
+      "loss": 0.6787,
+      "step": 1210
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 4.00154202004626e-05,
+      "loss": 0.0701,
+      "step": 1211
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.993831919814958e-05,
+      "loss": 0.3083,
+      "step": 1212
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.9861218195836545e-05,
+      "loss": 0.2759,
+      "step": 1213
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3.978411719352352e-05,
+      "loss": 0.0831,
+      "step": 1214
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3.9707016191210486e-05,
+      "loss": 0.1143,
+      "step": 1215
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.962991518889746e-05,
+      "loss": 0.1777,
+      "step": 1216
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.955281418658443e-05,
+      "loss": 0.0634,
+      "step": 1217
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.9475713184271395e-05,
+      "loss": 0.6279,
+      "step": 1218
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.939861218195837e-05,
+      "loss": 0.1274,
+      "step": 1219
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 3.932151117964534e-05,
+      "loss": 0.1797,
+      "step": 1220
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.924441017733231e-05,
+      "loss": 0.047,
+      "step": 1221
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.916730917501928e-05,
+      "loss": 0.2412,
+      "step": 1222
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.909020817270625e-05,
+      "loss": 0.1384,
+      "step": 1223
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.9013107170393213e-05,
+      "loss": 0.0779,
+      "step": 1224
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.893600616808019e-05,
+      "loss": 0.0634,
+      "step": 1225
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.8858905165767155e-05,
+      "loss": 0.0521,
+      "step": 1226
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.878180416345413e-05,
+      "loss": 0.0359,
+      "step": 1227
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.87047031611411e-05,
+      "loss": 0.0331,
+      "step": 1228
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.862760215882807e-05,
+      "loss": 0.0671,
+      "step": 1229
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.855050115651503e-05,
+      "loss": 0.5938,
+      "step": 1230
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.8473400154202006e-05,
+      "loss": 0.0434,
+      "step": 1231
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.839629915188897e-05,
+      "loss": 0.0723,
+      "step": 1232
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.831919814957595e-05,
+      "loss": 1.1172,
+      "step": 1233
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.8242097147262915e-05,
+      "loss": 0.0707,
+      "step": 1234
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.816499614494989e-05,
+      "loss": 0.0488,
+      "step": 1235
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.8087895142636857e-05,
+      "loss": 0.0216,
+      "step": 1236
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.8010794140323824e-05,
+      "loss": 0.0418,
+      "step": 1237
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.79336931380108e-05,
+      "loss": 0.0322,
+      "step": 1238
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.7856592135697766e-05,
+      "loss": 0.0721,
+      "step": 1239
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.777949113338474e-05,
+      "loss": 0.0328,
+      "step": 1240
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.770239013107171e-05,
+      "loss": 0.2676,
+      "step": 1241
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.762528912875868e-05,
+      "loss": 0.0698,
+      "step": 1242
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.754818812644564e-05,
+      "loss": 0.0381,
+      "step": 1243
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.7471087124132616e-05,
+      "loss": 0.7563,
+      "step": 1244
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.7393986121819584e-05,
+      "loss": 0.3145,
+      "step": 1245
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.731688511950656e-05,
+      "loss": 0.5269,
+      "step": 1246
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.7239784117193525e-05,
+      "loss": 0.1073,
+      "step": 1247
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.71626831148805e-05,
+      "loss": 0.4731,
+      "step": 1248
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.708558211256746e-05,
+      "loss": 0.0227,
+      "step": 1249
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.7008481110254434e-05,
+      "loss": 0.0535,
+      "step": 1250
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.69313801079414e-05,
+      "loss": 0.1163,
+      "step": 1251
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.6854279105628376e-05,
+      "loss": 0.3574,
+      "step": 1252
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.6777178103315343e-05,
+      "loss": 0.1658,
+      "step": 1253
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.670007710100232e-05,
+      "loss": 0.3359,
+      "step": 1254
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.6622976098689285e-05,
+      "loss": 0.1759,
+      "step": 1255
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.654587509637625e-05,
+      "loss": 0.0425,
+      "step": 1256
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.646877409406323e-05,
+      "loss": 0.1345,
+      "step": 1257
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.6391673091750194e-05,
+      "loss": 0.0467,
+      "step": 1258
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.631457208943717e-05,
+      "loss": 0.0379,
+      "step": 1259
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.6237471087124136e-05,
+      "loss": 0.032,
+      "step": 1260
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.61603700848111e-05,
+      "loss": 0.0507,
+      "step": 1261
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.608326908249807e-05,
+      "loss": 0.074,
+      "step": 1262
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.6006168080185045e-05,
+      "loss": 0.0307,
+      "step": 1263
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.592906707787201e-05,
+      "loss": 0.0342,
+      "step": 1264
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.5851966075558987e-05,
+      "loss": 0.245,
+      "step": 1265
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.5774865073245954e-05,
+      "loss": 0.0395,
+      "step": 1266
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.569776407093293e-05,
+      "loss": 0.126,
+      "step": 1267
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.562066306861989e-05,
+      "loss": 0.0478,
+      "step": 1268
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.554356206630686e-05,
+      "loss": 0.15,
+      "step": 1269
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.546646106399383e-05,
+      "loss": 0.1848,
+      "step": 1270
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.5389360061680805e-05,
+      "loss": 0.3103,
+      "step": 1271
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.531225905936777e-05,
+      "loss": 0.1377,
+      "step": 1272
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.5235158057054746e-05,
+      "loss": 0.0238,
+      "step": 1273
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.5158057054741714e-05,
+      "loss": 0.0805,
+      "step": 1274
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.508095605242868e-05,
+      "loss": 0.0368,
+      "step": 1275
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.5003855050115655e-05,
+      "loss": 0.0886,
+      "step": 1276
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.492675404780262e-05,
+      "loss": 0.1375,
+      "step": 1277
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.48496530454896e-05,
+      "loss": 0.0594,
+      "step": 1278
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.4772552043176564e-05,
+      "loss": 0.0891,
+      "step": 1279
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.469545104086353e-05,
+      "loss": 0.1381,
+      "step": 1280
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.46183500385505e-05,
+      "loss": 0.0968,
+      "step": 1281
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.4541249036237473e-05,
+      "loss": 0.0474,
+      "step": 1282
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.446414803392444e-05,
+      "loss": 0.3225,
+      "step": 1283
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.4387047031611415e-05,
+      "loss": 0.1726,
+      "step": 1284
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.430994602929838e-05,
+      "loss": 0.0319,
+      "step": 1285
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.423284502698536e-05,
+      "loss": 0.032,
+      "step": 1286
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.415574402467232e-05,
+      "loss": 0.2448,
+      "step": 1287
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.407864302235929e-05,
+      "loss": 0.0291,
+      "step": 1288
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.400154202004626e-05,
+      "loss": 0.0595,
+      "step": 1289
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.392444101773323e-05,
+      "loss": 0.0415,
+      "step": 1290
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.38473400154202e-05,
+      "loss": 0.0753,
+      "step": 1291
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.3770239013107175e-05,
+      "loss": 0.2374,
+      "step": 1292
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.369313801079414e-05,
+      "loss": 0.3269,
+      "step": 1293
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3.361603700848111e-05,
+      "loss": 0.1378,
+      "step": 1294
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.3538936006168084e-05,
+      "loss": 0.0358,
+      "step": 1295
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.346183500385505e-05,
+      "loss": 0.0627,
+      "step": 1296
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.3384734001542026e-05,
+      "loss": 0.1888,
+      "step": 1297
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.330763299922899e-05,
+      "loss": 0.2642,
+      "step": 1298
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.323053199691596e-05,
+      "loss": 0.054,
+      "step": 1299
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.315343099460293e-05,
+      "loss": 0.1151,
+      "step": 1300
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3.30763299922899e-05,
+      "loss": 0.0368,
+      "step": 1301
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.299922898997687e-05,
+      "loss": 0.0775,
+      "step": 1302
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.2922127987663844e-05,
+      "loss": 0.0641,
+      "step": 1303
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 3.284502698535081e-05,
+      "loss": 0.1465,
+      "step": 1304
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 3.2767925983037785e-05,
+      "loss": 0.0525,
+      "step": 1305
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3.2690824980724746e-05,
+      "loss": 0.0259,
+      "step": 1306
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3.261372397841172e-05,
+      "loss": 0.1158,
+      "step": 1307
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 3.253662297609869e-05,
+      "loss": 0.0955,
+      "step": 1308
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 3.245952197378566e-05,
+      "loss": 0.1311,
+      "step": 1309
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 3.238242097147263e-05,
+      "loss": 0.0464,
+      "step": 1310
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 3.2305319969159603e-05,
+      "loss": 0.1364,
+      "step": 1311
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 3.222821896684657e-05,
+      "loss": 0.0242,
+      "step": 1312
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 3.215111796453354e-05,
+      "loss": 0.0263,
+      "step": 1313
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.207401696222051e-05,
+      "loss": 0.1909,
+      "step": 1314
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.199691595990748e-05,
+      "loss": 0.376,
+      "step": 1315
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.1919814957594454e-05,
+      "loss": 0.0668,
+      "step": 1316
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.184271395528142e-05,
+      "loss": 0.0223,
+      "step": 1317
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.176561295296839e-05,
+      "loss": 0.5801,
+      "step": 1318
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.1688511950655356e-05,
+      "loss": 0.0648,
+      "step": 1319
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.161141094834233e-05,
+      "loss": 0.0425,
+      "step": 1320
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.15343099460293e-05,
+      "loss": 0.0193,
+      "step": 1321
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.145720894371627e-05,
+      "loss": 0.2277,
+      "step": 1322
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.138010794140324e-05,
+      "loss": 0.1279,
+      "step": 1323
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.130300693909021e-05,
+      "loss": 0.04,
+      "step": 1324
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.1225905936777175e-05,
+      "loss": 0.1127,
+      "step": 1325
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.114880493446415e-05,
+      "loss": 0.0294,
+      "step": 1326
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.1071703932151116e-05,
+      "loss": 0.0673,
+      "step": 1327
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.099460292983809e-05,
+      "loss": 0.0827,
+      "step": 1328
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.091750192752506e-05,
+      "loss": 0.036,
+      "step": 1329
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.084040092521203e-05,
+      "loss": 0.2029,
+      "step": 1330
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.0763299922899e-05,
+      "loss": 0.0685,
+      "step": 1331
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.068619892058597e-05,
+      "loss": 0.0183,
+      "step": 1332
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.060909791827294e-05,
+      "loss": 0.0461,
+      "step": 1333
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.053199691595991e-05,
+      "loss": 0.1296,
+      "step": 1334
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.045489591364688e-05,
+      "loss": 0.124,
+      "step": 1335
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.037779491133385e-05,
+      "loss": 0.0424,
+      "step": 1336
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.0300693909020818e-05,
+      "loss": 0.0349,
+      "step": 1337
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.022359290670779e-05,
+      "loss": 0.4756,
+      "step": 1338
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.014649190439476e-05,
+      "loss": 1.252,
+      "step": 1339
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.006939090208173e-05,
+      "loss": 0.1207,
+      "step": 1340
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.99922898997687e-05,
+      "loss": 0.0313,
+      "step": 1341
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.9915188897455672e-05,
+      "loss": 0.0612,
+      "step": 1342
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.9838087895142636e-05,
+      "loss": 0.0229,
+      "step": 1343
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.9760986892829607e-05,
+      "loss": 0.2864,
+      "step": 1344
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.9683885890516577e-05,
+      "loss": 0.0343,
+      "step": 1345
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.9606784888203548e-05,
+      "loss": 0.1375,
+      "step": 1346
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.952968388589052e-05,
+      "loss": 0.2072,
+      "step": 1347
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.945258288357749e-05,
+      "loss": 0.045,
+      "step": 1348
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.937548188126446e-05,
+      "loss": 0.7822,
+      "step": 1349
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.9298380878951425e-05,
+      "loss": 0.1543,
+      "step": 1350
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 2.9221279876638396e-05,
+      "loss": 0.0316,
+      "step": 1351
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 2.9144178874325366e-05,
+      "loss": 0.3008,
+      "step": 1352
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 2.9067077872012337e-05,
+      "loss": 0.0248,
+      "step": 1353
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 2.8989976869699308e-05,
+      "loss": 0.1029,
+      "step": 1354
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 2.891287586738628e-05,
+      "loss": 0.0385,
+      "step": 1355
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 2.8835774865073246e-05,
+      "loss": 0.0415,
+      "step": 1356
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 2.8758673862760217e-05,
+      "loss": 0.0443,
+      "step": 1357
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 2.8681572860447188e-05,
+      "loss": 0.0783,
+      "step": 1358
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 2.860447185813416e-05,
+      "loss": 0.0406,
+      "step": 1359
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 2.852737085582113e-05,
+      "loss": 0.1394,
+      "step": 1360
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 2.84502698535081e-05,
+      "loss": 0.0256,
+      "step": 1361
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 2.8373168851195064e-05,
+      "loss": 0.1747,
+      "step": 1362
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 2.8296067848882035e-05,
+      "loss": 0.0202,
+      "step": 1363
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 2.8218966846569006e-05,
+      "loss": 0.0293,
+      "step": 1364
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 2.8141865844255977e-05,
+      "loss": 0.0223,
+      "step": 1365
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 2.8064764841942948e-05,
+      "loss": 0.2147,
+      "step": 1366
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 2.798766383962992e-05,
+      "loss": 0.0194,
+      "step": 1367
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 2.7910562837316882e-05,
+      "loss": 0.0928,
+      "step": 1368
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 2.7833461835003853e-05,
+      "loss": 0.1842,
+      "step": 1369
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 2.7756360832690824e-05,
+      "loss": 0.0608,
+      "step": 1370
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 2.7679259830377795e-05,
+      "loss": 0.0203,
+      "step": 1371
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 2.7602158828064766e-05,
+      "loss": 0.0433,
+      "step": 1372
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 2.7525057825751737e-05,
+      "loss": 0.1021,
+      "step": 1373
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 2.7447956823438707e-05,
+      "loss": 0.052,
+      "step": 1374
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 2.7370855821125675e-05,
+      "loss": 2.0566,
+      "step": 1375
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 2.7293754818812646e-05,
+      "loss": 0.0335,
+      "step": 1376
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 2.7216653816499616e-05,
+      "loss": 0.0696,
+      "step": 1377
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.7139552814186587e-05,
+      "loss": 0.0443,
+      "step": 1378
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.7062451811873558e-05,
+      "loss": 0.8979,
+      "step": 1379
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.698535080956053e-05,
+      "loss": 0.157,
+      "step": 1380
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.6908249807247493e-05,
+      "loss": 0.0579,
+      "step": 1381
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 2.6831148804934464e-05,
+      "loss": 0.0414,
+      "step": 1382
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 2.6754047802621435e-05,
+      "loss": 0.0839,
+      "step": 1383
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.6676946800308405e-05,
+      "loss": 0.0617,
+      "step": 1384
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.6599845797995376e-05,
+      "loss": 0.3477,
+      "step": 1385
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.6522744795682347e-05,
+      "loss": 0.0482,
+      "step": 1386
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.644564379336931e-05,
+      "loss": 0.0583,
+      "step": 1387
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.6368542791056282e-05,
+      "loss": 0.396,
+      "step": 1388
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.6291441788743253e-05,
+      "loss": 0.0522,
+      "step": 1389
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.6214340786430224e-05,
+      "loss": 0.0469,
+      "step": 1390
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.6137239784117194e-05,
+      "loss": 0.1166,
+      "step": 1391
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.6060138781804165e-05,
+      "loss": 0.0473,
+      "step": 1392
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.5983037779491136e-05,
+      "loss": 0.058,
+      "step": 1393
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.5905936777178103e-05,
+      "loss": 0.0407,
+      "step": 1394
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.5828835774865074e-05,
+      "loss": 0.0257,
+      "step": 1395
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 2.5751734772552045e-05,
+      "loss": 0.1958,
+      "step": 1396
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 2.5674633770239016e-05,
+      "loss": 0.1718,
+      "step": 1397
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 2.5597532767925987e-05,
+      "loss": 0.0364,
+      "step": 1398
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 2.5520431765612957e-05,
+      "loss": 0.0173,
+      "step": 1399
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 2.544333076329992e-05,
+      "loss": 0.2186,
+      "step": 1400
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 2.5366229760986892e-05,
+      "loss": 0.1542,
+      "step": 1401
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 2.5289128758673863e-05,
+      "loss": 0.0203,
+      "step": 1402
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 2.5212027756360834e-05,
+      "loss": 0.069,
+      "step": 1403
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 2.5134926754047805e-05,
+      "loss": 0.1361,
+      "step": 1404
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 2.5057825751734776e-05,
+      "loss": 0.0933,
+      "step": 1405
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 2.4980724749421743e-05,
+      "loss": 0.2473,
+      "step": 1406
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 2.4903623747108714e-05,
+      "loss": 0.0387,
+      "step": 1407
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 2.482652274479568e-05,
+      "loss": 0.0208,
+      "step": 1408
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 2.4749421742482652e-05,
+      "loss": 0.0451,
+      "step": 1409
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 2.4672320740169623e-05,
+      "loss": 0.105,
+      "step": 1410
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 2.4595219737856594e-05,
+      "loss": 0.032,
+      "step": 1411
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 2.4518118735543565e-05,
+      "loss": 0.0439,
+      "step": 1412
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 2.4441017733230535e-05,
+      "loss": 0.0574,
+      "step": 1413
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 2.4363916730917503e-05,
+      "loss": 0.3,
+      "step": 1414
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 2.4286815728604474e-05,
+      "loss": 0.0779,
+      "step": 1415
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 2.4209714726291444e-05,
+      "loss": 0.0789,
+      "step": 1416
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 2.4132613723978412e-05,
+      "loss": 0.096,
+      "step": 1417
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 2.4055512721665383e-05,
+      "loss": 0.0473,
+      "step": 1418
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 2.3978411719352354e-05,
+      "loss": 0.0511,
+      "step": 1419
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 2.390131071703932e-05,
+      "loss": 0.1207,
+      "step": 1420
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 2.3824209714726292e-05,
+      "loss": 0.0303,
+      "step": 1421
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 2.3747108712413263e-05,
+      "loss": 0.0403,
+      "step": 1422
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 2.3670007710100233e-05,
+      "loss": 0.022,
+      "step": 1423
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 2.35929067077872e-05,
+      "loss": 0.0309,
+      "step": 1424
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 2.351580570547417e-05,
+      "loss": 0.3826,
+      "step": 1425
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 2.3438704703161142e-05,
+      "loss": 0.0673,
+      "step": 1426
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 2.336160370084811e-05,
+      "loss": 0.0311,
+      "step": 1427
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 2.328450269853508e-05,
+      "loss": 0.058,
+      "step": 1428
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 2.320740169622205e-05,
+      "loss": 0.0322,
+      "step": 1429
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 2.3130300693909022e-05,
+      "loss": 0.038,
+      "step": 1430
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 2.3053199691595993e-05,
+      "loss": 0.5073,
+      "step": 1431
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 2.2976098689282964e-05,
+      "loss": 0.053,
+      "step": 1432
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 2.289899768696993e-05,
+      "loss": 1.5156,
+      "step": 1433
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 2.2821896684656902e-05,
+      "loss": 0.5967,
+      "step": 1434
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 2.2744795682343873e-05,
+      "loss": 0.0256,
+      "step": 1435
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 2.266769468003084e-05,
+      "loss": 0.0145,
+      "step": 1436
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 2.259059367771781e-05,
+      "loss": 0.2117,
+      "step": 1437
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 2.2513492675404782e-05,
+      "loss": 0.2798,
+      "step": 1438
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 2.243639167309175e-05,
+      "loss": 0.2605,
+      "step": 1439
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 2.235929067077872e-05,
+      "loss": 0.1632,
+      "step": 1440
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 2.228218966846569e-05,
+      "loss": 0.0331,
+      "step": 1441
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 2.220508866615266e-05,
+      "loss": 0.0378,
+      "step": 1442
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 2.212798766383963e-05,
+      "loss": 0.0256,
+      "step": 1443
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 2.20508866615266e-05,
+      "loss": 0.2312,
+      "step": 1444
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 2.197378565921357e-05,
+      "loss": 0.0327,
+      "step": 1445
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 2.189668465690054e-05,
+      "loss": 0.0392,
+      "step": 1446
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 2.181958365458751e-05,
+      "loss": 0.0163,
+      "step": 1447
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 2.174248265227448e-05,
+      "loss": 0.057,
+      "step": 1448
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 2.166538164996145e-05,
+      "loss": 0.0286,
+      "step": 1449
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 2.1588280647648422e-05,
+      "loss": 0.0273,
+      "step": 1450
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 2.1511179645335393e-05,
+      "loss": 0.0637,
+      "step": 1451
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 2.143407864302236e-05,
+      "loss": 0.0767,
+      "step": 1452
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 2.135697764070933e-05,
+      "loss": 0.0452,
+      "step": 1453
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 2.12798766383963e-05,
+      "loss": 0.0183,
+      "step": 1454
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 2.120277563608327e-05,
+      "loss": 0.0488,
+      "step": 1455
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 2.112567463377024e-05,
+      "loss": 0.0421,
+      "step": 1456
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 2.104857363145721e-05,
+      "loss": 0.0626,
+      "step": 1457
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 2.0971472629144178e-05,
+      "loss": 0.0265,
+      "step": 1458
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 2.089437162683115e-05,
+      "loss": 0.0345,
+      "step": 1459
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 2.081727062451812e-05,
+      "loss": 0.0214,
+      "step": 1460
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 2.0740169622205087e-05,
+      "loss": 0.0139,
+      "step": 1461
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 2.0663068619892058e-05,
+      "loss": 0.0647,
+      "step": 1462
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 2.058596761757903e-05,
+      "loss": 0.0426,
+      "step": 1463
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 2.0508866615266e-05,
+      "loss": 0.0713,
+      "step": 1464
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 2.043176561295297e-05,
+      "loss": 0.0924,
+      "step": 1465
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 2.035466461063994e-05,
+      "loss": 0.0199,
+      "step": 1466
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 2.0277563608326912e-05,
+      "loss": 0.4607,
+      "step": 1467
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 2.020046260601388e-05,
+      "loss": 0.0692,
+      "step": 1468
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 2.012336160370085e-05,
+      "loss": 0.0586,
+      "step": 1469
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 2.004626060138782e-05,
+      "loss": 0.054,
+      "step": 1470
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 1.996915959907479e-05,
+      "loss": 0.1055,
+      "step": 1471
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 1.989205859676176e-05,
+      "loss": 0.0463,
+      "step": 1472
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 1.981495759444873e-05,
+      "loss": 0.1218,
+      "step": 1473
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 1.9737856592135698e-05,
+      "loss": 0.0574,
+      "step": 1474
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 1.966075558982267e-05,
+      "loss": 0.0231,
+      "step": 1475
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 1.958365458750964e-05,
+      "loss": 0.0258,
+      "step": 1476
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 1.9506553585196607e-05,
+      "loss": 0.0295,
+      "step": 1477
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 1.9429452582883578e-05,
+      "loss": 0.0712,
+      "step": 1478
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 1.935235158057055e-05,
+      "loss": 0.0254,
+      "step": 1479
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 1.9275250578257516e-05,
+      "loss": 0.0225,
+      "step": 1480
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 1.9198149575944487e-05,
+      "loss": 0.2045,
+      "step": 1481
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 1.9121048573631457e-05,
+      "loss": 0.0278,
+      "step": 1482
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 1.9043947571318428e-05,
+      "loss": 0.0602,
+      "step": 1483
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 1.89668465690054e-05,
+      "loss": 0.1967,
+      "step": 1484
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 1.888974556669237e-05,
+      "loss": 0.5464,
+      "step": 1485
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 1.881264456437934e-05,
+      "loss": 0.0376,
+      "step": 1486
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 1.8735543562066308e-05,
+      "loss": 0.0255,
+      "step": 1487
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 1.865844255975328e-05,
+      "loss": 0.016,
+      "step": 1488
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 1.858134155744025e-05,
+      "loss": 0.0684,
+      "step": 1489
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 1.8504240555127217e-05,
+      "loss": 0.0241,
+      "step": 1490
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 1.8427139552814188e-05,
+      "loss": 0.1476,
+      "step": 1491
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 1.835003855050116e-05,
+      "loss": 0.0277,
+      "step": 1492
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 1.8272937548188126e-05,
+      "loss": 0.0216,
+      "step": 1493
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 1.8195836545875097e-05,
+      "loss": 0.1067,
+      "step": 1494
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 1.8118735543562068e-05,
+      "loss": 0.0264,
+      "step": 1495
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 1.8041634541249035e-05,
+      "loss": 0.2571,
+      "step": 1496
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 1.7964533538936006e-05,
+      "loss": 0.2415,
+      "step": 1497
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 1.7887432536622977e-05,
+      "loss": 0.0395,
+      "step": 1498
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 1.7810331534309944e-05,
+      "loss": 0.2196,
+      "step": 1499
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 1.7733230531996915e-05,
+      "loss": 0.0191,
+      "step": 1500
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 1.7656129529683886e-05,
+      "loss": 0.0415,
+      "step": 1501
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 1.7579028527370857e-05,
+      "loss": 0.1031,
+      "step": 1502
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 1.7501927525057828e-05,
+      "loss": 0.1393,
+      "step": 1503
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 1.74248265227448e-05,
+      "loss": 0.0608,
+      "step": 1504
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 1.7347725520431766e-05,
+      "loss": 0.0417,
+      "step": 1505
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 1.7270624518118737e-05,
+      "loss": 0.0481,
+      "step": 1506
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 1.7193523515805708e-05,
+      "loss": 0.0301,
+      "step": 1507
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 1.711642251349268e-05,
+      "loss": 0.0247,
+      "step": 1508
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 1.7039321511179646e-05,
+      "loss": 0.027,
+      "step": 1509
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 1.6962220508866617e-05,
+      "loss": 0.0264,
+      "step": 1510
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 1.6885119506553587e-05,
+      "loss": 0.0904,
+      "step": 1511
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 1.6808018504240555e-05,
+      "loss": 0.0398,
+      "step": 1512
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 1.6730917501927526e-05,
+      "loss": 0.1307,
+      "step": 1513
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 1.6653816499614496e-05,
+      "loss": 0.026,
+      "step": 1514
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 1.6576715497301464e-05,
+      "loss": 0.0423,
+      "step": 1515
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 1.6499614494988435e-05,
+      "loss": 0.121,
+      "step": 1516
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 1.6422513492675406e-05,
+      "loss": 0.129,
+      "step": 1517
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 1.6345412490362373e-05,
+      "loss": 0.1036,
+      "step": 1518
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 1.6268311488049344e-05,
+      "loss": 0.0513,
+      "step": 1519
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 1.6191210485736315e-05,
+      "loss": 0.1328,
+      "step": 1520
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 1.6114109483423285e-05,
+      "loss": 0.0222,
+      "step": 1521
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 1.6037008481110256e-05,
+      "loss": 0.6226,
+      "step": 1522
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 1.5959907478797227e-05,
+      "loss": 0.0723,
+      "step": 1523
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 1.5882806476484195e-05,
+      "loss": 0.0766,
+      "step": 1524
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 1.5805705474171165e-05,
+      "loss": 0.1259,
+      "step": 1525
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 1.5728604471858136e-05,
+      "loss": 0.1005,
+      "step": 1526
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 1.5651503469545104e-05,
+      "loss": 0.0206,
+      "step": 1527
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 1.5574402467232074e-05,
+      "loss": 0.0689,
+      "step": 1528
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 1.5497301464919045e-05,
+      "loss": 0.0417,
+      "step": 1529
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 1.5420200462606016e-05,
+      "loss": 0.0649,
+      "step": 1530
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 1.5343099460292983e-05,
+      "loss": 0.0685,
+      "step": 1531
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 1.5265998457979954e-05,
+      "loss": 0.0252,
+      "step": 1532
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 1.5188897455666925e-05,
+      "loss": 0.0339,
+      "step": 1533
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 1.5111796453353894e-05,
+      "loss": 0.0458,
+      "step": 1534
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 1.5034695451040865e-05,
+      "loss": 0.3564,
+      "step": 1535
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 1.4957594448727836e-05,
+      "loss": 0.0216,
+      "step": 1536
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 1.4880493446414803e-05,
+      "loss": 0.0901,
+      "step": 1537
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 1.4803392444101774e-05,
+      "loss": 1.0742,
+      "step": 1538
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 1.4726291441788745e-05,
+      "loss": 0.063,
+      "step": 1539
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 1.4649190439475712e-05,
+      "loss": 0.0418,
+      "step": 1540
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 1.4572089437162683e-05,
+      "loss": 0.0296,
+      "step": 1541
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 1.4494988434849654e-05,
+      "loss": 0.0439,
+      "step": 1542
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 1.4417887432536623e-05,
+      "loss": 0.0259,
+      "step": 1543
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 1.4340786430223594e-05,
+      "loss": 0.0437,
+      "step": 1544
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 1.4263685427910565e-05,
+      "loss": 0.0643,
+      "step": 1545
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 1.4186584425597532e-05,
+      "loss": 0.0818,
+      "step": 1546
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 1.4109483423284503e-05,
+      "loss": 0.0188,
+      "step": 1547
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 1.4032382420971474e-05,
+      "loss": 0.0188,
+      "step": 1548
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 1.3955281418658441e-05,
+      "loss": 0.0739,
+      "step": 1549
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 1.3878180416345412e-05,
+      "loss": 0.0196,
+      "step": 1550
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 1.3801079414032383e-05,
+      "loss": 0.0594,
+      "step": 1551
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 1.3723978411719354e-05,
+      "loss": 0.0257,
+      "step": 1552
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 1.3646877409406323e-05,
+      "loss": 0.0193,
+      "step": 1553
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 1.3569776407093294e-05,
+      "loss": 0.0196,
+      "step": 1554
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 1.3492675404780264e-05,
+      "loss": 0.0707,
+      "step": 1555
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 1.3415574402467232e-05,
+      "loss": 0.0248,
+      "step": 1556
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.3338473400154203e-05,
+      "loss": 0.1671,
+      "step": 1557
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.3261372397841174e-05,
+      "loss": 0.0324,
+      "step": 1558
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.3184271395528141e-05,
+      "loss": 0.0627,
+      "step": 1559
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.3107170393215112e-05,
+      "loss": 0.0214,
+      "step": 1560
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.3030069390902083e-05,
+      "loss": 0.0823,
+      "step": 1561
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.2952968388589052e-05,
+      "loss": 0.0399,
+      "step": 1562
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.2875867386276023e-05,
+      "loss": 0.0221,
+      "step": 1563
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.2798766383962993e-05,
+      "loss": 0.0396,
+      "step": 1564
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.272166538164996e-05,
+      "loss": 0.0399,
+      "step": 1565
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.2644564379336932e-05,
+      "loss": 0.1547,
+      "step": 1566
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.2567463377023902e-05,
+      "loss": 0.0283,
+      "step": 1567
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.2490362374710872e-05,
+      "loss": 0.0372,
+      "step": 1568
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.241326137239784e-05,
+      "loss": 0.0327,
+      "step": 1569
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 1.2336160370084811e-05,
+      "loss": 0.074,
+      "step": 1570
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 1.2259059367771782e-05,
+      "loss": 0.0184,
+      "step": 1571
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 1.2181958365458751e-05,
+      "loss": 0.0191,
+      "step": 1572
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 1.2104857363145722e-05,
+      "loss": 0.0364,
+      "step": 1573
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 1.2027756360832691e-05,
+      "loss": 0.0248,
+      "step": 1574
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 1.195065535851966e-05,
+      "loss": 0.0814,
+      "step": 1575
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 1.1873554356206631e-05,
+      "loss": 0.0331,
+      "step": 1576
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 1.17964533538936e-05,
+      "loss": 0.0178,
+      "step": 1577
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 1.1719352351580571e-05,
+      "loss": 0.0286,
+      "step": 1578
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 1.164225134926754e-05,
+      "loss": 0.0837,
+      "step": 1579
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 1.1565150346954511e-05,
+      "loss": 0.0404,
+      "step": 1580
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 1.1488049344641482e-05,
+      "loss": 0.0345,
+      "step": 1581
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 1.1410948342328451e-05,
+      "loss": 0.0608,
+      "step": 1582
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 1.133384734001542e-05,
+      "loss": 0.0253,
+      "step": 1583
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 1.1256746337702391e-05,
+      "loss": 0.0305,
+      "step": 1584
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 1.117964533538936e-05,
+      "loss": 0.0732,
+      "step": 1585
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 1.110254433307633e-05,
+      "loss": 0.0217,
+      "step": 1586
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 1.10254433307633e-05,
+      "loss": 0.074,
+      "step": 1587
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 1.094834232845027e-05,
+      "loss": 0.4382,
+      "step": 1588
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 1.087124132613724e-05,
+      "loss": 0.0224,
+      "step": 1589
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 1.0794140323824211e-05,
+      "loss": 0.0476,
+      "step": 1590
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 1.071703932151118e-05,
+      "loss": 0.093,
+      "step": 1591
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 1.063993831919815e-05,
+      "loss": 0.0462,
+      "step": 1592
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 1.056283731688512e-05,
+      "loss": 0.047,
+      "step": 1593
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 1.0485736314572089e-05,
+      "loss": 0.0217,
+      "step": 1594
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 1.040863531225906e-05,
+      "loss": 0.055,
+      "step": 1595
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 1.0331534309946029e-05,
+      "loss": 0.0162,
+      "step": 1596
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 1.0254433307633e-05,
+      "loss": 0.0975,
+      "step": 1597
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 1.017733230531997e-05,
+      "loss": 0.0387,
+      "step": 1598
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 1.010023130300694e-05,
+      "loss": 0.2158,
+      "step": 1599
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 1.002313030069391e-05,
+      "loss": 0.021,
+      "step": 1600
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 9.94602929838088e-06,
+      "loss": 0.101,
+      "step": 1601
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 9.868928296067849e-06,
+      "loss": 0.0249,
+      "step": 1602
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 9.79182729375482e-06,
+      "loss": 0.1372,
+      "step": 1603
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 9.714726291441789e-06,
+      "loss": 0.0396,
+      "step": 1604
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 9.637625289128758e-06,
+      "loss": 0.0177,
+      "step": 1605
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 9.560524286815729e-06,
+      "loss": 0.0507,
+      "step": 1606
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 9.4834232845027e-06,
+      "loss": 0.0288,
+      "step": 1607
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 9.40632228218967e-06,
+      "loss": 0.0832,
+      "step": 1608
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 9.32922127987664e-06,
+      "loss": 0.0263,
+      "step": 1609
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 9.252120277563609e-06,
+      "loss": 0.0646,
+      "step": 1610
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 9.17501927525058e-06,
+      "loss": 0.0214,
+      "step": 1611
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 9.097918272937549e-06,
+      "loss": 0.0277,
+      "step": 1612
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 9.020817270624518e-06,
+      "loss": 0.0399,
+      "step": 1613
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 8.943716268311488e-06,
+      "loss": 0.0242,
+      "step": 1614
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 8.866615265998458e-06,
+      "loss": 0.0188,
+      "step": 1615
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 8.789514263685428e-06,
+      "loss": 0.05,
+      "step": 1616
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 8.7124132613724e-06,
+      "loss": 0.0198,
+      "step": 1617
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 8.635312259059368e-06,
+      "loss": 0.0463,
+      "step": 1618
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 8.55821125674634e-06,
+      "loss": 0.0265,
+      "step": 1619
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 8.481110254433308e-06,
+      "loss": 0.0194,
+      "step": 1620
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 8.404009252120277e-06,
+      "loss": 0.0318,
+      "step": 1621
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 8.326908249807248e-06,
+      "loss": 0.0227,
+      "step": 1622
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 8.249807247494217e-06,
+      "loss": 0.199,
+      "step": 1623
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 8.172706245181186e-06,
+      "loss": 0.0348,
+      "step": 1624
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 8.095605242868157e-06,
+      "loss": 0.0229,
+      "step": 1625
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 8.018504240555128e-06,
+      "loss": 0.035,
+      "step": 1626
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 7.941403238242097e-06,
+      "loss": 0.0202,
+      "step": 1627
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 7.864302235929068e-06,
+      "loss": 0.0164,
+      "step": 1628
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 7.787201233616037e-06,
+      "loss": 0.5103,
+      "step": 1629
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 7.710100231303008e-06,
+      "loss": 0.0965,
+      "step": 1630
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 7.632999228989977e-06,
+      "loss": 0.0231,
+      "step": 1631
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 7.555898226676947e-06,
+      "loss": 0.0202,
+      "step": 1632
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 7.478797224363918e-06,
+      "loss": 0.0408,
+      "step": 1633
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 7.401696222050887e-06,
+      "loss": 0.0841,
+      "step": 1634
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 7.324595219737856e-06,
+      "loss": 0.0142,
+      "step": 1635
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 7.247494217424827e-06,
+      "loss": 0.0222,
+      "step": 1636
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 7.170393215111797e-06,
+      "loss": 0.0135,
+      "step": 1637
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 7.093292212798766e-06,
+      "loss": 0.0306,
+      "step": 1638
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 7.016191210485737e-06,
+      "loss": 0.0396,
+      "step": 1639
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 6.939090208172706e-06,
+      "loss": 0.0534,
+      "step": 1640
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 6.861989205859677e-06,
+      "loss": 0.0715,
+      "step": 1641
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 6.784888203546647e-06,
+      "loss": 0.0379,
+      "step": 1642
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 6.707787201233616e-06,
+      "loss": 0.0532,
+      "step": 1643
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 6.630686198920587e-06,
+      "loss": 0.0283,
+      "step": 1644
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 6.553585196607556e-06,
+      "loss": 0.3547,
+      "step": 1645
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 6.476484194294526e-06,
+      "loss": 0.0982,
+      "step": 1646
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 6.399383191981497e-06,
+      "loss": 0.0181,
+      "step": 1647
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 6.322282189668466e-06,
+      "loss": 0.0502,
+      "step": 1648
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 6.245181187355436e-06,
+      "loss": 0.0156,
+      "step": 1649
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 6.168080185042406e-06,
+      "loss": 0.025,
+      "step": 1650
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 6.090979182729376e-06,
+      "loss": 0.0163,
+      "step": 1651
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 6.013878180416346e-06,
+      "loss": 0.0195,
+      "step": 1652
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.936777178103316e-06,
+      "loss": 0.0259,
+      "step": 1653
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.859676175790286e-06,
+      "loss": 0.3464,
+      "step": 1654
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.782575173477256e-06,
+      "loss": 0.0249,
+      "step": 1655
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.7054741711642256e-06,
+      "loss": 0.0187,
+      "step": 1656
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.6283731688511955e-06,
+      "loss": 0.046,
+      "step": 1657
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.551272166538165e-06,
+      "loss": 0.5039,
+      "step": 1658
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.474171164225135e-06,
+      "loss": 0.0365,
+      "step": 1659
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.3970701619121054e-06,
+      "loss": 0.0234,
+      "step": 1660
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.319969159599075e-06,
+      "loss": 0.0196,
+      "step": 1661
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.2428681572860445e-06,
+      "loss": 0.0544,
+      "step": 1662
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.1657671549730145e-06,
+      "loss": 0.0895,
+      "step": 1663
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.088666152659985e-06,
+      "loss": 0.025,
+      "step": 1664
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.011565150346955e-06,
+      "loss": 0.0798,
+      "step": 1665
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 4.934464148033924e-06,
+      "loss": 0.0504,
+      "step": 1666
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 4.857363145720894e-06,
+      "loss": 0.0326,
+      "step": 1667
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 4.780262143407864e-06,
+      "loss": 0.1388,
+      "step": 1668
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 4.703161141094835e-06,
+      "loss": 0.0321,
+      "step": 1669
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 4.626060138781804e-06,
+      "loss": 0.033,
+      "step": 1670
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 4.548959136468774e-06,
+      "loss": 0.0492,
+      "step": 1671
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 4.471858134155744e-06,
+      "loss": 0.0257,
+      "step": 1672
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 4.394757131842714e-06,
+      "loss": 0.0324,
+      "step": 1673
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 4.317656129529684e-06,
+      "loss": 0.1075,
+      "step": 1674
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 4.240555127216654e-06,
+      "loss": 0.0892,
+      "step": 1675
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 4.163454124903624e-06,
+      "loss": 0.0868,
+      "step": 1676
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 4.086353122590593e-06,
+      "loss": 0.0593,
+      "step": 1677
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 4.009252120277564e-06,
+      "loss": 0.0444,
+      "step": 1678
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 3.932151117964534e-06,
+      "loss": 0.0438,
+      "step": 1679
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 3.855050115651504e-06,
+      "loss": 0.0236,
+      "step": 1680
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 3.7779491133384736e-06,
+      "loss": 0.0242,
+      "step": 1681
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 3.7008481110254435e-06,
+      "loss": 0.0336,
+      "step": 1682
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 3.6237471087124135e-06,
+      "loss": 0.0374,
+      "step": 1683
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 3.546646106399383e-06,
+      "loss": 0.1829,
+      "step": 1684
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 3.469545104086353e-06,
+      "loss": 0.1229,
+      "step": 1685
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3.3924441017733234e-06,
+      "loss": 0.0445,
+      "step": 1686
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3.3153430994602934e-06,
+      "loss": 0.0167,
+      "step": 1687
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 3.238242097147263e-06,
+      "loss": 0.0654,
+      "step": 1688
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 3.161141094834233e-06,
+      "loss": 0.04,
+      "step": 1689
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 3.084040092521203e-06,
+      "loss": 0.0208,
+      "step": 1690
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 3.006939090208173e-06,
+      "loss": 0.0469,
+      "step": 1691
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 2.929838087895143e-06,
+      "loss": 0.1353,
+      "step": 1692
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 2.8527370855821128e-06,
+      "loss": 0.0205,
+      "step": 1693
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 2.7756360832690823e-06,
+      "loss": 0.0163,
+      "step": 1694
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 2.6985350809560527e-06,
+      "loss": 0.0667,
+      "step": 1695
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 2.6214340786430223e-06,
+      "loss": 0.0952,
+      "step": 1696
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 2.5443330763299927e-06,
+      "loss": 0.0162,
+      "step": 1697
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 2.467232074016962e-06,
+      "loss": 0.2007,
+      "step": 1698
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 2.390131071703932e-06,
+      "loss": 0.014,
+      "step": 1699
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 2.313030069390902e-06,
+      "loss": 0.0358,
+      "step": 1700
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 2.235929067077872e-06,
+      "loss": 0.0195,
+      "step": 1701
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 2.158828064764842e-06,
+      "loss": 0.0242,
+      "step": 1702
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 2.081727062451812e-06,
+      "loss": 0.0172,
+      "step": 1703
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 2.004626060138782e-06,
+      "loss": 0.0331,
+      "step": 1704
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 1.927525057825752e-06,
+      "loss": 0.0256,
+      "step": 1705
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 1.8504240555127218e-06,
+      "loss": 0.0442,
+      "step": 1706
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 1.7733230531996915e-06,
+      "loss": 0.0504,
+      "step": 1707
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 1.6962220508866617e-06,
+      "loss": 0.018,
+      "step": 1708
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 1.6191210485736315e-06,
+      "loss": 0.0514,
+      "step": 1709
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 1.5420200462606014e-06,
+      "loss": 0.2229,
+      "step": 1710
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 1.4649190439475714e-06,
+      "loss": 0.0291,
+      "step": 1711
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 1.3878180416345412e-06,
+      "loss": 0.0298,
+      "step": 1712
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 1.3107170393215111e-06,
+      "loss": 0.0421,
+      "step": 1713
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 1.233616037008481e-06,
+      "loss": 0.0378,
+      "step": 1714
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 1.156515034695451e-06,
+      "loss": 0.017,
+      "step": 1715
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 1.079414032382421e-06,
+      "loss": 0.0206,
+      "step": 1716
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 1.002313030069391e-06,
+      "loss": 0.0207,
+      "step": 1717
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 9.252120277563609e-07,
+      "loss": 0.025,
+      "step": 1718
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 8.481110254433309e-07,
+      "loss": 0.0151,
+      "step": 1719
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 7.710100231303007e-07,
+      "loss": 0.1569,
+      "step": 1720
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 6.939090208172706e-07,
+      "loss": 0.019,
+      "step": 1721
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 6.168080185042406e-07,
+      "loss": 0.0298,
+      "step": 1722
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.397070161912105e-07,
+      "loss": 0.0221,
+      "step": 1723
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 4.6260601387818044e-07,
+      "loss": 0.0191,
+      "step": 1724
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 3.8550501156515036e-07,
+      "loss": 0.0321,
+      "step": 1725
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 3.084040092521203e-07,
+      "loss": 0.0206,
+      "step": 1726
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 2.3130300693909022e-07,
+      "loss": 0.4175,
+      "step": 1727
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 1.5420200462606014e-07,
+      "loss": 0.0243,
+      "step": 1728
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 7.710100231303007e-08,
+      "loss": 0.0995,
+      "step": 1729
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 0.0,
+      "loss": 0.1494,
+      "step": 1730
+    },
+    {
+      "epoch": 10.0,
+      "step": 1730,
+      "total_flos": 3.3456234411117773e+21,
+      "train_loss": 1.1823103138477127,
+      "train_runtime": 927.423,
+      "train_samples_per_second": 239.308,
+      "train_steps_per_second": 1.865
+    }
+  ],
+  "max_steps": 1730,
+  "num_train_epochs": 10,
+  "total_flos": 3.3456234411117773e+21,
+  "trial_name": null,
+  "trial_params": null
+}