diff --git "a/INLegalLlama/CPT/llama2_cpt_checkpoint_3000_seq_2048/trainer_state.json" "b/INLegalLlama/CPT/llama2_cpt_checkpoint_3000_seq_2048/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/INLegalLlama/CPT/llama2_cpt_checkpoint_3000_seq_2048/trainer_state.json"
@@ -0,0 +1,18064 @@
+{
+  "best_metric": 1.6164995431900024,
+  "best_model_checkpoint": "/content/drive/MyDrive/LegalLLM/Medium-LLAMA2-max_seq_len-4096/checkpoint-3000",
+  "epoch": 0.3470113646221914,
+  "global_step": 3000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3094688221709008e-07,
+      "loss": 2.0846,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6189376443418015e-07,
+      "loss": 2.0732,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.928406466512703e-07,
+      "loss": 2.0352,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.237875288683603e-07,
+      "loss": 1.9817,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1547344110854505e-06,
+      "loss": 2.092,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3856812933025406e-06,
+      "loss": 2.0729,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6166281755196307e-06,
+      "loss": 2.0448,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8475750577367206e-06,
+      "loss": 1.9562,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0785219399538105e-06,
+      "loss": 2.0066,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.309468822170901e-06,
+      "loss": 1.921,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5404157043879907e-06,
+      "loss": 2.0488,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7713625866050812e-06,
+      "loss": 1.9696,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.002309468822171e-06,
+      "loss": 2.0639,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2332563510392614e-06,
+      "loss": 2.0748,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.464203233256351e-06,
+      "loss": 2.1176,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6951501154734412e-06,
+      "loss": 1.9372,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.926096997690531e-06,
+      "loss": 2.0593,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.157043879907621e-06,
+      "loss": 2.0553,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.3879907621247115e-06,
+      "loss": 1.9244,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.618937644341802e-06,
+      "loss": 2.0149,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.849884526558892e-06,
+      "loss": 2.0162,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.080831408775981e-06,
+      "loss": 2.0238,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.311778290993072e-06,
+      "loss": 2.0295,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.5427251732101625e-06,
+      "loss": 1.9885,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.773672055427252e-06,
+      "loss": 2.0215,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.004618937644342e-06,
+      "loss": 2.0808,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.235565819861432e-06,
+      "loss": 2.0348,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.466512702078523e-06,
+      "loss": 1.9852,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.6974595842956126e-06,
+      "loss": 2.0443,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.928406466512702e-06,
+      "loss": 2.0176,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.159353348729793e-06,
+      "loss": 2.0441,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.3903002309468824e-06,
+      "loss": 1.8899,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.621247113163972e-06,
+      "loss": 2.0094,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.852193995381063e-06,
+      "loss": 2.006,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.083140877598153e-06,
+      "loss": 2.0552,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.314087759815242e-06,
+      "loss": 1.9298,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.545034642032333e-06,
+      "loss": 2.02,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.775981524249423e-06,
+      "loss": 2.1267,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.006928406466514e-06,
+      "loss": 2.1137,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.237875288683604e-06,
+      "loss": 2.0607,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.468822170900693e-06,
+      "loss": 2.0968,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.699769053117783e-06,
+      "loss": 2.1135,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.930715935334874e-06,
+      "loss": 2.1301,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0161662817551963e-05,
+      "loss": 2.1014,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0392609699769053e-05,
+      "loss": 1.9971,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0623556581986144e-05,
+      "loss": 2.174,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0854503464203233e-05,
+      "loss": 2.2499,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1085450346420325e-05,
+      "loss": 2.2668,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1316397228637414e-05,
+      "loss": 2.5706,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1547344110854504e-05,
+      "loss": 2.589,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 1.9031355381011963,
+      "eval_runtime": 4346.2195,
+      "eval_samples_per_second": 2.816,
+      "eval_steps_per_second": 0.352,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1778290993071595e-05,
+      "loss": 1.9493,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2009237875288684e-05,
+      "loss": 1.8829,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2240184757505774e-05,
+      "loss": 1.8731,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2471131639722865e-05,
+      "loss": 1.9774,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2702078521939955e-05,
+      "loss": 1.9232,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2933025404157046e-05,
+      "loss": 1.934,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3163972286374135e-05,
+      "loss": 1.8979,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3394919168591225e-05,
+      "loss": 1.8454,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3625866050808314e-05,
+      "loss": 1.9291,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3856812933025404e-05,
+      "loss": 1.8771,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4087759815242497e-05,
+      "loss": 1.8711,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4318706697459586e-05,
+      "loss": 1.9267,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4549653579676676e-05,
+      "loss": 1.8762,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4780600461893765e-05,
+      "loss": 1.9624,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5011547344110854e-05,
+      "loss": 1.7954,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5242494226327944e-05,
+      "loss": 1.8662,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5473441108545038e-05,
+      "loss": 1.8557,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5704387990762125e-05,
+      "loss": 1.8732,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5935334872979216e-05,
+      "loss": 1.791,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6166281755196306e-05,
+      "loss": 1.9255,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6397228637413393e-05,
+      "loss": 1.8677,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6628175519630484e-05,
+      "loss": 1.8697,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6859122401847578e-05,
+      "loss": 2.0248,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7090069284064665e-05,
+      "loss": 1.8714,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7321016166281756e-05,
+      "loss": 1.8957,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7551963048498846e-05,
+      "loss": 1.7782,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7782909930715937e-05,
+      "loss": 1.9138,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8013856812933027e-05,
+      "loss": 1.8843,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8244803695150118e-05,
+      "loss": 1.89,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8475750577367208e-05,
+      "loss": 1.8072,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8706697459584295e-05,
+      "loss": 1.7224,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8937644341801386e-05,
+      "loss": 1.9273,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9168591224018476e-05,
+      "loss": 1.8944,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9399538106235567e-05,
+      "loss": 1.8449,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9630484988452657e-05,
+      "loss": 1.8296,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9861431870669748e-05,
+      "loss": 1.8449,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0092378752886835e-05,
+      "loss": 1.832,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0323325635103926e-05,
+      "loss": 1.989,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0554272517321016e-05,
+      "loss": 1.9413,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0785219399538107e-05,
+      "loss": 1.904,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1016166281755197e-05,
+      "loss": 1.8749,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1247113163972288e-05,
+      "loss": 1.9168,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1478060046189378e-05,
+      "loss": 1.9331,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1709006928406465e-05,
+      "loss": 1.8452,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1939953810623556e-05,
+      "loss": 1.9158,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.217090069284065e-05,
+      "loss": 1.9955,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2401847575057737e-05,
+      "loss": 2.0556,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2632794457274828e-05,
+      "loss": 2.0924,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2863741339491918e-05,
+      "loss": 2.0482,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.309468822170901e-05,
+      "loss": 2.2229,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 1.8028513193130493,
+      "eval_runtime": 4347.4093,
+      "eval_samples_per_second": 2.815,
+      "eval_steps_per_second": 0.352,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3325635103926096e-05,
+      "loss": 1.7893,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.355658198614319e-05,
+      "loss": 1.8141,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.378752886836028e-05,
+      "loss": 1.9125,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4018475750577367e-05,
+      "loss": 1.773,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4249422632794458e-05,
+      "loss": 1.7314,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.448036951501155e-05,
+      "loss": 1.7397,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.471131639722864e-05,
+      "loss": 1.7579,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.494226327944573e-05,
+      "loss": 1.8101,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.517321016166282e-05,
+      "loss": 1.8257,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.540415704387991e-05,
+      "loss": 1.7937,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5635103926096998e-05,
+      "loss": 1.7566,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.586605080831409e-05,
+      "loss": 1.8354,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.609699769053118e-05,
+      "loss": 1.7677,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.632794457274827e-05,
+      "loss": 1.8853,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6558891454965356e-05,
+      "loss": 1.9083,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.678983833718245e-05,
+      "loss": 1.7963,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.702078521939954e-05,
+      "loss": 1.7374,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7251732101616628e-05,
+      "loss": 1.6736,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7482678983833722e-05,
+      "loss": 1.845,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.771362586605081e-05,
+      "loss": 1.7152,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.79445727482679e-05,
+      "loss": 1.9232,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8175519630484993e-05,
+      "loss": 1.7572,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.840646651270208e-05,
+      "loss": 1.7617,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.863741339491917e-05,
+      "loss": 1.8282,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8868360277136258e-05,
+      "loss": 1.7843,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9099307159353352e-05,
+      "loss": 1.7754,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9330254041570436e-05,
+      "loss": 1.8754,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.956120092378753e-05,
+      "loss": 1.7872,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9792147806004624e-05,
+      "loss": 1.771,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0023094688221707e-05,
+      "loss": 1.9019,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.02540415704388e-05,
+      "loss": 1.8207,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.048498845265589e-05,
+      "loss": 1.8205,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.071593533487298e-05,
+      "loss": 1.8155,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0946882217090076e-05,
+      "loss": 1.7486,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.117782909930716e-05,
+      "loss": 1.7522,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.140877598152425e-05,
+      "loss": 1.83,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.163972286374134e-05,
+      "loss": 1.8351,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.187066974595843e-05,
+      "loss": 1.7584,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2101616628175526e-05,
+      "loss": 1.7932,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.233256351039261e-05,
+      "loss": 1.7944,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.25635103926097e-05,
+      "loss": 1.803,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.279445727482679e-05,
+      "loss": 1.8246,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.302540415704388e-05,
+      "loss": 1.9016,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.325635103926097e-05,
+      "loss": 1.8471,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.348729792147806e-05,
+      "loss": 1.8209,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3718244803695156e-05,
+      "loss": 1.9718,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.394919168591224e-05,
+      "loss": 1.8385,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.418013856812933e-05,
+      "loss": 1.9649,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.441108545034642e-05,
+      "loss": 1.9567,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.464203233256351e-05,
+      "loss": 2.0799,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4872979214780605e-05,
+      "loss": 1.7452,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.510392609699769e-05,
+      "loss": 1.7384,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5334872979214786e-05,
+      "loss": 1.7826,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.556581986143187e-05,
+      "loss": 1.7008,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.579676674364896e-05,
+      "loss": 1.7251,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6027713625866054e-05,
+      "loss": 1.7472,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.625866050808314e-05,
+      "loss": 1.8222,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6489607390300235e-05,
+      "loss": 1.808,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.672055427251732e-05,
+      "loss": 1.8216,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6951501154734416e-05,
+      "loss": 1.718,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7182448036951504e-05,
+      "loss": 1.7371,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.741339491916859e-05,
+      "loss": 1.763,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7644341801385685e-05,
+      "loss": 1.8253,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.787528868360277e-05,
+      "loss": 1.658,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8106235565819866e-05,
+      "loss": 1.7982,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.833718244803695e-05,
+      "loss": 1.7464,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.856812933025405e-05,
+      "loss": 1.8344,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8799076212471134e-05,
+      "loss": 1.7035,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.903002309468822e-05,
+      "loss": 1.7534,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9260969976905315e-05,
+      "loss": 1.7304,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.94919168591224e-05,
+      "loss": 1.7613,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9722863741339496e-05,
+      "loss": 1.8009,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.995381062355658e-05,
+      "loss": 1.779,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.018475750577367e-05,
+      "loss": 1.6867,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0415704387990764e-05,
+      "loss": 1.7308,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.064665127020785e-05,
+      "loss": 1.8133,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0877598152424945e-05,
+      "loss": 1.6895,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.110854503464203e-05,
+      "loss": 1.7995,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1339491916859126e-05,
+      "loss": 1.7661,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1570438799076213e-05,
+      "loss": 1.6933,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.18013856812933e-05,
+      "loss": 1.7878,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2032332563510394e-05,
+      "loss": 1.7622,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.226327944572748e-05,
+      "loss": 1.8288,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2494226327944576e-05,
+      "loss": 1.6556,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.272517321016167e-05,
+      "loss": 1.6565,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2956120092378757e-05,
+      "loss": 1.7726,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3187066974595844e-05,
+      "loss": 1.6462,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.341801385681293e-05,
+      "loss": 1.6818,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3648960739030025e-05,
+      "loss": 1.696,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.387990762124711e-05,
+      "loss": 1.8292,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4110854503464206e-05,
+      "loss": 1.8906,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.43418013856813e-05,
+      "loss": 1.8841,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.457274826789839e-05,
+      "loss": 1.7492,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4803695150115474e-05,
+      "loss": 1.8522,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.503464203233256e-05,
+      "loss": 1.8968,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5265588914549655e-05,
+      "loss": 1.8029,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.549653579676675e-05,
+      "loss": 1.7943,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5727482678983836e-05,
+      "loss": 1.8515,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.595842956120093e-05,
+      "loss": 1.8426,
+      "step": 199
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.618937644341802e-05,
+      "loss": 1.8904,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6420323325635104e-05,
+      "loss": 1.659,
+      "step": 201
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.665127020785219e-05,
+      "loss": 1.7144,
+      "step": 202
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6882217090069285e-05,
+      "loss": 1.8116,
+      "step": 203
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.711316397228638e-05,
+      "loss": 1.5886,
+      "step": 204
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7344110854503466e-05,
+      "loss": 1.7973,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.757505773672056e-05,
+      "loss": 1.7768,
+      "step": 206
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.780600461893765e-05,
+      "loss": 1.7139,
+      "step": 207
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8036951501154735e-05,
+      "loss": 1.7802,
+      "step": 208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.826789838337183e-05,
+      "loss": 1.7805,
+      "step": 209
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8498845265588916e-05,
+      "loss": 1.7139,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.872979214780601e-05,
+      "loss": 1.7308,
+      "step": 211
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.89607390300231e-05,
+      "loss": 1.7976,
+      "step": 212
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.919168591224019e-05,
+      "loss": 1.7218,
+      "step": 213
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.942263279445728e-05,
+      "loss": 1.7034,
+      "step": 214
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9653579676674365e-05,
+      "loss": 1.7749,
+      "step": 215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.988452655889146e-05,
+      "loss": 1.7661,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.011547344110855e-05,
+      "loss": 1.6896,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.034642032332564e-05,
+      "loss": 1.8193,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.057736720554273e-05,
+      "loss": 1.8205,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.080831408775982e-05,
+      "loss": 1.763,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.103926096997691e-05,
+      "loss": 1.7446,
+      "step": 221
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.1270207852193995e-05,
+      "loss": 1.8132,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.150115473441108e-05,
+      "loss": 1.728,
+      "step": 223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.173210161662818e-05,
+      "loss": 1.6494,
+      "step": 224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.196304849884527e-05,
+      "loss": 1.7572,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.219399538106236e-05,
+      "loss": 1.6832,
+      "step": 226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.242494226327945e-05,
+      "loss": 1.6696,
+      "step": 227
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.265588914549654e-05,
+      "loss": 1.7514,
+      "step": 228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.2886836027713626e-05,
+      "loss": 1.6937,
+      "step": 229
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.311778290993071e-05,
+      "loss": 1.7217,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.334872979214781e-05,
+      "loss": 1.6909,
+      "step": 231
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.35796766743649e-05,
+      "loss": 1.8286,
+      "step": 232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.381062355658199e-05,
+      "loss": 1.8051,
+      "step": 233
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.404157043879908e-05,
+      "loss": 1.7914,
+      "step": 234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.427251732101617e-05,
+      "loss": 1.7863,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.4503464203233256e-05,
+      "loss": 1.7585,
+      "step": 236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.4734411085450356e-05,
+      "loss": 1.7482,
+      "step": 237
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.4965357967667444e-05,
+      "loss": 1.8119,
+      "step": 238
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.519630484988453e-05,
+      "loss": 1.7155,
+      "step": 239
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.542725173210162e-05,
+      "loss": 1.8023,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.565819861431871e-05,
+      "loss": 1.7472,
+      "step": 241
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.58891454965358e-05,
+      "loss": 1.805,
+      "step": 242
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.6120092378752886e-05,
+      "loss": 1.7592,
+      "step": 243
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.635103926096999e-05,
+      "loss": 1.7485,
+      "step": 244
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.6581986143187074e-05,
+      "loss": 1.7665,
+      "step": 245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.681293302540416e-05,
+      "loss": 1.842,
+      "step": 246
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.704387990762125e-05,
+      "loss": 1.9278,
+      "step": 247
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.727482678983834e-05,
+      "loss": 1.7335,
+      "step": 248
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.750577367205543e-05,
+      "loss": 1.803,
+      "step": 249
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.7736720554272516e-05,
+      "loss": 1.9681,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.796766743648962e-05,
+      "loss": 1.6879,
+      "step": 251
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.8198614318706704e-05,
+      "loss": 1.7048,
+      "step": 252
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.842956120092379e-05,
+      "loss": 1.7272,
+      "step": 253
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.866050808314087e-05,
+      "loss": 1.668,
+      "step": 254
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.889145496535797e-05,
+      "loss": 1.7625,
+      "step": 255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.912240184757506e-05,
+      "loss": 1.6881,
+      "step": 256
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.935334872979215e-05,
+      "loss": 1.7622,
+      "step": 257
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.958429561200925e-05,
+      "loss": 1.8298,
+      "step": 258
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9815242494226335e-05,
+      "loss": 1.713,
+      "step": 259
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.0046189376443415e-05,
+      "loss": 1.7587,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.0277136258660516e-05,
+      "loss": 1.7898,
+      "step": 261
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.05080831408776e-05,
+      "loss": 1.6853,
+      "step": 262
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.073903002309469e-05,
+      "loss": 1.8425,
+      "step": 263
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.096997690531178e-05,
+      "loss": 1.7564,
+      "step": 264
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.120092378752888e-05,
+      "loss": 1.7372,
+      "step": 265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.143187066974596e-05,
+      "loss": 1.6618,
+      "step": 266
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.166281755196305e-05,
+      "loss": 1.6271,
+      "step": 267
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.189376443418015e-05,
+      "loss": 1.818,
+      "step": 268
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.212471131639723e-05,
+      "loss": 1.7138,
+      "step": 269
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.235565819861431e-05,
+      "loss": 1.6889,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.25866050808314e-05,
+      "loss": 1.7502,
+      "step": 271
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.28175519630485e-05,
+      "loss": 1.7369,
+      "step": 272
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.304849884526559e-05,
+      "loss": 1.7414,
+      "step": 273
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.327944572748268e-05,
+      "loss": 1.7803,
+      "step": 274
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.351039260969978e-05,
+      "loss": 1.7649,
+      "step": 275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.374133949191686e-05,
+      "loss": 1.6532,
+      "step": 276
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.397228637413395e-05,
+      "loss": 1.6783,
+      "step": 277
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.420323325635105e-05,
+      "loss": 1.7418,
+      "step": 278
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.443418013856814e-05,
+      "loss": 1.7078,
+      "step": 279
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.466512702078523e-05,
+      "loss": 1.593,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.489607390300231e-05,
+      "loss": 1.6632,
+      "step": 281
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.51270207852194e-05,
+      "loss": 1.7226,
+      "step": 282
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.535796766743649e-05,
+      "loss": 1.6491,
+      "step": 283
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.558891454965357e-05,
+      "loss": 1.8326,
+      "step": 284
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.581986143187067e-05,
+      "loss": 1.744,
+      "step": 285
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.605080831408776e-05,
+      "loss": 1.6603,
+      "step": 286
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.628175519630485e-05,
+      "loss": 1.7561,
+      "step": 287
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.651270207852194e-05,
+      "loss": 1.6338,
+      "step": 288
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.674364896073904e-05,
+      "loss": 1.7346,
+      "step": 289
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.697459584295612e-05,
+      "loss": 1.7542,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.720554272517321e-05,
+      "loss": 1.9143,
+      "step": 291
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.743648960739031e-05,
+      "loss": 1.7637,
+      "step": 292
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.76674364896074e-05,
+      "loss": 1.7614,
+      "step": 293
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.789838337182449e-05,
+      "loss": 1.7997,
+      "step": 294
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.812933025404157e-05,
+      "loss": 1.6856,
+      "step": 295
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.836027713625866e-05,
+      "loss": 1.6449,
+      "step": 296
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.859122401847575e-05,
+      "loss": 1.7912,
+      "step": 297
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.882217090069283e-05,
+      "loss": 1.8067,
+      "step": 298
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.905311778290994e-05,
+      "loss": 1.8372,
+      "step": 299
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.928406466512702e-05,
+      "loss": 1.8121,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.951501154734411e-05,
+      "loss": 1.6257,
+      "step": 301
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.974595842956121e-05,
+      "loss": 1.6421,
+      "step": 302
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.99769053117783e-05,
+      "loss": 1.7146,
+      "step": 303
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.020785219399538e-05,
+      "loss": 1.7722,
+      "step": 304
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.043879907621247e-05,
+      "loss": 1.7451,
+      "step": 305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.066974595842957e-05,
+      "loss": 1.6878,
+      "step": 306
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.090069284064666e-05,
+      "loss": 1.7257,
+      "step": 307
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.113163972286375e-05,
+      "loss": 1.7836,
+      "step": 308
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.136258660508083e-05,
+      "loss": 1.7758,
+      "step": 309
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.159353348729792e-05,
+      "loss": 1.6699,
+      "step": 310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.182448036951501e-05,
+      "loss": 1.7711,
+      "step": 311
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.205542725173211e-05,
+      "loss": 1.8168,
+      "step": 312
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.22863741339492e-05,
+      "loss": 1.7687,
+      "step": 313
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.251732101616628e-05,
+      "loss": 1.7733,
+      "step": 314
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.274826789838337e-05,
+      "loss": 1.7339,
+      "step": 315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.297921478060047e-05,
+      "loss": 1.6978,
+      "step": 316
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.321016166281756e-05,
+      "loss": 1.7598,
+      "step": 317
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.344110854503465e-05,
+      "loss": 1.8048,
+      "step": 318
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.367205542725175e-05,
+      "loss": 1.701,
+      "step": 319
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.390300230946883e-05,
+      "loss": 1.7014,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.413394919168592e-05,
+      "loss": 1.6823,
+      "step": 321
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.436489607390301e-05,
+      "loss": 1.7688,
+      "step": 322
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.45958429561201e-05,
+      "loss": 1.7721,
+      "step": 323
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.482678983833718e-05,
+      "loss": 1.6708,
+      "step": 324
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.505773672055427e-05,
+      "loss": 1.58,
+      "step": 325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.528868360277137e-05,
+      "loss": 1.7666,
+      "step": 326
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.551963048498846e-05,
+      "loss": 1.7138,
+      "step": 327
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.575057736720554e-05,
+      "loss": 1.6147,
+      "step": 328
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.598152424942263e-05,
+      "loss": 1.6749,
+      "step": 329
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.621247113163973e-05,
+      "loss": 1.7002,
+      "step": 330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.644341801385682e-05,
+      "loss": 1.7286,
+      "step": 331
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.66743648960739e-05,
+      "loss": 1.6233,
+      "step": 332
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.6905311778291e-05,
+      "loss": 1.7853,
+      "step": 333
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.71362586605081e-05,
+      "loss": 1.841,
+      "step": 334
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.736720554272518e-05,
+      "loss": 1.6616,
+      "step": 335
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.759815242494227e-05,
+      "loss": 1.7413,
+      "step": 336
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.782909930715935e-05,
+      "loss": 1.6992,
+      "step": 337
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.806004618937644e-05,
+      "loss": 1.7349,
+      "step": 338
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.829099307159353e-05,
+      "loss": 1.6088,
+      "step": 339
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.852193995381063e-05,
+      "loss": 1.8487,
+      "step": 340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.875288683602772e-05,
+      "loss": 1.7514,
+      "step": 341
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.89838337182448e-05,
+      "loss": 1.734,
+      "step": 342
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.92147806004619e-05,
+      "loss": 1.7416,
+      "step": 343
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.944572748267899e-05,
+      "loss": 1.7704,
+      "step": 344
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.967667436489608e-05,
+      "loss": 1.7362,
+      "step": 345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.990762124711317e-05,
+      "loss": 1.7067,
+      "step": 346
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.013856812933027e-05,
+      "loss": 1.8517,
+      "step": 347
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.036951501154734e-05,
+      "loss": 1.8275,
+      "step": 348
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.060046189376443e-05,
+      "loss": 1.7033,
+      "step": 349
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.083140877598153e-05,
+      "loss": 1.86,
+      "step": 350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.106235565819862e-05,
+      "loss": 1.785,
+      "step": 351
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.12933025404157e-05,
+      "loss": 1.6494,
+      "step": 352
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.15242494226328e-05,
+      "loss": 1.6693,
+      "step": 353
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.175519630484989e-05,
+      "loss": 1.7269,
+      "step": 354
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.198614318706698e-05,
+      "loss": 1.6691,
+      "step": 355
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.221709006928406e-05,
+      "loss": 1.6573,
+      "step": 356
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.244803695150117e-05,
+      "loss": 1.7651,
+      "step": 357
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.267898383371825e-05,
+      "loss": 1.8549,
+      "step": 358
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.290993071593534e-05,
+      "loss": 1.6775,
+      "step": 359
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.314087759815243e-05,
+      "loss": 1.8279,
+      "step": 360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.337182448036951e-05,
+      "loss": 1.6998,
+      "step": 361
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.36027713625866e-05,
+      "loss": 1.7014,
+      "step": 362
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.383371824480369e-05,
+      "loss": 1.8117,
+      "step": 363
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.406466512702079e-05,
+      "loss": 1.6871,
+      "step": 364
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.429561200923788e-05,
+      "loss": 1.5849,
+      "step": 365
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.452655889145496e-05,
+      "loss": 1.7148,
+      "step": 366
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.475750577367206e-05,
+      "loss": 1.7061,
+      "step": 367
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.498845265588915e-05,
+      "loss": 1.6919,
+      "step": 368
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.521939953810624e-05,
+      "loss": 1.7411,
+      "step": 369
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.545034642032334e-05,
+      "loss": 1.6994,
+      "step": 370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.568129330254043e-05,
+      "loss": 1.7334,
+      "step": 371
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.591224018475751e-05,
+      "loss": 1.7558,
+      "step": 372
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.61431870669746e-05,
+      "loss": 1.6788,
+      "step": 373
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.637413394919169e-05,
+      "loss": 1.5368,
+      "step": 374
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.660508083140877e-05,
+      "loss": 1.6376,
+      "step": 375
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.683602771362586e-05,
+      "loss": 1.6468,
+      "step": 376
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.706697459584296e-05,
+      "loss": 1.8013,
+      "step": 377
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.729792147806005e-05,
+      "loss": 1.7509,
+      "step": 378
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.752886836027714e-05,
+      "loss": 1.6814,
+      "step": 379
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.775981524249422e-05,
+      "loss": 1.683,
+      "step": 380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.799076212471132e-05,
+      "loss": 1.6735,
+      "step": 381
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.822170900692841e-05,
+      "loss": 1.6959,
+      "step": 382
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.84526558891455e-05,
+      "loss": 1.6169,
+      "step": 383
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.86836027713626e-05,
+      "loss": 1.651,
+      "step": 384
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.891454965357969e-05,
+      "loss": 1.8153,
+      "step": 385
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.914549653579677e-05,
+      "loss": 1.6773,
+      "step": 386
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.937644341801386e-05,
+      "loss": 1.6487,
+      "step": 387
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.960739030023095e-05,
+      "loss": 1.6444,
+      "step": 388
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.983833718244804e-05,
+      "loss": 1.8085,
+      "step": 389
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.006928406466512e-05,
+      "loss": 1.6826,
+      "step": 390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.030023094688222e-05,
+      "loss": 1.7592,
+      "step": 391
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.053117782909931e-05,
+      "loss": 1.8147,
+      "step": 392
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.07621247113164e-05,
+      "loss": 1.6817,
+      "step": 393
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.09930715935335e-05,
+      "loss": 1.6965,
+      "step": 394
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.122401847575059e-05,
+      "loss": 1.6711,
+      "step": 395
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.145496535796767e-05,
+      "loss": 1.6746,
+      "step": 396
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.168591224018476e-05,
+      "loss": 1.8663,
+      "step": 397
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.191685912240186e-05,
+      "loss": 1.7098,
+      "step": 398
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.214780600461895e-05,
+      "loss": 1.8029,
+      "step": 399
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.237875288683603e-05,
+      "loss": 1.9676,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.260969976905312e-05,
+      "loss": 1.6455,
+      "step": 401
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.284064665127021e-05,
+      "loss": 1.7672,
+      "step": 402
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.30715935334873e-05,
+      "loss": 1.7412,
+      "step": 403
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.330254041570438e-05,
+      "loss": 1.7341,
+      "step": 404
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.353348729792148e-05,
+      "loss": 1.7876,
+      "step": 405
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.376443418013857e-05,
+      "loss": 1.6994,
+      "step": 406
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.399538106235566e-05,
+      "loss": 1.7065,
+      "step": 407
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.422632794457276e-05,
+      "loss": 1.7601,
+      "step": 408
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.445727482678985e-05,
+      "loss": 1.8049,
+      "step": 409
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.468822170900693e-05,
+      "loss": 1.5474,
+      "step": 410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.491916859122403e-05,
+      "loss": 1.7281,
+      "step": 411
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.515011547344112e-05,
+      "loss": 1.8323,
+      "step": 412
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.538106235565821e-05,
+      "loss": 1.6463,
+      "step": 413
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.56120092378753e-05,
+      "loss": 1.6124,
+      "step": 414
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.584295612009238e-05,
+      "loss": 1.7445,
+      "step": 415
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.607390300230947e-05,
+      "loss": 1.6723,
+      "step": 416
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.630484988452656e-05,
+      "loss": 1.6537,
+      "step": 417
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.653579676674366e-05,
+      "loss": 1.6003,
+      "step": 418
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.676674364896074e-05,
+      "loss": 1.824,
+      "step": 419
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.699769053117783e-05,
+      "loss": 1.7469,
+      "step": 420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.722863741339492e-05,
+      "loss": 1.7074,
+      "step": 421
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.745958429561202e-05,
+      "loss": 1.7642,
+      "step": 422
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.76905311778291e-05,
+      "loss": 1.7154,
+      "step": 423
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.79214780600462e-05,
+      "loss": 1.679,
+      "step": 424
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.81524249422633e-05,
+      "loss": 1.682,
+      "step": 425
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.838337182448038e-05,
+      "loss": 1.7589,
+      "step": 426
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.861431870669747e-05,
+      "loss": 1.7403,
+      "step": 427
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.884526558891456e-05,
+      "loss": 1.7977,
+      "step": 428
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.907621247113164e-05,
+      "loss": 1.5648,
+      "step": 429
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.930715935334873e-05,
+      "loss": 1.8315,
+      "step": 430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.953810623556582e-05,
+      "loss": 1.5583,
+      "step": 431
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.976905311778292e-05,
+      "loss": 1.741,
+      "step": 432
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001,
+      "loss": 1.6083,
+      "step": 433
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999999634117307e-05,
+      "loss": 1.4568,
+      "step": 434
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999998536469281e-05,
+      "loss": 1.5528,
+      "step": 435
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999996707056081e-05,
+      "loss": 1.6205,
+      "step": 436
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999994145877978e-05,
+      "loss": 1.7533,
+      "step": 437
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999990852935344e-05,
+      "loss": 1.7803,
+      "step": 438
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999986828228662e-05,
+      "loss": 1.7526,
+      "step": 439
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999982071758523e-05,
+      "loss": 1.7433,
+      "step": 440
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999976583525619e-05,
+      "loss": 1.8143,
+      "step": 441
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999970363530755e-05,
+      "loss": 1.8028,
+      "step": 442
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999963411774843e-05,
+      "loss": 1.6967,
+      "step": 443
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999955728258899e-05,
+      "loss": 1.5162,
+      "step": 444
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999947312984047e-05,
+      "loss": 1.7631,
+      "step": 445
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999938165951519e-05,
+      "loss": 1.7156,
+      "step": 446
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999928287162654e-05,
+      "loss": 1.7056,
+      "step": 447
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9999176766189e-05,
+      "loss": 1.8366,
+      "step": 448
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999906334321804e-05,
+      "loss": 1.8317,
+      "step": 449
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999894260273033e-05,
+      "loss": 1.6757,
+      "step": 450
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999881454474348e-05,
+      "loss": 1.7926,
+      "step": 451
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999867916927627e-05,
+      "loss": 1.6726,
+      "step": 452
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99985364763485e-05,
+      "loss": 1.6581,
+      "step": 453
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999838646598106e-05,
+      "loss": 1.8693,
+      "step": 454
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999822913819587e-05,
+      "loss": 1.6388,
+      "step": 455
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9998064493016e-05,
+      "loss": 1.6629,
+      "step": 456
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999789253046555e-05,
+      "loss": 1.6222,
+      "step": 457
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999771325056966e-05,
+      "loss": 1.7521,
+      "step": 458
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999752665335455e-05,
+      "loss": 1.6815,
+      "step": 459
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99973327388476e-05,
+      "loss": 1.6851,
+      "step": 460
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99971315070771e-05,
+      "loss": 1.7556,
+      "step": 461
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999692295807257e-05,
+      "loss": 1.7038,
+      "step": 462
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999670709186451e-05,
+      "loss": 1.6948,
+      "step": 463
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99964839084845e-05,
+      "loss": 1.6525,
+      "step": 464
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999625340796521e-05,
+      "loss": 1.6326,
+      "step": 465
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99960155903404e-05,
+      "loss": 1.7389,
+      "step": 466
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999577045564483e-05,
+      "loss": 1.7795,
+      "step": 467
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99955180039144e-05,
+      "loss": 1.6156,
+      "step": 468
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999525823518608e-05,
+      "loss": 1.7019,
+      "step": 469
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999499114949783e-05,
+      "loss": 1.6759,
+      "step": 470
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99947167468888e-05,
+      "loss": 1.672,
+      "step": 471
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999443502739912e-05,
+      "loss": 1.6926,
+      "step": 472
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999414599107001e-05,
+      "loss": 1.6887,
+      "step": 473
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999384963794381e-05,
+      "loss": 1.7496,
+      "step": 474
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999354596806385e-05,
+      "loss": 1.6766,
+      "step": 475
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99932349814746e-05,
+      "loss": 1.649,
+      "step": 476
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999291667822157e-05,
+      "loss": 1.736,
+      "step": 477
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999259105835132e-05,
+      "loss": 1.7733,
+      "step": 478
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999225812191153e-05,
+      "loss": 1.675,
+      "step": 479
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999191786895093e-05,
+      "loss": 1.7006,
+      "step": 480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99915702995193e-05,
+      "loss": 1.5517,
+      "step": 481
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999121541366752e-05,
+      "loss": 1.8047,
+      "step": 482
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999085321144752e-05,
+      "loss": 1.5923,
+      "step": 483
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999048369291231e-05,
+      "loss": 1.5327,
+      "step": 484
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999010685811598e-05,
+      "loss": 1.5378,
+      "step": 485
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.998972270711367e-05,
+      "loss": 1.5273,
+      "step": 486
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.998933123996161e-05,
+      "loss": 1.7053,
+      "step": 487
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99889324567171e-05,
+      "loss": 1.8069,
+      "step": 488
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.998852635743848e-05,
+      "loss": 1.6969,
+      "step": 489
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.998811294218519e-05,
+      "loss": 1.7853,
+      "step": 490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.998769221101776e-05,
+      "loss": 1.6408,
+      "step": 491
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.998726416399773e-05,
+      "loss": 1.8463,
+      "step": 492
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.998682880118776e-05,
+      "loss": 1.6882,
+      "step": 493
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.998638612265158e-05,
+      "loss": 1.7104,
+      "step": 494
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.998593612845396e-05,
+      "loss": 1.6034,
+      "step": 495
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.998547881866076e-05,
+      "loss": 1.7127,
+      "step": 496
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.998501419333893e-05,
+      "loss": 1.7397,
+      "step": 497
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.998454225255643e-05,
+      "loss": 1.9687,
+      "step": 498
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.998406299638236e-05,
+      "loss": 1.8718,
+      "step": 499
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.998357642488685e-05,
+      "loss": 1.8287,
+      "step": 500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.998308253814111e-05,
+      "loss": 1.7594,
+      "step": 501
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.998258133621744e-05,
+      "loss": 1.6541,
+      "step": 502
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.998207281918916e-05,
+      "loss": 1.7286,
+      "step": 503
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.998155698713074e-05,
+      "loss": 1.596,
+      "step": 504
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.998103384011762e-05,
+      "loss": 1.5839,
+      "step": 505
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.998050337822639e-05,
+      "loss": 1.6686,
+      "step": 506
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.997996560153468e-05,
+      "loss": 1.7363,
+      "step": 507
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.997942051012122e-05,
+      "loss": 1.8112,
+      "step": 508
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.997886810406576e-05,
+      "loss": 1.6508,
+      "step": 509
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.997830838344913e-05,
+      "loss": 1.8195,
+      "step": 510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.997774134835329e-05,
+      "loss": 1.6837,
+      "step": 511
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99771669988612e-05,
+      "loss": 1.6527,
+      "step": 512
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.997658533505692e-05,
+      "loss": 1.6977,
+      "step": 513
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.997599635702558e-05,
+      "loss": 1.7285,
+      "step": 514
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.997540006485337e-05,
+      "loss": 1.6837,
+      "step": 515
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99747964586276e-05,
+      "loss": 1.7058,
+      "step": 516
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.997418553843655e-05,
+      "loss": 1.6407,
+      "step": 517
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.997356730436966e-05,
+      "loss": 1.7551,
+      "step": 518
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.997294175651743e-05,
+      "loss": 1.6461,
+      "step": 519
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.997230889497138e-05,
+      "loss": 1.7397,
+      "step": 520
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.997166871982413e-05,
+      "loss": 1.711,
+      "step": 521
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.997102123116939e-05,
+      "loss": 1.622,
+      "step": 522
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.997036642910193e-05,
+      "loss": 1.7626,
+      "step": 523
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.996970431371755e-05,
+      "loss": 1.6134,
+      "step": 524
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.996903488511318e-05,
+      "loss": 1.6663,
+      "step": 525
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.996835814338678e-05,
+      "loss": 1.6415,
+      "step": 526
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.996767408863738e-05,
+      "loss": 1.7471,
+      "step": 527
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.996698272096513e-05,
+      "loss": 1.5339,
+      "step": 528
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99662840404712e-05,
+      "loss": 1.7099,
+      "step": 529
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.996557804725781e-05,
+      "loss": 1.4499,
+      "step": 530
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.996486474142833e-05,
+      "loss": 1.5677,
+      "step": 531
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.996414412308713e-05,
+      "loss": 1.5127,
+      "step": 532
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.996341619233968e-05,
+      "loss": 1.5516,
+      "step": 533
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.996268094929253e-05,
+      "loss": 1.629,
+      "step": 534
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.996193839405326e-05,
+      "loss": 1.4941,
+      "step": 535
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.996118852673055e-05,
+      "loss": 1.4412,
+      "step": 536
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.996043134743417e-05,
+      "loss": 1.5978,
+      "step": 537
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.995966685627491e-05,
+      "loss": 1.712,
+      "step": 538
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.995889505336466e-05,
+      "loss": 1.5433,
+      "step": 539
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.995811593881636e-05,
+      "loss": 1.718,
+      "step": 540
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.995732951274408e-05,
+      "loss": 1.709,
+      "step": 541
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.995653577526288e-05,
+      "loss": 1.7087,
+      "step": 542
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.995573472648893e-05,
+      "loss": 1.6869,
+      "step": 543
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.995492636653949e-05,
+      "loss": 1.6937,
+      "step": 544
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.995411069553284e-05,
+      "loss": 1.6139,
+      "step": 545
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.995328771358836e-05,
+      "loss": 1.7852,
+      "step": 546
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.995245742082651e-05,
+      "loss": 1.655,
+      "step": 547
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.995161981736878e-05,
+      "loss": 1.7025,
+      "step": 548
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.995077490333778e-05,
+      "loss": 1.7687,
+      "step": 549
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.994992267885716e-05,
+      "loss": 1.675,
+      "step": 550
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.994906314405163e-05,
+      "loss": 1.7029,
+      "step": 551
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.994819629904702e-05,
+      "loss": 1.6822,
+      "step": 552
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.994732214397016e-05,
+      "loss": 1.7043,
+      "step": 553
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.9946440678949e-05,
+      "loss": 1.6889,
+      "step": 554
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.994555190411255e-05,
+      "loss": 1.8042,
+      "step": 555
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.994465581959087e-05,
+      "loss": 1.7054,
+      "step": 556
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.994375242551512e-05,
+      "loss": 1.6955,
+      "step": 557
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.994284172201751e-05,
+      "loss": 1.686,
+      "step": 558
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.994192370923132e-05,
+      "loss": 1.6027,
+      "step": 559
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.994099838729091e-05,
+      "loss": 1.6767,
+      "step": 560
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99400657563317e-05,
+      "loss": 1.6605,
+      "step": 561
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.993912581649018e-05,
+      "loss": 1.5944,
+      "step": 562
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.993817856790393e-05,
+      "loss": 1.6722,
+      "step": 563
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.993722401071155e-05,
+      "loss": 1.6135,
+      "step": 564
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.993626214505279e-05,
+      "loss": 1.697,
+      "step": 565
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.993529297106836e-05,
+      "loss": 1.5683,
+      "step": 566
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.993431648890016e-05,
+      "loss": 1.6133,
+      "step": 567
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.993333269869107e-05,
+      "loss": 1.6507,
+      "step": 568
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.993234160058508e-05,
+      "loss": 1.6447,
+      "step": 569
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.993134319472724e-05,
+      "loss": 1.5835,
+      "step": 570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.993033748126365e-05,
+      "loss": 1.7586,
+      "step": 571
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.992932446034153e-05,
+      "loss": 1.6713,
+      "step": 572
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.992830413210914e-05,
+      "loss": 1.7306,
+      "step": 573
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.992727649671578e-05,
+      "loss": 1.7494,
+      "step": 574
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.992624155431185e-05,
+      "loss": 1.7282,
+      "step": 575
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.992519930504884e-05,
+      "loss": 1.6271,
+      "step": 576
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99241497490793e-05,
+      "loss": 1.6177,
+      "step": 577
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.992309288655676e-05,
+      "loss": 1.6843,
+      "step": 578
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.992202871763599e-05,
+      "loss": 1.7529,
+      "step": 579
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.992095724247266e-05,
+      "loss": 1.4018,
+      "step": 580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.991987846122362e-05,
+      "loss": 1.6443,
+      "step": 581
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.991879237404675e-05,
+      "loss": 1.5648,
+      "step": 582
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.991769898110101e-05,
+      "loss": 1.5628,
+      "step": 583
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99165982825464e-05,
+      "loss": 1.6047,
+      "step": 584
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.991549027854404e-05,
+      "loss": 1.6917,
+      "step": 585
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.991437496925604e-05,
+      "loss": 1.5437,
+      "step": 586
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.991325235484568e-05,
+      "loss": 1.5215,
+      "step": 587
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.991212243547723e-05,
+      "loss": 1.7474,
+      "step": 588
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.991098521131606e-05,
+      "loss": 1.6638,
+      "step": 589
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.990984068252862e-05,
+      "loss": 1.7324,
+      "step": 590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.990868884928242e-05,
+      "loss": 1.8047,
+      "step": 591
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9907529711746e-05,
+      "loss": 1.7238,
+      "step": 592
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.990636327008902e-05,
+      "loss": 1.6249,
+      "step": 593
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.990518952448221e-05,
+      "loss": 1.6595,
+      "step": 594
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.990400847509733e-05,
+      "loss": 1.6668,
+      "step": 595
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.990282012210725e-05,
+      "loss": 1.5852,
+      "step": 596
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.990162446568587e-05,
+      "loss": 1.772,
+      "step": 597
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.990042150600817e-05,
+      "loss": 1.6899,
+      "step": 598
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.989921124325026e-05,
+      "loss": 1.7686,
+      "step": 599
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.989799367758921e-05,
+      "loss": 1.8567,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.989676880920322e-05,
+      "loss": 1.6866,
+      "step": 601
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.989553663827157e-05,
+      "loss": 1.6774,
+      "step": 602
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.98942971649746e-05,
+      "loss": 1.713,
+      "step": 603
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.989305038949368e-05,
+      "loss": 1.7198,
+      "step": 604
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.989179631201131e-05,
+      "loss": 1.6643,
+      "step": 605
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.989053493271101e-05,
+      "loss": 1.6197,
+      "step": 606
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.988926625177739e-05,
+      "loss": 1.7276,
+      "step": 607
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.988799026939613e-05,
+      "loss": 1.722,
+      "step": 608
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.988670698575397e-05,
+      "loss": 1.6293,
+      "step": 609
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.988541640103874e-05,
+      "loss": 1.7995,
+      "step": 610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.988411851543929e-05,
+      "loss": 1.5932,
+      "step": 611
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.98828133291456e-05,
+      "loss": 1.6573,
+      "step": 612
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.988150084234866e-05,
+      "loss": 1.6455,
+      "step": 613
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.988018105524057e-05,
+      "loss": 1.7073,
+      "step": 614
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.987885396801449e-05,
+      "loss": 1.7621,
+      "step": 615
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.987751958086463e-05,
+      "loss": 1.6232,
+      "step": 616
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.987617789398629e-05,
+      "loss": 1.6043,
+      "step": 617
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.987482890757584e-05,
+      "loss": 1.6486,
+      "step": 618
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.98734726218307e-05,
+      "loss": 1.6044,
+      "step": 619
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.987210903694933e-05,
+      "loss": 1.7044,
+      "step": 620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.987073815313136e-05,
+      "loss": 1.6631,
+      "step": 621
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.986935997057738e-05,
+      "loss": 1.6302,
+      "step": 622
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.986797448948911e-05,
+      "loss": 1.6848,
+      "step": 623
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.986658171006932e-05,
+      "loss": 1.76,
+      "step": 624
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.986518163252183e-05,
+      "loss": 1.7416,
+      "step": 625
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.986377425705155e-05,
+      "loss": 1.7527,
+      "step": 626
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.986235958386448e-05,
+      "loss": 1.631,
+      "step": 627
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.986093761316762e-05,
+      "loss": 1.7874,
+      "step": 628
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.985950834516911e-05,
+      "loss": 1.293,
+      "step": 629
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.985807178007812e-05,
+      "loss": 1.5749,
+      "step": 630
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.98566279181049e-05,
+      "loss": 1.7587,
+      "step": 631
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.985517675946075e-05,
+      "loss": 1.7806,
+      "step": 632
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.985371830435806e-05,
+      "loss": 1.7243,
+      "step": 633
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.985225255301029e-05,
+      "loss": 1.604,
+      "step": 634
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.985077950563193e-05,
+      "loss": 1.5491,
+      "step": 635
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.98492991624386e-05,
+      "loss": 1.5912,
+      "step": 636
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.984781152364693e-05,
+      "loss": 1.7186,
+      "step": 637
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.984631658947463e-05,
+      "loss": 1.5918,
+      "step": 638
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.984481436014052e-05,
+      "loss": 1.6988,
+      "step": 639
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.984330483586442e-05,
+      "loss": 1.5665,
+      "step": 640
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.984178801686729e-05,
+      "loss": 1.6682,
+      "step": 641
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.984026390337109e-05,
+      "loss": 1.6753,
+      "step": 642
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.983873249559891e-05,
+      "loss": 1.6103,
+      "step": 643
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.983719379377485e-05,
+      "loss": 1.6239,
+      "step": 644
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.983564779812411e-05,
+      "loss": 1.708,
+      "step": 645
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.983409450887296e-05,
+      "loss": 1.635,
+      "step": 646
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.983253392624872e-05,
+      "loss": 1.6234,
+      "step": 647
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.983096605047978e-05,
+      "loss": 1.6666,
+      "step": 648
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.982939088179561e-05,
+      "loss": 1.6849,
+      "step": 649
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.982780842042676e-05,
+      "loss": 1.6762,
+      "step": 650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98262186666048e-05,
+      "loss": 1.7721,
+      "step": 651
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.982462162056241e-05,
+      "loss": 1.7067,
+      "step": 652
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.982301728253331e-05,
+      "loss": 1.5818,
+      "step": 653
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98214056527523e-05,
+      "loss": 1.5865,
+      "step": 654
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.981978673145528e-05,
+      "loss": 1.6983,
+      "step": 655
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.981816051887914e-05,
+      "loss": 1.7546,
+      "step": 656
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.981652701526192e-05,
+      "loss": 1.5319,
+      "step": 657
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.981488622084265e-05,
+      "loss": 1.6988,
+      "step": 658
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98132381358615e-05,
+      "loss": 1.7877,
+      "step": 659
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.981158276055964e-05,
+      "loss": 1.6788,
+      "step": 660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.980992009517937e-05,
+      "loss": 1.7596,
+      "step": 661
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.980825013996402e-05,
+      "loss": 1.6508,
+      "step": 662
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.980657289515796e-05,
+      "loss": 1.7169,
+      "step": 663
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98048883610067e-05,
+      "loss": 1.6611,
+      "step": 664
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.980319653775677e-05,
+      "loss": 1.6162,
+      "step": 665
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.980149742565575e-05,
+      "loss": 1.6943,
+      "step": 666
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.979979102495231e-05,
+      "loss": 1.6026,
+      "step": 667
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.979807733589623e-05,
+      "loss": 1.737,
+      "step": 668
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.979635635873829e-05,
+      "loss": 1.6476,
+      "step": 669
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.979462809373035e-05,
+      "loss": 1.6968,
+      "step": 670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.979289254112534e-05,
+      "loss": 1.7627,
+      "step": 671
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.979114970117728e-05,
+      "loss": 1.6058,
+      "step": 672
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.978939957414125e-05,
+      "loss": 1.738,
+      "step": 673
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.978764216027336e-05,
+      "loss": 1.6799,
+      "step": 674
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.978587745983084e-05,
+      "loss": 1.6146,
+      "step": 675
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.978410547307193e-05,
+      "loss": 1.6552,
+      "step": 676
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.9782326200256e-05,
+      "loss": 1.6824,
+      "step": 677
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.978053964164342e-05,
+      "loss": 1.683,
+      "step": 678
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.977874579749568e-05,
+      "loss": 1.7087,
+      "step": 679
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.97769446680753e-05,
+      "loss": 1.4994,
+      "step": 680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.97751362536459e-05,
+      "loss": 1.7006,
+      "step": 681
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.977332055447212e-05,
+      "loss": 1.725,
+      "step": 682
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.97714975708197e-05,
+      "loss": 1.7816,
+      "step": 683
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.976966730295546e-05,
+      "loss": 1.5774,
+      "step": 684
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.976782975114725e-05,
+      "loss": 1.6021,
+      "step": 685
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.9765984915664e-05,
+      "loss": 1.5923,
+      "step": 686
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.976413279677572e-05,
+      "loss": 1.6613,
+      "step": 687
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.976227339475346e-05,
+      "loss": 1.6915,
+      "step": 688
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.976040670986935e-05,
+      "loss": 1.6586,
+      "step": 689
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.975853274239659e-05,
+      "loss": 1.5673,
+      "step": 690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.975665149260944e-05,
+      "loss": 1.6825,
+      "step": 691
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.975476296078324e-05,
+      "loss": 1.7254,
+      "step": 692
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.975286714719434e-05,
+      "loss": 1.6855,
+      "step": 693
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.975096405212025e-05,
+      "loss": 1.6873,
+      "step": 694
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.974905367583946e-05,
+      "loss": 1.5699,
+      "step": 695
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.974713601863157e-05,
+      "loss": 1.7141,
+      "step": 696
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.974521108077723e-05,
+      "loss": 1.7238,
+      "step": 697
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.974327886255816e-05,
+      "loss": 1.6936,
+      "step": 698
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.974133936425717e-05,
+      "loss": 1.5895,
+      "step": 699
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.97393925861581e-05,
+      "loss": 1.678,
+      "step": 700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.973743852854583e-05,
+      "loss": 1.6346,
+      "step": 701
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.973547719170639e-05,
+      "loss": 1.6816,
+      "step": 702
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.973350857592682e-05,
+      "loss": 1.6705,
+      "step": 703
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.97315326814952e-05,
+      "loss": 1.65,
+      "step": 704
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.972954950870074e-05,
+      "loss": 1.6791,
+      "step": 705
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.972755905783368e-05,
+      "loss": 1.6795,
+      "step": 706
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.972556132918532e-05,
+      "loss": 1.748,
+      "step": 707
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.972355632304803e-05,
+      "loss": 1.6527,
+      "step": 708
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.972154403971527e-05,
+      "loss": 1.7168,
+      "step": 709
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.971952447948153e-05,
+      "loss": 1.6781,
+      "step": 710
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.971749764264238e-05,
+      "loss": 1.6769,
+      "step": 711
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.971546352949444e-05,
+      "loss": 1.6027,
+      "step": 712
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.971342214033542e-05,
+      "loss": 1.7586,
+      "step": 713
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.971137347546409e-05,
+      "loss": 1.5895,
+      "step": 714
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.970931753518027e-05,
+      "loss": 1.6647,
+      "step": 715
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.970725431978488e-05,
+      "loss": 1.6961,
+      "step": 716
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.970518382957984e-05,
+      "loss": 1.6574,
+      "step": 717
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.970310606486817e-05,
+      "loss": 1.7379,
+      "step": 718
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.970102102595398e-05,
+      "loss": 1.5466,
+      "step": 719
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.969892871314242e-05,
+      "loss": 1.675,
+      "step": 720
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.96968291267397e-05,
+      "loss": 1.5696,
+      "step": 721
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.96947222670531e-05,
+      "loss": 1.6681,
+      "step": 722
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.969260813439098e-05,
+      "loss": 1.8135,
+      "step": 723
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.969048672906274e-05,
+      "loss": 1.6739,
+      "step": 724
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.968835805137884e-05,
+      "loss": 1.5291,
+      "step": 725
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.968622210165084e-05,
+      "loss": 1.5936,
+      "step": 726
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.968407888019134e-05,
+      "loss": 1.6107,
+      "step": 727
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.968192838731399e-05,
+      "loss": 1.7133,
+      "step": 728
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.967977062333353e-05,
+      "loss": 1.7448,
+      "step": 729
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.967760558856577e-05,
+      "loss": 1.7029,
+      "step": 730
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.967543328332755e-05,
+      "loss": 1.5677,
+      "step": 731
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.96732537079368e-05,
+      "loss": 1.6727,
+      "step": 732
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.96710668627125e-05,
+      "loss": 1.5228,
+      "step": 733
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.966887274797473e-05,
+      "loss": 1.5142,
+      "step": 734
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.966667136404458e-05,
+      "loss": 1.5364,
+      "step": 735
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.966446271124424e-05,
+      "loss": 1.4629,
+      "step": 736
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.966224678989693e-05,
+      "loss": 1.6864,
+      "step": 737
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.966002360032698e-05,
+      "loss": 1.5376,
+      "step": 738
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.965779314285975e-05,
+      "loss": 1.6366,
+      "step": 739
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.96555554178217e-05,
+      "loss": 1.6577,
+      "step": 740
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.96533104255403e-05,
+      "loss": 1.6743,
+      "step": 741
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.965105816634411e-05,
+      "loss": 1.6709,
+      "step": 742
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.964879864056276e-05,
+      "loss": 1.6781,
+      "step": 743
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.964653184852694e-05,
+      "loss": 1.7374,
+      "step": 744
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.964425779056843e-05,
+      "loss": 1.6789,
+      "step": 745
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.964197646702e-05,
+      "loss": 1.5924,
+      "step": 746
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.963968787821556e-05,
+      "loss": 1.8135,
+      "step": 747
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.963739202449004e-05,
+      "loss": 1.7995,
+      "step": 748
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.963508890617945e-05,
+      "loss": 1.725,
+      "step": 749
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.963277852362083e-05,
+      "loss": 1.7897,
+      "step": 750
+    },
+    {
+      "epoch": 0.09,
+      "eval_loss": 1.6744195222854614,
+      "eval_runtime": 4348.5249,
+      "eval_samples_per_second": 2.815,
+      "eval_steps_per_second": 0.352,
+      "step": 750
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.963046087715235e-05,
+      "loss": 1.696,
+      "step": 751
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.96281359671132e-05,
+      "loss": 1.6665,
+      "step": 752
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.962580379384363e-05,
+      "loss": 1.5601,
+      "step": 753
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.962346435768494e-05,
+      "loss": 1.6852,
+      "step": 754
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.962111765897955e-05,
+      "loss": 1.7244,
+      "step": 755
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.961876369807088e-05,
+      "loss": 1.671,
+      "step": 756
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.961640247530345e-05,
+      "loss": 1.7275,
+      "step": 757
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.961403399102284e-05,
+      "loss": 1.6171,
+      "step": 758
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.961165824557567e-05,
+      "loss": 1.6079,
+      "step": 759
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.960927523930966e-05,
+      "loss": 1.6228,
+      "step": 760
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.960688497257353e-05,
+      "loss": 1.7507,
+      "step": 761
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.960448744571715e-05,
+      "loss": 1.4979,
+      "step": 762
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.960208265909136e-05,
+      "loss": 1.7367,
+      "step": 763
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.959967061304813e-05,
+      "loss": 1.6772,
+      "step": 764
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.959725130794048e-05,
+      "loss": 1.7043,
+      "step": 765
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.959482474412248e-05,
+      "loss": 1.7185,
+      "step": 766
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.959239092194925e-05,
+      "loss": 1.5921,
+      "step": 767
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.958994984177701e-05,
+      "loss": 1.6878,
+      "step": 768
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.958750150396298e-05,
+      "loss": 1.6324,
+      "step": 769
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.958504590886552e-05,
+      "loss": 1.6169,
+      "step": 770
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.9582583056844e-05,
+      "loss": 1.6401,
+      "step": 771
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.958011294825889e-05,
+      "loss": 1.5786,
+      "step": 772
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.957763558347163e-05,
+      "loss": 1.7911,
+      "step": 773
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.957515096284488e-05,
+      "loss": 1.6009,
+      "step": 774
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.95726590867422e-05,
+      "loss": 1.6425,
+      "step": 775
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.957015995552833e-05,
+      "loss": 1.6658,
+      "step": 776
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.956765356956899e-05,
+      "loss": 1.6148,
+      "step": 777
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.956513992923102e-05,
+      "loss": 1.7182,
+      "step": 778
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.95626190348823e-05,
+      "loss": 1.6179,
+      "step": 779
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.956009088689174e-05,
+      "loss": 1.6032,
+      "step": 780
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.955755548562937e-05,
+      "loss": 1.6677,
+      "step": 781
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.955501283146627e-05,
+      "loss": 1.6444,
+      "step": 782
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.955246292477453e-05,
+      "loss": 1.6189,
+      "step": 783
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.954990576592735e-05,
+      "loss": 1.5911,
+      "step": 784
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.9547341355299e-05,
+      "loss": 1.5075,
+      "step": 785
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.954476969326474e-05,
+      "loss": 1.5013,
+      "step": 786
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.954219078020098e-05,
+      "loss": 1.4645,
+      "step": 787
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.953960461648514e-05,
+      "loss": 1.7852,
+      "step": 788
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.953701120249572e-05,
+      "loss": 1.6583,
+      "step": 789
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.953441053861226e-05,
+      "loss": 1.8058,
+      "step": 790
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.95318026252154e-05,
+      "loss": 1.7017,
+      "step": 791
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.952918746268677e-05,
+      "loss": 1.5097,
+      "step": 792
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.952656505140917e-05,
+      "loss": 1.5855,
+      "step": 793
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.952393539176635e-05,
+      "loss": 1.5333,
+      "step": 794
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.952129848414319e-05,
+      "loss": 1.7133,
+      "step": 795
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.95186543289256e-05,
+      "loss": 1.6831,
+      "step": 796
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.951600292650055e-05,
+      "loss": 1.5898,
+      "step": 797
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.951334427725611e-05,
+      "loss": 1.7113,
+      "step": 798
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.951067838158135e-05,
+      "loss": 1.6701,
+      "step": 799
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.950800523986647e-05,
+      "loss": 1.744,
+      "step": 800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.950532485250266e-05,
+      "loss": 1.5537,
+      "step": 801
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.95026372198822e-05,
+      "loss": 1.782,
+      "step": 802
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.949994234239845e-05,
+      "loss": 1.7007,
+      "step": 803
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.949724022044583e-05,
+      "loss": 1.7217,
+      "step": 804
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.949453085441978e-05,
+      "loss": 1.7183,
+      "step": 805
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.94918142447168e-05,
+      "loss": 1.7363,
+      "step": 806
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.948909039173453e-05,
+      "loss": 1.5789,
+      "step": 807
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.948635929587158e-05,
+      "loss": 1.695,
+      "step": 808
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.948362095752766e-05,
+      "loss": 1.6405,
+      "step": 809
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.948087537710353e-05,
+      "loss": 1.6058,
+      "step": 810
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.947812255500103e-05,
+      "loss": 1.6195,
+      "step": 811
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.947536249162303e-05,
+      "loss": 1.6619,
+      "step": 812
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.947259518737347e-05,
+      "loss": 1.7094,
+      "step": 813
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.946982064265736e-05,
+      "loss": 1.7041,
+      "step": 814
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.946703885788077e-05,
+      "loss": 1.7196,
+      "step": 815
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.946424983345082e-05,
+      "loss": 1.6481,
+      "step": 816
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.946145356977568e-05,
+      "loss": 1.6516,
+      "step": 817
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.945865006726461e-05,
+      "loss": 1.6645,
+      "step": 818
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.945583932632788e-05,
+      "loss": 1.7008,
+      "step": 819
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.945302134737688e-05,
+      "loss": 1.7061,
+      "step": 820
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.945019613082403e-05,
+      "loss": 1.6723,
+      "step": 821
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.94473636770828e-05,
+      "loss": 1.7297,
+      "step": 822
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.944452398656774e-05,
+      "loss": 1.6317,
+      "step": 823
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.944167705969443e-05,
+      "loss": 1.6116,
+      "step": 824
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.943882289687952e-05,
+      "loss": 1.6809,
+      "step": 825
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.943596149854075e-05,
+      "loss": 1.6142,
+      "step": 826
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.943309286509687e-05,
+      "loss": 1.7707,
+      "step": 827
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.943021699696775e-05,
+      "loss": 1.7086,
+      "step": 828
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.942733389457426e-05,
+      "loss": 1.6646,
+      "step": 829
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.942444355833834e-05,
+      "loss": 1.6475,
+      "step": 830
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.9421545988683e-05,
+      "loss": 1.3557,
+      "step": 831
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.941864118603232e-05,
+      "loss": 1.6325,
+      "step": 832
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.941572915081143e-05,
+      "loss": 1.6872,
+      "step": 833
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.941280988344651e-05,
+      "loss": 1.5432,
+      "step": 834
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.940988338436481e-05,
+      "loss": 1.4032,
+      "step": 835
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.940694965399462e-05,
+      "loss": 1.5763,
+      "step": 836
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.94040086927653e-05,
+      "loss": 1.7203,
+      "step": 837
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.940106050110728e-05,
+      "loss": 1.5735,
+      "step": 838
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.939810507945204e-05,
+      "loss": 1.6542,
+      "step": 839
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.93951424282321e-05,
+      "loss": 1.5734,
+      "step": 840
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.939217254788104e-05,
+      "loss": 1.7334,
+      "step": 841
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.938919543883355e-05,
+      "loss": 1.6186,
+      "step": 842
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.938621110152533e-05,
+      "loss": 1.6191,
+      "step": 843
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.938321953639314e-05,
+      "loss": 1.532,
+      "step": 844
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.938022074387479e-05,
+      "loss": 1.6486,
+      "step": 845
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.937721472440916e-05,
+      "loss": 1.5388,
+      "step": 846
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.937420147843623e-05,
+      "loss": 1.6527,
+      "step": 847
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.937118100639696e-05,
+      "loss": 1.6268,
+      "step": 848
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.936815330873341e-05,
+      "loss": 1.5742,
+      "step": 849
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.93651183858887e-05,
+      "loss": 1.6839,
+      "step": 850
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.936207623830702e-05,
+      "loss": 1.6376,
+      "step": 851
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.935902686643357e-05,
+      "loss": 1.7089,
+      "step": 852
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.935597027071464e-05,
+      "loss": 1.5612,
+      "step": 853
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.935290645159759e-05,
+      "loss": 1.7967,
+      "step": 854
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.934983540953079e-05,
+      "loss": 1.6716,
+      "step": 855
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.93467571449637e-05,
+      "loss": 1.6283,
+      "step": 856
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.934367165834689e-05,
+      "loss": 1.7034,
+      "step": 857
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.934057895013185e-05,
+      "loss": 1.635,
+      "step": 858
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.933747902077124e-05,
+      "loss": 1.6598,
+      "step": 859
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.933437187071877e-05,
+      "loss": 1.534,
+      "step": 860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.933125750042913e-05,
+      "loss": 1.6978,
+      "step": 861
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.932813591035817e-05,
+      "loss": 1.6782,
+      "step": 862
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.932500710096271e-05,
+      "loss": 1.8354,
+      "step": 863
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.932187107270068e-05,
+      "loss": 1.6947,
+      "step": 864
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.931872782603103e-05,
+      "loss": 1.546,
+      "step": 865
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.931557736141379e-05,
+      "loss": 1.623,
+      "step": 866
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.931241967931005e-05,
+      "loss": 1.623,
+      "step": 867
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.930925478018192e-05,
+      "loss": 1.6758,
+      "step": 868
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.930608266449263e-05,
+      "loss": 1.6195,
+      "step": 869
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.93029033327064e-05,
+      "loss": 1.6796,
+      "step": 870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.929971678528855e-05,
+      "loss": 1.5877,
+      "step": 871
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.929652302270545e-05,
+      "loss": 1.6801,
+      "step": 872
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.929332204542448e-05,
+      "loss": 1.6055,
+      "step": 873
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.929011385391416e-05,
+      "loss": 1.6239,
+      "step": 874
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.928689844864398e-05,
+      "loss": 1.7406,
+      "step": 875
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.928367583008454e-05,
+      "loss": 1.6157,
+      "step": 876
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.928044599870749e-05,
+      "loss": 1.6495,
+      "step": 877
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.92772089549855e-05,
+      "loss": 1.6119,
+      "step": 878
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.927396469939235e-05,
+      "loss": 1.6971,
+      "step": 879
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.927071323240283e-05,
+      "loss": 1.626,
+      "step": 880
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.92674545544928e-05,
+      "loss": 1.6421,
+      "step": 881
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.926418866613918e-05,
+      "loss": 1.7128,
+      "step": 882
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.926091556781996e-05,
+      "loss": 1.7437,
+      "step": 883
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.925763526001415e-05,
+      "loss": 1.6921,
+      "step": 884
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.925434774320184e-05,
+      "loss": 1.6402,
+      "step": 885
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.925105301786415e-05,
+      "loss": 1.5448,
+      "step": 886
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.92477510844833e-05,
+      "loss": 1.5466,
+      "step": 887
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.924444194354253e-05,
+      "loss": 1.4402,
+      "step": 888
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.924112559552615e-05,
+      "loss": 1.4662,
+      "step": 889
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.923780204091948e-05,
+      "loss": 1.5989,
+      "step": 890
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.923447128020897e-05,
+      "loss": 1.7245,
+      "step": 891
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.923113331388207e-05,
+      "loss": 1.6998,
+      "step": 892
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.922778814242733e-05,
+      "loss": 1.6881,
+      "step": 893
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.922443576633428e-05,
+      "loss": 1.6461,
+      "step": 894
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.92210761860936e-05,
+      "loss": 1.6427,
+      "step": 895
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.921770940219695e-05,
+      "loss": 1.6078,
+      "step": 896
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.921433541513705e-05,
+      "loss": 1.7476,
+      "step": 897
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.921095422540771e-05,
+      "loss": 1.6747,
+      "step": 898
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.92075658335038e-05,
+      "loss": 1.7584,
+      "step": 899
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.92041702399212e-05,
+      "loss": 1.6921,
+      "step": 900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.920076744515685e-05,
+      "loss": 1.6434,
+      "step": 901
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.91973574497088e-05,
+      "loss": 1.5742,
+      "step": 902
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.919394025407607e-05,
+      "loss": 1.7078,
+      "step": 903
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.91905158587588e-05,
+      "loss": 1.6758,
+      "step": 904
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.918708426425818e-05,
+      "loss": 1.7028,
+      "step": 905
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.91836454710764e-05,
+      "loss": 1.6157,
+      "step": 906
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.918019947971673e-05,
+      "loss": 1.634,
+      "step": 907
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.917674629068355e-05,
+      "loss": 1.6845,
+      "step": 908
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.91732859044822e-05,
+      "loss": 1.686,
+      "step": 909
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.916981832161914e-05,
+      "loss": 1.7121,
+      "step": 910
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.916634354260186e-05,
+      "loss": 1.8033,
+      "step": 911
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.91628615679389e-05,
+      "loss": 1.6488,
+      "step": 912
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.915937239813986e-05,
+      "loss": 1.5833,
+      "step": 913
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.915587603371537e-05,
+      "loss": 1.6983,
+      "step": 914
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.915237247517718e-05,
+      "loss": 1.6266,
+      "step": 915
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.914886172303801e-05,
+      "loss": 1.6485,
+      "step": 916
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.914534377781168e-05,
+      "loss": 1.6246,
+      "step": 917
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.914181864001304e-05,
+      "loss": 1.586,
+      "step": 918
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.913828631015803e-05,
+      "loss": 1.7629,
+      "step": 919
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.91347467887636e-05,
+      "loss": 1.7097,
+      "step": 920
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.913120007634778e-05,
+      "loss": 1.6797,
+      "step": 921
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.912764617342963e-05,
+      "loss": 1.6406,
+      "step": 922
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.912408508052927e-05,
+      "loss": 1.6531,
+      "step": 923
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.91205167981679e-05,
+      "loss": 1.6553,
+      "step": 924
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.911694132686774e-05,
+      "loss": 1.6999,
+      "step": 925
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.911335866715205e-05,
+      "loss": 1.6673,
+      "step": 926
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.91097688195452e-05,
+      "loss": 1.6849,
+      "step": 927
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.910617178457254e-05,
+      "loss": 1.6062,
+      "step": 928
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.910256756276052e-05,
+      "loss": 1.731,
+      "step": 929
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.909895615463665e-05,
+      "loss": 1.5787,
+      "step": 930
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.909533756072944e-05,
+      "loss": 1.5705,
+      "step": 931
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.909171178156849e-05,
+      "loss": 1.5927,
+      "step": 932
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.908807881768445e-05,
+      "loss": 1.625,
+      "step": 933
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.908443866960902e-05,
+      "loss": 1.545,
+      "step": 934
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.908079133787495e-05,
+      "loss": 1.5208,
+      "step": 935
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.907713682301602e-05,
+      "loss": 1.7268,
+      "step": 936
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.90734751255671e-05,
+      "loss": 1.457,
+      "step": 937
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.906980624606406e-05,
+      "loss": 1.6833,
+      "step": 938
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.906613018504389e-05,
+      "loss": 1.6644,
+      "step": 939
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.906244694304457e-05,
+      "loss": 1.636,
+      "step": 940
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.905875652060514e-05,
+      "loss": 1.6528,
+      "step": 941
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.905505891826574e-05,
+      "loss": 1.5559,
+      "step": 942
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.905135413656751e-05,
+      "loss": 1.6882,
+      "step": 943
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.904764217605265e-05,
+      "loss": 1.5119,
+      "step": 944
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.904392303726441e-05,
+      "loss": 1.6743,
+      "step": 945
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.904019672074714e-05,
+      "loss": 1.5569,
+      "step": 946
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.903646322704614e-05,
+      "loss": 1.5326,
+      "step": 947
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.903272255670785e-05,
+      "loss": 1.6281,
+      "step": 948
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.902897471027972e-05,
+      "loss": 1.6137,
+      "step": 949
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.902521968831026e-05,
+      "loss": 1.6935,
+      "step": 950
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.902145749134906e-05,
+      "loss": 1.7324,
+      "step": 951
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.901768811994667e-05,
+      "loss": 1.6051,
+      "step": 952
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.901391157465479e-05,
+      "loss": 1.5942,
+      "step": 953
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.901012785602612e-05,
+      "loss": 1.5761,
+      "step": 954
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.900633696461441e-05,
+      "loss": 1.6864,
+      "step": 955
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.900253890097449e-05,
+      "loss": 1.6731,
+      "step": 956
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.899873366566219e-05,
+      "loss": 1.7475,
+      "step": 957
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.899492125923445e-05,
+      "loss": 1.6763,
+      "step": 958
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.899110168224918e-05,
+      "loss": 1.6459,
+      "step": 959
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.898727493526544e-05,
+      "loss": 1.7011,
+      "step": 960
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.898344101884325e-05,
+      "loss": 1.4971,
+      "step": 961
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.897959993354374e-05,
+      "loss": 1.8238,
+      "step": 962
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.897575167992906e-05,
+      "loss": 1.6501,
+      "step": 963
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.897189625856239e-05,
+      "loss": 1.6591,
+      "step": 964
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.8968033670008e-05,
+      "loss": 1.6619,
+      "step": 965
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.896416391483118e-05,
+      "loss": 1.6575,
+      "step": 966
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.896028699359832e-05,
+      "loss": 1.6125,
+      "step": 967
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.895640290687677e-05,
+      "loss": 1.6051,
+      "step": 968
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.895251165523501e-05,
+      "loss": 1.5627,
+      "step": 969
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.894861323924252e-05,
+      "loss": 1.706,
+      "step": 970
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.894470765946986e-05,
+      "loss": 1.7494,
+      "step": 971
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.894079491648862e-05,
+      "loss": 1.6834,
+      "step": 972
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.89368750108714e-05,
+      "loss": 1.6633,
+      "step": 973
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.893294794319197e-05,
+      "loss": 1.6198,
+      "step": 974
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.8929013714025e-05,
+      "loss": 1.6857,
+      "step": 975
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.892507232394632e-05,
+      "loss": 1.7124,
+      "step": 976
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.892112377353273e-05,
+      "loss": 1.6604,
+      "step": 977
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.891716806336212e-05,
+      "loss": 1.7058,
+      "step": 978
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.891320519401345e-05,
+      "loss": 1.5829,
+      "step": 979
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.890923516606669e-05,
+      "loss": 1.4607,
+      "step": 980
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.890525798010283e-05,
+      "loss": 1.6805,
+      "step": 981
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.890127363670396e-05,
+      "loss": 1.6594,
+      "step": 982
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.889728213645322e-05,
+      "loss": 1.4878,
+      "step": 983
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.889328347993477e-05,
+      "loss": 1.5211,
+      "step": 984
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.888927766773381e-05,
+      "loss": 1.5707,
+      "step": 985
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.888526470043662e-05,
+      "loss": 1.69,
+      "step": 986
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.88812445786305e-05,
+      "loss": 1.6166,
+      "step": 987
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.887721730290383e-05,
+      "loss": 1.4937,
+      "step": 988
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.887318287384598e-05,
+      "loss": 1.6902,
+      "step": 989
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.886914129204742e-05,
+      "loss": 1.676,
+      "step": 990
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.886509255809965e-05,
+      "loss": 1.7416,
+      "step": 991
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.886103667259519e-05,
+      "loss": 1.6612,
+      "step": 992
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.885697363612767e-05,
+      "loss": 1.3844,
+      "step": 993
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.88529034492917e-05,
+      "loss": 1.5381,
+      "step": 994
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.884882611268298e-05,
+      "loss": 1.626,
+      "step": 995
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.884474162689823e-05,
+      "loss": 1.489,
+      "step": 996
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.884064999253524e-05,
+      "loss": 1.6029,
+      "step": 997
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.883655121019282e-05,
+      "loss": 1.7026,
+      "step": 998
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.883244528047084e-05,
+      "loss": 1.7179,
+      "step": 999
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.882833220397022e-05,
+      "loss": 1.6522,
+      "step": 1000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.882421198129293e-05,
+      "loss": 1.548,
+      "step": 1001
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.882008461304195e-05,
+      "loss": 1.6774,
+      "step": 1002
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.881595009982136e-05,
+      "loss": 1.7477,
+      "step": 1003
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.881180844223625e-05,
+      "loss": 1.7137,
+      "step": 1004
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.880765964089276e-05,
+      "loss": 1.6867,
+      "step": 1005
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.880350369639808e-05,
+      "loss": 1.6688,
+      "step": 1006
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.879934060936047e-05,
+      "loss": 1.6474,
+      "step": 1007
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.879517038038916e-05,
+      "loss": 1.6832,
+      "step": 1008
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.879099301009453e-05,
+      "loss": 1.6372,
+      "step": 1009
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.878680849908792e-05,
+      "loss": 1.5778,
+      "step": 1010
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.878261684798174e-05,
+      "loss": 1.6846,
+      "step": 1011
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.877841805738948e-05,
+      "loss": 1.7359,
+      "step": 1012
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.877421212792563e-05,
+      "loss": 1.6067,
+      "step": 1013
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.876999906020573e-05,
+      "loss": 1.6308,
+      "step": 1014
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.876577885484639e-05,
+      "loss": 1.699,
+      "step": 1015
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.876155151246524e-05,
+      "loss": 1.5827,
+      "step": 1016
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.875731703368098e-05,
+      "loss": 1.5175,
+      "step": 1017
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.875307541911333e-05,
+      "loss": 1.7313,
+      "step": 1018
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.874882666938305e-05,
+      "loss": 1.6211,
+      "step": 1019
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.8744570785112e-05,
+      "loss": 1.6393,
+      "step": 1020
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.874030776692298e-05,
+      "loss": 1.6215,
+      "step": 1021
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.873603761543992e-05,
+      "loss": 1.6651,
+      "step": 1022
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.873176033128781e-05,
+      "loss": 1.5342,
+      "step": 1023
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.872747591509259e-05,
+      "loss": 1.5715,
+      "step": 1024
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.872318436748133e-05,
+      "loss": 1.6652,
+      "step": 1025
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.871888568908209e-05,
+      "loss": 1.6933,
+      "step": 1026
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.871457988052401e-05,
+      "loss": 1.5858,
+      "step": 1027
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.871026694243725e-05,
+      "loss": 1.6224,
+      "step": 1028
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.870594687545304e-05,
+      "loss": 1.644,
+      "step": 1029
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.870161968020361e-05,
+      "loss": 1.517,
+      "step": 1030
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.869728535732227e-05,
+      "loss": 1.6623,
+      "step": 1031
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.869294390744336e-05,
+      "loss": 1.4544,
+      "step": 1032
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.868859533120226e-05,
+      "loss": 1.7489,
+      "step": 1033
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.86842396292354e-05,
+      "loss": 1.6724,
+      "step": 1034
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.867987680218025e-05,
+      "loss": 1.4977,
+      "step": 1035
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.867550685067535e-05,
+      "loss": 1.4725,
+      "step": 1036
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.867112977536021e-05,
+      "loss": 1.3146,
+      "step": 1037
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.866674557687545e-05,
+      "loss": 1.5802,
+      "step": 1038
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.866235425586271e-05,
+      "loss": 1.631,
+      "step": 1039
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.865795581296469e-05,
+      "loss": 1.5665,
+      "step": 1040
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.865355024882508e-05,
+      "loss": 1.6591,
+      "step": 1041
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.864913756408867e-05,
+      "loss": 1.722,
+      "step": 1042
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.864471775940127e-05,
+      "loss": 1.5733,
+      "step": 1043
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.864029083540974e-05,
+      "loss": 1.664,
+      "step": 1044
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.863585679276195e-05,
+      "loss": 1.7661,
+      "step": 1045
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.863141563210685e-05,
+      "loss": 1.5577,
+      "step": 1046
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.862696735409444e-05,
+      "loss": 1.6889,
+      "step": 1047
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.86225119593757e-05,
+      "loss": 1.7192,
+      "step": 1048
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.861804944860271e-05,
+      "loss": 1.6843,
+      "step": 1049
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.861357982242855e-05,
+      "loss": 1.5724,
+      "step": 1050
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.86091030815074e-05,
+      "loss": 1.8944,
+      "step": 1051
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.860461922649443e-05,
+      "loss": 1.6383,
+      "step": 1052
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.860012825804585e-05,
+      "loss": 1.6737,
+      "step": 1053
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.859563017681896e-05,
+      "loss": 1.6832,
+      "step": 1054
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.859112498347202e-05,
+      "loss": 1.5982,
+      "step": 1055
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.858661267866443e-05,
+      "loss": 1.602,
+      "step": 1056
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.858209326305655e-05,
+      "loss": 1.6182,
+      "step": 1057
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.857756673730982e-05,
+      "loss": 1.7023,
+      "step": 1058
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.857303310208671e-05,
+      "loss": 1.6981,
+      "step": 1059
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.856849235805073e-05,
+      "loss": 1.698,
+      "step": 1060
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.856394450586642e-05,
+      "loss": 1.6756,
+      "step": 1061
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.855938954619939e-05,
+      "loss": 1.653,
+      "step": 1062
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.855482747971628e-05,
+      "loss": 1.7553,
+      "step": 1063
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.855025830708473e-05,
+      "loss": 1.7174,
+      "step": 1064
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.854568202897347e-05,
+      "loss": 1.5547,
+      "step": 1065
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.854109864605228e-05,
+      "loss": 1.7043,
+      "step": 1066
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.853650815899189e-05,
+      "loss": 1.5754,
+      "step": 1067
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.853191056846419e-05,
+      "loss": 1.6618,
+      "step": 1068
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.852730587514203e-05,
+      "loss": 1.6643,
+      "step": 1069
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.852269407969932e-05,
+      "loss": 1.7109,
+      "step": 1070
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.8518075182811e-05,
+      "loss": 1.6197,
+      "step": 1071
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.851344918515307e-05,
+      "loss": 1.7104,
+      "step": 1072
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.850881608740257e-05,
+      "loss": 1.6336,
+      "step": 1073
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.850417589023755e-05,
+      "loss": 1.643,
+      "step": 1074
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.849952859433713e-05,
+      "loss": 1.6973,
+      "step": 1075
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.849487420038143e-05,
+      "loss": 1.6077,
+      "step": 1076
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.84902127090517e-05,
+      "loss": 1.587,
+      "step": 1077
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.848554412103008e-05,
+      "loss": 1.3603,
+      "step": 1078
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.848086843699989e-05,
+      "loss": 1.6419,
+      "step": 1079
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.847618565764541e-05,
+      "loss": 1.6188,
+      "step": 1080
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.847149578365197e-05,
+      "loss": 1.6472,
+      "step": 1081
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.846679881570597e-05,
+      "loss": 1.5982,
+      "step": 1082
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.84620947544948e-05,
+      "loss": 1.6044,
+      "step": 1083
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.845738360070695e-05,
+      "loss": 1.6941,
+      "step": 1084
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.845266535503188e-05,
+      "loss": 1.5988,
+      "step": 1085
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.844794001816013e-05,
+      "loss": 1.524,
+      "step": 1086
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.844320759078327e-05,
+      "loss": 1.6763,
+      "step": 1087
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.843846807359389e-05,
+      "loss": 1.5858,
+      "step": 1088
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.843372146728566e-05,
+      "loss": 1.7129,
+      "step": 1089
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.842896777255324e-05,
+      "loss": 1.7051,
+      "step": 1090
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.842420699009236e-05,
+      "loss": 1.6468,
+      "step": 1091
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.841943912059977e-05,
+      "loss": 1.8096,
+      "step": 1092
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.841466416477325e-05,
+      "loss": 1.7657,
+      "step": 1093
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.840988212331166e-05,
+      "loss": 1.6477,
+      "step": 1094
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.840509299691483e-05,
+      "loss": 1.7033,
+      "step": 1095
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.840029678628368e-05,
+      "loss": 1.5561,
+      "step": 1096
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.839549349212016e-05,
+      "loss": 1.6126,
+      "step": 1097
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.839068311512725e-05,
+      "loss": 1.6522,
+      "step": 1098
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.838586565600894e-05,
+      "loss": 1.6589,
+      "step": 1099
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.838104111547029e-05,
+      "loss": 1.7199,
+      "step": 1100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.837620949421738e-05,
+      "loss": 1.676,
+      "step": 1101
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.837137079295736e-05,
+      "loss": 1.6636,
+      "step": 1102
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.836652501239834e-05,
+      "loss": 1.6012,
+      "step": 1103
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.836167215324957e-05,
+      "loss": 1.6135,
+      "step": 1104
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.835681221622123e-05,
+      "loss": 1.627,
+      "step": 1105
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.835194520202465e-05,
+      "loss": 1.5147,
+      "step": 1106
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.834707111137207e-05,
+      "loss": 1.6872,
+      "step": 1107
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.834218994497686e-05,
+      "loss": 1.5475,
+      "step": 1108
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.833730170355338e-05,
+      "loss": 1.7223,
+      "step": 1109
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.833240638781704e-05,
+      "loss": 1.5809,
+      "step": 1110
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.83275039984843e-05,
+      "loss": 1.6208,
+      "step": 1111
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.832259453627263e-05,
+      "loss": 1.6922,
+      "step": 1112
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.831767800190055e-05,
+      "loss": 1.6112,
+      "step": 1113
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.831275439608758e-05,
+      "loss": 1.618,
+      "step": 1114
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.830782371955436e-05,
+      "loss": 1.6526,
+      "step": 1115
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.830288597302247e-05,
+      "loss": 1.5867,
+      "step": 1116
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.829794115721455e-05,
+      "loss": 1.677,
+      "step": 1117
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.829298927285434e-05,
+      "loss": 1.6575,
+      "step": 1118
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.828803032066653e-05,
+      "loss": 1.6053,
+      "step": 1119
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.828306430137688e-05,
+      "loss": 1.6053,
+      "step": 1120
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.827809121571219e-05,
+      "loss": 1.6534,
+      "step": 1121
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.827311106440027e-05,
+      "loss": 1.5998,
+      "step": 1122
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.826812384817002e-05,
+      "loss": 1.7216,
+      "step": 1123
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.826312956775128e-05,
+      "loss": 1.7335,
+      "step": 1124
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.8258128223875e-05,
+      "loss": 1.6979,
+      "step": 1125
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.825311981727318e-05,
+      "loss": 1.7056,
+      "step": 1126
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.824810434867877e-05,
+      "loss": 1.6914,
+      "step": 1127
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.82430818188258e-05,
+      "loss": 1.6645,
+      "step": 1128
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.823805222844937e-05,
+      "loss": 1.6887,
+      "step": 1129
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.823301557828552e-05,
+      "loss": 1.6542,
+      "step": 1130
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.822797186907143e-05,
+      "loss": 1.501,
+      "step": 1131
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.822292110154525e-05,
+      "loss": 1.7378,
+      "step": 1132
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.821786327644615e-05,
+      "loss": 1.5521,
+      "step": 1133
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.821279839451439e-05,
+      "loss": 1.5522,
+      "step": 1134
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.820772645649121e-05,
+      "loss": 1.4824,
+      "step": 1135
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.820264746311892e-05,
+      "loss": 1.4851,
+      "step": 1136
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.819756141514084e-05,
+      "loss": 1.6901,
+      "step": 1137
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.819246831330132e-05,
+      "loss": 1.6016,
+      "step": 1138
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.818736815834576e-05,
+      "loss": 1.5273,
+      "step": 1139
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.818226095102057e-05,
+      "loss": 1.5302,
+      "step": 1140
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.817714669207323e-05,
+      "loss": 1.6513,
+      "step": 1141
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.81720253822522e-05,
+      "loss": 1.6392,
+      "step": 1142
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.816689702230704e-05,
+      "loss": 1.5935,
+      "step": 1143
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.816176161298826e-05,
+      "loss": 1.6646,
+      "step": 1144
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.815661915504746e-05,
+      "loss": 1.6226,
+      "step": 1145
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.815146964923725e-05,
+      "loss": 1.6305,
+      "step": 1146
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.814631309631127e-05,
+      "loss": 1.6634,
+      "step": 1147
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.814114949702422e-05,
+      "loss": 1.6646,
+      "step": 1148
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.813597885213178e-05,
+      "loss": 1.5877,
+      "step": 1149
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.813080116239071e-05,
+      "loss": 1.6967,
+      "step": 1150
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.812561642855878e-05,
+      "loss": 1.5477,
+      "step": 1151
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.81204246513948e-05,
+      "loss": 1.6401,
+      "step": 1152
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.811522583165855e-05,
+      "loss": 1.6626,
+      "step": 1153
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.811001997011097e-05,
+      "loss": 1.6059,
+      "step": 1154
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.81048070675139e-05,
+      "loss": 1.7272,
+      "step": 1155
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.809958712463028e-05,
+      "loss": 1.6279,
+      "step": 1156
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.809436014222407e-05,
+      "loss": 1.6461,
+      "step": 1157
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.808912612106025e-05,
+      "loss": 1.4729,
+      "step": 1158
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.808388506190484e-05,
+      "loss": 1.5849,
+      "step": 1159
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.807863696552489e-05,
+      "loss": 1.7291,
+      "step": 1160
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.807338183268844e-05,
+      "loss": 1.5895,
+      "step": 1161
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.806811966416463e-05,
+      "loss": 1.6119,
+      "step": 1162
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.806285046072358e-05,
+      "loss": 1.6594,
+      "step": 1163
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.805757422313649e-05,
+      "loss": 1.5694,
+      "step": 1164
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.805229095217548e-05,
+      "loss": 1.6055,
+      "step": 1165
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.804700064861386e-05,
+      "loss": 1.6233,
+      "step": 1166
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.804170331322579e-05,
+      "loss": 1.6705,
+      "step": 1167
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.803639894678664e-05,
+      "loss": 1.7675,
+      "step": 1168
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.803108755007267e-05,
+      "loss": 1.6547,
+      "step": 1169
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.802576912386121e-05,
+      "loss": 1.6368,
+      "step": 1170
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.802044366893066e-05,
+      "loss": 1.6457,
+      "step": 1171
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.80151111860604e-05,
+      "loss": 1.677,
+      "step": 1172
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.800977167603087e-05,
+      "loss": 1.6945,
+      "step": 1173
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.800442513962352e-05,
+      "loss": 1.6528,
+      "step": 1174
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.79990715776208e-05,
+      "loss": 1.6695,
+      "step": 1175
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.799371099080625e-05,
+      "loss": 1.6908,
+      "step": 1176
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.798834337996441e-05,
+      "loss": 1.5928,
+      "step": 1177
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.798296874588085e-05,
+      "loss": 1.6976,
+      "step": 1178
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.797758708934213e-05,
+      "loss": 1.6687,
+      "step": 1179
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.797219841113591e-05,
+      "loss": 1.6552,
+      "step": 1180
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.796680271205083e-05,
+      "loss": 1.5784,
+      "step": 1181
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.796139999287657e-05,
+      "loss": 1.5568,
+      "step": 1182
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.79559902544038e-05,
+      "loss": 1.7018,
+      "step": 1183
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.795057349742431e-05,
+      "loss": 1.7148,
+      "step": 1184
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.794514972273081e-05,
+      "loss": 1.5901,
+      "step": 1185
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.793971893111711e-05,
+      "loss": 1.5049,
+      "step": 1186
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.793428112337801e-05,
+      "loss": 1.5757,
+      "step": 1187
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.792883630030937e-05,
+      "loss": 1.7333,
+      "step": 1188
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.792338446270804e-05,
+      "loss": 1.616,
+      "step": 1189
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.791792561137192e-05,
+      "loss": 1.6733,
+      "step": 1190
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.791245974709992e-05,
+      "loss": 1.6497,
+      "step": 1191
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.7906986870692e-05,
+      "loss": 1.5437,
+      "step": 1192
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.790150698294912e-05,
+      "loss": 1.388,
+      "step": 1193
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.789602008467329e-05,
+      "loss": 1.3259,
+      "step": 1194
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.789052617666752e-05,
+      "loss": 1.6174,
+      "step": 1195
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.788502525973587e-05,
+      "loss": 1.5411,
+      "step": 1196
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.787951733468342e-05,
+      "loss": 1.7207,
+      "step": 1197
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.787400240231627e-05,
+      "loss": 1.6548,
+      "step": 1198
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.786848046344154e-05,
+      "loss": 1.7569,
+      "step": 1199
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.786295151886738e-05,
+      "loss": 1.7464,
+      "step": 1200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.785741556940298e-05,
+      "loss": 1.6567,
+      "step": 1201
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.785187261585853e-05,
+      "loss": 1.5857,
+      "step": 1202
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.784632265904527e-05,
+      "loss": 1.798,
+      "step": 1203
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.784076569977546e-05,
+      "loss": 1.6756,
+      "step": 1204
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.783520173886236e-05,
+      "loss": 1.5211,
+      "step": 1205
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.782963077712029e-05,
+      "loss": 1.5004,
+      "step": 1206
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.782405281536455e-05,
+      "loss": 1.6212,
+      "step": 1207
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.781846785441152e-05,
+      "loss": 1.5991,
+      "step": 1208
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.781287589507857e-05,
+      "loss": 1.7031,
+      "step": 1209
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.780727693818409e-05,
+      "loss": 1.5324,
+      "step": 1210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.780167098454751e-05,
+      "loss": 1.6562,
+      "step": 1211
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.779605803498927e-05,
+      "loss": 1.6286,
+      "step": 1212
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.779043809033088e-05,
+      "loss": 1.5874,
+      "step": 1213
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.778481115139479e-05,
+      "loss": 1.6277,
+      "step": 1214
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.777917721900454e-05,
+      "loss": 1.6461,
+      "step": 1215
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.777353629398467e-05,
+      "loss": 1.7341,
+      "step": 1216
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.776788837716075e-05,
+      "loss": 1.5645,
+      "step": 1217
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.776223346935935e-05,
+      "loss": 1.6332,
+      "step": 1218
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.775657157140813e-05,
+      "loss": 1.7013,
+      "step": 1219
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.77509026841357e-05,
+      "loss": 1.5809,
+      "step": 1220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.774522680837168e-05,
+      "loss": 1.6794,
+      "step": 1221
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.773954394494682e-05,
+      "loss": 1.6823,
+      "step": 1222
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.773385409469278e-05,
+      "loss": 1.639,
+      "step": 1223
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.772815725844231e-05,
+      "loss": 1.4752,
+      "step": 1224
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.772245343702913e-05,
+      "loss": 1.7273,
+      "step": 1225
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.771674263128805e-05,
+      "loss": 1.5713,
+      "step": 1226
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.771102484205483e-05,
+      "loss": 1.6799,
+      "step": 1227
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.770530007016632e-05,
+      "loss": 1.58,
+      "step": 1228
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.769956831646033e-05,
+      "loss": 1.5718,
+      "step": 1229
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.769382958177573e-05,
+      "loss": 1.7414,
+      "step": 1230
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.76880838669524e-05,
+      "loss": 1.5359,
+      "step": 1231
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.768233117283123e-05,
+      "loss": 1.5189,
+      "step": 1232
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.767657150025418e-05,
+      "loss": 1.4845,
+      "step": 1233
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.767080485006416e-05,
+      "loss": 1.7151,
+      "step": 1234
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.766503122310516e-05,
+      "loss": 1.4603,
+      "step": 1235
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.765925062022216e-05,
+      "loss": 1.5236,
+      "step": 1236
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.765346304226116e-05,
+      "loss": 1.6764,
+      "step": 1237
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.764766849006922e-05,
+      "loss": 1.475,
+      "step": 1238
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.764186696449433e-05,
+      "loss": 1.6632,
+      "step": 1239
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.763605846638562e-05,
+      "loss": 1.544,
+      "step": 1240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.763024299659316e-05,
+      "loss": 1.6625,
+      "step": 1241
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.762442055596806e-05,
+      "loss": 1.6738,
+      "step": 1242
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.761859114536246e-05,
+      "loss": 1.5271,
+      "step": 1243
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.761275476562951e-05,
+      "loss": 1.6294,
+      "step": 1244
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.760691141762337e-05,
+      "loss": 1.6547,
+      "step": 1245
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.760106110219926e-05,
+      "loss": 1.5507,
+      "step": 1246
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.759520382021336e-05,
+      "loss": 1.4926,
+      "step": 1247
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.758933957252291e-05,
+      "loss": 1.6088,
+      "step": 1248
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.758346835998617e-05,
+      "loss": 1.7528,
+      "step": 1249
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.75775901834624e-05,
+      "loss": 1.7717,
+      "step": 1250
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.75717050438119e-05,
+      "loss": 1.5835,
+      "step": 1251
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.756581294189597e-05,
+      "loss": 1.6642,
+      "step": 1252
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.755991387857695e-05,
+      "loss": 1.6769,
+      "step": 1253
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.755400785471816e-05,
+      "loss": 1.58,
+      "step": 1254
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.7548094871184e-05,
+      "loss": 1.5188,
+      "step": 1255
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.754217492883981e-05,
+      "loss": 1.605,
+      "step": 1256
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.753624802855204e-05,
+      "loss": 1.5866,
+      "step": 1257
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.753031417118806e-05,
+      "loss": 1.6225,
+      "step": 1258
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.752437335761635e-05,
+      "loss": 1.5256,
+      "step": 1259
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.751842558870635e-05,
+      "loss": 1.6228,
+      "step": 1260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.751247086532854e-05,
+      "loss": 1.5643,
+      "step": 1261
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.75065091883544e-05,
+      "loss": 1.6523,
+      "step": 1262
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.750054055865646e-05,
+      "loss": 1.6484,
+      "step": 1263
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.74945649771082e-05,
+      "loss": 1.6189,
+      "step": 1264
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.748858244458422e-05,
+      "loss": 1.6427,
+      "step": 1265
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.748259296196006e-05,
+      "loss": 1.5903,
+      "step": 1266
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.74765965301123e-05,
+      "loss": 1.6592,
+      "step": 1267
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.747059314991853e-05,
+      "loss": 1.6103,
+      "step": 1268
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.746458282225739e-05,
+      "loss": 1.685,
+      "step": 1269
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.745856554800848e-05,
+      "loss": 1.5843,
+      "step": 1270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.745254132805244e-05,
+      "loss": 1.649,
+      "step": 1271
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.744651016327095e-05,
+      "loss": 1.6806,
+      "step": 1272
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.744047205454671e-05,
+      "loss": 1.6369,
+      "step": 1273
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.743442700276338e-05,
+      "loss": 1.6756,
+      "step": 1274
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.74283750088057e-05,
+      "loss": 1.6011,
+      "step": 1275
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.742231607355936e-05,
+      "loss": 1.5772,
+      "step": 1276
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.741625019791115e-05,
+      "loss": 1.627,
+      "step": 1277
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.74101773827488e-05,
+      "loss": 1.6797,
+      "step": 1278
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.74040976289611e-05,
+      "loss": 1.6088,
+      "step": 1279
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.739801093743781e-05,
+      "loss": 1.3166,
+      "step": 1280
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.739191730906978e-05,
+      "loss": 1.572,
+      "step": 1281
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.738581674474881e-05,
+      "loss": 1.4797,
+      "step": 1282
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.737970924536774e-05,
+      "loss": 1.5584,
+      "step": 1283
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.737359481182041e-05,
+      "loss": 1.6243,
+      "step": 1284
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.73674734450017e-05,
+      "loss": 1.4483,
+      "step": 1285
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.736134514580747e-05,
+      "loss": 1.3759,
+      "step": 1286
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.735520991513465e-05,
+      "loss": 1.4831,
+      "step": 1287
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.734906775388112e-05,
+      "loss": 1.6383,
+      "step": 1288
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.734291866294582e-05,
+      "loss": 1.5084,
+      "step": 1289
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.733676264322868e-05,
+      "loss": 1.7044,
+      "step": 1290
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.733059969563067e-05,
+      "loss": 1.6632,
+      "step": 1291
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.732442982105373e-05,
+      "loss": 1.6034,
+      "step": 1292
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.731825302040085e-05,
+      "loss": 1.656,
+      "step": 1293
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.731206929457604e-05,
+      "loss": 1.8776,
+      "step": 1294
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.730587864448428e-05,
+      "loss": 1.5892,
+      "step": 1295
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.72996810710316e-05,
+      "loss": 1.5481,
+      "step": 1296
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.729347657512504e-05,
+      "loss": 1.6325,
+      "step": 1297
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.728726515767265e-05,
+      "loss": 1.5969,
+      "step": 1298
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.72810468195835e-05,
+      "loss": 1.6622,
+      "step": 1299
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.727482156176764e-05,
+      "loss": 1.8498,
+      "step": 1300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.726858938513615e-05,
+      "loss": 1.6009,
+      "step": 1301
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.726235029060115e-05,
+      "loss": 1.6948,
+      "step": 1302
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.725610427907575e-05,
+      "loss": 1.6578,
+      "step": 1303
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.724985135147404e-05,
+      "loss": 1.5366,
+      "step": 1304
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.724359150871121e-05,
+      "loss": 1.78,
+      "step": 1305
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.723732475170337e-05,
+      "loss": 1.6246,
+      "step": 1306
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.723105108136767e-05,
+      "loss": 1.5264,
+      "step": 1307
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.722477049862233e-05,
+      "loss": 1.6941,
+      "step": 1308
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.721848300438648e-05,
+      "loss": 1.6207,
+      "step": 1309
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.721218859958035e-05,
+      "loss": 1.5308,
+      "step": 1310
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.720588728512511e-05,
+      "loss": 1.5974,
+      "step": 1311
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.7199579061943e-05,
+      "loss": 1.542,
+      "step": 1312
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.719326393095725e-05,
+      "loss": 1.5083,
+      "step": 1313
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.71869418930921e-05,
+      "loss": 1.6111,
+      "step": 1314
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.718061294927279e-05,
+      "loss": 1.6312,
+      "step": 1315
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.717427710042559e-05,
+      "loss": 1.6274,
+      "step": 1316
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.716793434747775e-05,
+      "loss": 1.6151,
+      "step": 1317
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.716158469135759e-05,
+      "loss": 1.5817,
+      "step": 1318
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.715522813299436e-05,
+      "loss": 1.5976,
+      "step": 1319
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.714886467331839e-05,
+      "loss": 1.5981,
+      "step": 1320
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.714249431326096e-05,
+      "loss": 1.6098,
+      "step": 1321
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.713611705375442e-05,
+      "loss": 1.5811,
+      "step": 1322
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.712973289573211e-05,
+      "loss": 1.5029,
+      "step": 1323
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.712334184012837e-05,
+      "loss": 1.5653,
+      "step": 1324
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.711694388787851e-05,
+      "loss": 1.5678,
+      "step": 1325
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.711053903991894e-05,
+      "loss": 1.5926,
+      "step": 1326
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.710412729718699e-05,
+      "loss": 1.5405,
+      "step": 1327
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.709770866062104e-05,
+      "loss": 1.6926,
+      "step": 1328
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.709128313116052e-05,
+      "loss": 1.2353,
+      "step": 1329
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.708485070974578e-05,
+      "loss": 1.5508,
+      "step": 1330
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.707841139731827e-05,
+      "loss": 1.5169,
+      "step": 1331
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.707196519482034e-05,
+      "loss": 1.6458,
+      "step": 1332
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.706551210319545e-05,
+      "loss": 1.6978,
+      "step": 1333
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.705905212338806e-05,
+      "loss": 1.5763,
+      "step": 1334
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.705258525634356e-05,
+      "loss": 1.4782,
+      "step": 1335
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.70461115030084e-05,
+      "loss": 1.6119,
+      "step": 1336
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.703963086433004e-05,
+      "loss": 1.4438,
+      "step": 1337
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.703314334125696e-05,
+      "loss": 1.6374,
+      "step": 1338
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.702664893473862e-05,
+      "loss": 1.6406,
+      "step": 1339
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.702014764572548e-05,
+      "loss": 1.6417,
+      "step": 1340
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.701363947516904e-05,
+      "loss": 1.5407,
+      "step": 1341
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.700712442402177e-05,
+      "loss": 1.6607,
+      "step": 1342
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.700060249323721e-05,
+      "loss": 1.514,
+      "step": 1343
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.699407368376982e-05,
+      "loss": 1.5952,
+      "step": 1344
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.698753799657514e-05,
+      "loss": 1.6766,
+      "step": 1345
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.698099543260968e-05,
+      "loss": 1.4661,
+      "step": 1346
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.697444599283095e-05,
+      "loss": 1.5984,
+      "step": 1347
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.69678896781975e-05,
+      "loss": 1.786,
+      "step": 1348
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.696132648966887e-05,
+      "loss": 1.7939,
+      "step": 1349
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.695475642820558e-05,
+      "loss": 1.8647,
+      "step": 1350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.694817949476918e-05,
+      "loss": 1.5458,
+      "step": 1351
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.694159569032226e-05,
+      "loss": 1.6438,
+      "step": 1352
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.693500501582834e-05,
+      "loss": 1.7194,
+      "step": 1353
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.6928407472252e-05,
+      "loss": 1.5244,
+      "step": 1354
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.692180306055881e-05,
+      "loss": 1.6537,
+      "step": 1355
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.691519178171535e-05,
+      "loss": 1.6286,
+      "step": 1356
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.690857363668922e-05,
+      "loss": 1.7968,
+      "step": 1357
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.690194862644896e-05,
+      "loss": 1.5863,
+      "step": 1358
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.689531675196419e-05,
+      "loss": 1.6933,
+      "step": 1359
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.68886780142055e-05,
+      "loss": 1.6449,
+      "step": 1360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.68820324141445e-05,
+      "loss": 1.5259,
+      "step": 1361
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.687537995275377e-05,
+      "loss": 1.6777,
+      "step": 1362
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.686872063100694e-05,
+      "loss": 1.6621,
+      "step": 1363
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.686205444987862e-05,
+      "loss": 1.6323,
+      "step": 1364
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.685538141034441e-05,
+      "loss": 1.7004,
+      "step": 1365
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.684870151338095e-05,
+      "loss": 1.6358,
+      "step": 1366
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.684201475996585e-05,
+      "loss": 1.7178,
+      "step": 1367
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.683532115107774e-05,
+      "loss": 1.6889,
+      "step": 1368
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.682862068769624e-05,
+      "loss": 1.5296,
+      "step": 1369
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.682191337080202e-05,
+      "loss": 1.5766,
+      "step": 1370
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.681519920137667e-05,
+      "loss": 1.5967,
+      "step": 1371
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.680847818040287e-05,
+      "loss": 1.6752,
+      "step": 1372
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.680175030886423e-05,
+      "loss": 1.7257,
+      "step": 1373
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.679501558774542e-05,
+      "loss": 1.6311,
+      "step": 1374
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.678827401803206e-05,
+      "loss": 1.583,
+      "step": 1375
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.678152560071082e-05,
+      "loss": 1.6601,
+      "step": 1376
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.677477033676934e-05,
+      "loss": 1.5467,
+      "step": 1377
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.676800822719629e-05,
+      "loss": 1.7424,
+      "step": 1378
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.67612392729813e-05,
+      "loss": 1.6182,
+      "step": 1379
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.675446347511506e-05,
+      "loss": 1.4666,
+      "step": 1380
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.67476808345892e-05,
+      "loss": 1.7103,
+      "step": 1381
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.674089135239639e-05,
+      "loss": 1.6386,
+      "step": 1382
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.673409502953029e-05,
+      "loss": 1.5822,
+      "step": 1383
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.672729186698557e-05,
+      "loss": 1.6779,
+      "step": 1384
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.672048186575789e-05,
+      "loss": 1.4983,
+      "step": 1385
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.671366502684393e-05,
+      "loss": 1.5975,
+      "step": 1386
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.670684135124131e-05,
+      "loss": 1.5894,
+      "step": 1387
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.670001083994875e-05,
+      "loss": 1.4553,
+      "step": 1388
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.669317349396589e-05,
+      "loss": 1.4554,
+      "step": 1389
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.668632931429338e-05,
+      "loss": 1.6541,
+      "step": 1390
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.667947830193292e-05,
+      "loss": 1.7196,
+      "step": 1391
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.667262045788716e-05,
+      "loss": 1.6761,
+      "step": 1392
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.666575578315977e-05,
+      "loss": 1.685,
+      "step": 1393
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.66588842787554e-05,
+      "loss": 1.5685,
+      "step": 1394
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.665200594567975e-05,
+      "loss": 1.4986,
+      "step": 1395
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.664512078493946e-05,
+      "loss": 1.5559,
+      "step": 1396
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.66382287975422e-05,
+      "loss": 1.6983,
+      "step": 1397
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.663132998449663e-05,
+      "loss": 1.7286,
+      "step": 1398
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.662442434681241e-05,
+      "loss": 1.8595,
+      "step": 1399
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.661751188550023e-05,
+      "loss": 1.6927,
+      "step": 1400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.661059260157171e-05,
+      "loss": 1.5721,
+      "step": 1401
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.660366649603953e-05,
+      "loss": 1.5526,
+      "step": 1402
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.659673356991734e-05,
+      "loss": 1.5603,
+      "step": 1403
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.65897938242198e-05,
+      "loss": 1.6768,
+      "step": 1404
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.658284725996256e-05,
+      "loss": 1.6087,
+      "step": 1405
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.657589387816229e-05,
+      "loss": 1.5965,
+      "step": 1406
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.65689336798366e-05,
+      "loss": 1.6279,
+      "step": 1407
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.656196666600417e-05,
+      "loss": 1.6119,
+      "step": 1408
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.655499283768461e-05,
+      "loss": 1.6594,
+      "step": 1409
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.65480121958986e-05,
+      "loss": 1.6169,
+      "step": 1410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.654102474166776e-05,
+      "loss": 1.6599,
+      "step": 1411
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.653403047601473e-05,
+      "loss": 1.6778,
+      "step": 1412
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.652702939996313e-05,
+      "loss": 1.7242,
+      "step": 1413
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.65200215145376e-05,
+      "loss": 1.5526,
+      "step": 1414
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.651300682076376e-05,
+      "loss": 1.6678,
+      "step": 1415
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.650598531966826e-05,
+      "loss": 1.6307,
+      "step": 1416
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.649895701227867e-05,
+      "loss": 1.5524,
+      "step": 1417
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.649192189962362e-05,
+      "loss": 1.6838,
+      "step": 1418
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.648487998273274e-05,
+      "loss": 1.6632,
+      "step": 1419
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.647783126263664e-05,
+      "loss": 1.5973,
+      "step": 1420
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.64707757403669e-05,
+      "loss": 1.5458,
+      "step": 1421
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.64637134169561e-05,
+      "loss": 1.607,
+      "step": 1422
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.645664429343788e-05,
+      "loss": 1.6772,
+      "step": 1423
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.64495683708468e-05,
+      "loss": 1.5565,
+      "step": 1424
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.644248565021845e-05,
+      "loss": 1.5613,
+      "step": 1425
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.643539613258941e-05,
+      "loss": 1.7476,
+      "step": 1426
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.642829981899723e-05,
+      "loss": 1.6301,
+      "step": 1427
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.642119671048051e-05,
+      "loss": 1.4848,
+      "step": 1428
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.64140868080788e-05,
+      "loss": 1.5895,
+      "step": 1429
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.640697011283263e-05,
+      "loss": 1.6816,
+      "step": 1430
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.639984662578361e-05,
+      "loss": 1.4818,
+      "step": 1431
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.639271634797422e-05,
+      "loss": 1.568,
+      "step": 1432
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.638557928044805e-05,
+      "loss": 1.5213,
+      "step": 1433
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.637843542424959e-05,
+      "loss": 1.606,
+      "step": 1434
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.637128478042438e-05,
+      "loss": 1.6555,
+      "step": 1435
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.636412735001896e-05,
+      "loss": 1.583,
+      "step": 1436
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.635696313408083e-05,
+      "loss": 1.2705,
+      "step": 1437
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.634979213365847e-05,
+      "loss": 1.5452,
+      "step": 1438
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.634261434980138e-05,
+      "loss": 1.6957,
+      "step": 1439
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.63354297835601e-05,
+      "loss": 1.7012,
+      "step": 1440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.632823843598607e-05,
+      "loss": 1.7038,
+      "step": 1441
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.632104030813178e-05,
+      "loss": 1.6075,
+      "step": 1442
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.63138354010507e-05,
+      "loss": 1.5904,
+      "step": 1443
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.630662371579728e-05,
+      "loss": 1.6799,
+      "step": 1444
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.629940525342697e-05,
+      "loss": 1.5336,
+      "step": 1445
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.629218001499623e-05,
+      "loss": 1.5634,
+      "step": 1446
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.628494800156249e-05,
+      "loss": 1.5073,
+      "step": 1447
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.627770921418418e-05,
+      "loss": 1.7181,
+      "step": 1448
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.627046365392069e-05,
+      "loss": 1.6304,
+      "step": 1449
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.626321132183249e-05,
+      "loss": 1.7013,
+      "step": 1450
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.625595221898092e-05,
+      "loss": 1.6572,
+      "step": 1451
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.624868634642841e-05,
+      "loss": 1.5864,
+      "step": 1452
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.624141370523831e-05,
+      "loss": 1.703,
+      "step": 1453
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.623413429647504e-05,
+      "loss": 1.5635,
+      "step": 1454
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.622684812120393e-05,
+      "loss": 1.6279,
+      "step": 1455
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.621955518049134e-05,
+      "loss": 1.7375,
+      "step": 1456
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.62122554754046e-05,
+      "loss": 1.4978,
+      "step": 1457
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.62049490070121e-05,
+      "loss": 1.7917,
+      "step": 1458
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.619763577638309e-05,
+      "loss": 1.6668,
+      "step": 1459
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.619031578458793e-05,
+      "loss": 1.6689,
+      "step": 1460
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.618298903269791e-05,
+      "loss": 1.5817,
+      "step": 1461
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.617565552178531e-05,
+      "loss": 1.5685,
+      "step": 1462
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.616831525292345e-05,
+      "loss": 1.6046,
+      "step": 1463
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.616096822718656e-05,
+      "loss": 1.6079,
+      "step": 1464
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.615361444564993e-05,
+      "loss": 1.5687,
+      "step": 1465
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.614625390938978e-05,
+      "loss": 1.5163,
+      "step": 1466
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.613888661948338e-05,
+      "loss": 1.6491,
+      "step": 1467
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.613151257700895e-05,
+      "loss": 1.6514,
+      "step": 1468
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.612413178304566e-05,
+      "loss": 1.5881,
+      "step": 1469
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.611674423867377e-05,
+      "loss": 1.6325,
+      "step": 1470
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.610934994497443e-05,
+      "loss": 1.6122,
+      "step": 1471
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.610194890302982e-05,
+      "loss": 1.7194,
+      "step": 1472
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.609454111392312e-05,
+      "loss": 1.5255,
+      "step": 1473
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.60871265787385e-05,
+      "loss": 1.7359,
+      "step": 1474
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.607970529856106e-05,
+      "loss": 1.558,
+      "step": 1475
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.607227727447696e-05,
+      "loss": 1.6231,
+      "step": 1476
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.606484250757329e-05,
+      "loss": 1.6291,
+      "step": 1477
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.605740099893816e-05,
+      "loss": 1.5252,
+      "step": 1478
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.604995274966069e-05,
+      "loss": 1.4787,
+      "step": 1479
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.60424977608309e-05,
+      "loss": 1.2098,
+      "step": 1480
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.603503603353987e-05,
+      "loss": 1.5763,
+      "step": 1481
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.602756756887965e-05,
+      "loss": 1.6365,
+      "step": 1482
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.602009236794328e-05,
+      "loss": 1.6186,
+      "step": 1483
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.601261043182477e-05,
+      "loss": 1.4322,
+      "step": 1484
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.600512176161914e-05,
+      "loss": 1.3571,
+      "step": 1485
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.599762635842236e-05,
+      "loss": 1.6119,
+      "step": 1486
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.59901242233314e-05,
+      "loss": 1.51,
+      "step": 1487
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.598261535744424e-05,
+      "loss": 1.6723,
+      "step": 1488
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.597509976185983e-05,
+      "loss": 1.6147,
+      "step": 1489
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.596757743767807e-05,
+      "loss": 1.6888,
+      "step": 1490
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.59600483859999e-05,
+      "loss": 1.6815,
+      "step": 1491
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.595251260792721e-05,
+      "loss": 1.439,
+      "step": 1492
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.59449701045629e-05,
+      "loss": 1.5024,
+      "step": 1493
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.593742087701083e-05,
+      "loss": 1.2304,
+      "step": 1494
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.592986492637584e-05,
+      "loss": 1.5983,
+      "step": 1495
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.592230225376377e-05,
+      "loss": 1.4782,
+      "step": 1496
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.591473286028147e-05,
+      "loss": 1.5767,
+      "step": 1497
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.59071567470367e-05,
+      "loss": 1.6626,
+      "step": 1498
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.589957391513826e-05,
+      "loss": 1.6201,
+      "step": 1499
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.589198436569595e-05,
+      "loss": 1.6516,
+      "step": 1500
+    },
+    {
+      "epoch": 0.17,
+      "eval_loss": 1.6439121961593628,
+      "eval_runtime": 4349.54,
+      "eval_samples_per_second": 2.814,
+      "eval_steps_per_second": 0.352,
+      "step": 1500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.58843880998205e-05,
+      "loss": 1.6714,
+      "step": 1501
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.587678511862365e-05,
+      "loss": 1.4784,
+      "step": 1502
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.586917542321812e-05,
+      "loss": 1.6485,
+      "step": 1503
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.58615590147176e-05,
+      "loss": 1.6559,
+      "step": 1504
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.585393589423679e-05,
+      "loss": 1.6756,
+      "step": 1505
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.584630606289136e-05,
+      "loss": 1.6327,
+      "step": 1506
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.583866952179795e-05,
+      "loss": 1.5767,
+      "step": 1507
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.583102627207418e-05,
+      "loss": 1.6421,
+      "step": 1508
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.582337631483871e-05,
+      "loss": 1.5391,
+      "step": 1509
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.581571965121108e-05,
+      "loss": 1.5531,
+      "step": 1510
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.580805628231189e-05,
+      "loss": 1.6123,
+      "step": 1511
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.58003862092627e-05,
+      "loss": 1.6823,
+      "step": 1512
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.579270943318605e-05,
+      "loss": 1.6648,
+      "step": 1513
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.578502595520544e-05,
+      "loss": 1.7396,
+      "step": 1514
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.57773357764454e-05,
+      "loss": 1.638,
+      "step": 1515
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.576963889803139e-05,
+      "loss": 1.5863,
+      "step": 1516
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.576193532108987e-05,
+      "loss": 1.6301,
+      "step": 1517
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.575422504674832e-05,
+      "loss": 1.6745,
+      "step": 1518
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.574650807613509e-05,
+      "loss": 1.698,
+      "step": 1519
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.573878441037967e-05,
+      "loss": 1.6438,
+      "step": 1520
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.573105405061237e-05,
+      "loss": 1.6556,
+      "step": 1521
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.572331699796459e-05,
+      "loss": 1.5875,
+      "step": 1522
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.571557325356865e-05,
+      "loss": 1.7114,
+      "step": 1523
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.570782281855789e-05,
+      "loss": 1.6739,
+      "step": 1524
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.57000656940666e-05,
+      "loss": 1.6094,
+      "step": 1525
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.569230188123006e-05,
+      "loss": 1.5379,
+      "step": 1526
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.568453138118454e-05,
+      "loss": 1.4603,
+      "step": 1527
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.567675419506725e-05,
+      "loss": 1.6977,
+      "step": 1528
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.566897032401644e-05,
+      "loss": 1.5499,
+      "step": 1529
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.566117976917126e-05,
+      "loss": 1.5698,
+      "step": 1530
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.565338253167193e-05,
+      "loss": 1.4632,
+      "step": 1531
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.564557861265956e-05,
+      "loss": 1.6092,
+      "step": 1532
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.563776801327631e-05,
+      "loss": 1.5548,
+      "step": 1533
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.562995073466526e-05,
+      "loss": 1.5463,
+      "step": 1534
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.562212677797049e-05,
+      "loss": 1.4529,
+      "step": 1535
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.561429614433706e-05,
+      "loss": 1.5003,
+      "step": 1536
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.560645883491104e-05,
+      "loss": 1.3016,
+      "step": 1537
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.559861485083943e-05,
+      "loss": 1.5677,
+      "step": 1538
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.55907641932702e-05,
+      "loss": 1.4766,
+      "step": 1539
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.558290686335234e-05,
+      "loss": 1.5778,
+      "step": 1540
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.557504286223578e-05,
+      "loss": 1.6078,
+      "step": 1541
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.556717219107145e-05,
+      "loss": 1.5506,
+      "step": 1542
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.555929485101124e-05,
+      "loss": 1.4303,
+      "step": 1543
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.555141084320804e-05,
+      "loss": 1.5734,
+      "step": 1544
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.554352016881567e-05,
+      "loss": 1.4864,
+      "step": 1545
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.553562282898898e-05,
+      "loss": 1.7216,
+      "step": 1546
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.552771882488377e-05,
+      "loss": 1.5608,
+      "step": 1547
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.551980815765678e-05,
+      "loss": 1.5024,
+      "step": 1548
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.55118908284658e-05,
+      "loss": 1.6475,
+      "step": 1549
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.550396683846954e-05,
+      "loss": 1.7335,
+      "step": 1550
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.54960361888277e-05,
+      "loss": 1.6845,
+      "step": 1551
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.548809888070095e-05,
+      "loss": 1.7218,
+      "step": 1552
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.548015491525097e-05,
+      "loss": 1.617,
+      "step": 1553
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.547220429364034e-05,
+      "loss": 1.5859,
+      "step": 1554
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.546424701703269e-05,
+      "loss": 1.5998,
+      "step": 1555
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.545628308659258e-05,
+      "loss": 1.6613,
+      "step": 1556
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.544831250348556e-05,
+      "loss": 1.6806,
+      "step": 1557
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.544033526887814e-05,
+      "loss": 1.635,
+      "step": 1558
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.54323513839378e-05,
+      "loss": 1.6568,
+      "step": 1559
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.542436084983303e-05,
+      "loss": 1.5108,
+      "step": 1560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.541636366773329e-05,
+      "loss": 1.5927,
+      "step": 1561
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.540835983880894e-05,
+      "loss": 1.6278,
+      "step": 1562
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.540034936423141e-05,
+      "loss": 1.6853,
+      "step": 1563
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.539233224517304e-05,
+      "loss": 1.5918,
+      "step": 1564
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.538430848280712e-05,
+      "loss": 1.7526,
+      "step": 1565
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.537627807830802e-05,
+      "loss": 1.5057,
+      "step": 1566
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.536824103285099e-05,
+      "loss": 1.4209,
+      "step": 1567
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.536019734761226e-05,
+      "loss": 1.6456,
+      "step": 1568
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.535214702376907e-05,
+      "loss": 1.5783,
+      "step": 1569
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.534409006249959e-05,
+      "loss": 1.5966,
+      "step": 1570
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.533602646498298e-05,
+      "loss": 1.6684,
+      "step": 1571
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.53279562323994e-05,
+      "loss": 1.8019,
+      "step": 1572
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.531987936592994e-05,
+      "loss": 1.6745,
+      "step": 1573
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.531179586675665e-05,
+      "loss": 1.6119,
+      "step": 1574
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.530370573606261e-05,
+      "loss": 1.6421,
+      "step": 1575
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.529560897503182e-05,
+      "loss": 1.6881,
+      "step": 1576
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.528750558484927e-05,
+      "loss": 1.5743,
+      "step": 1577
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.527939556670089e-05,
+      "loss": 1.7027,
+      "step": 1578
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.527127892177366e-05,
+      "loss": 1.6342,
+      "step": 1579
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.526315565125542e-05,
+      "loss": 1.5503,
+      "step": 1580
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.525502575633508e-05,
+      "loss": 1.6355,
+      "step": 1581
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.524688923820244e-05,
+      "loss": 1.5466,
+      "step": 1582
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.523874609804833e-05,
+      "loss": 1.543,
+      "step": 1583
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.523059633706451e-05,
+      "loss": 1.4428,
+      "step": 1584
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.52224399564437e-05,
+      "loss": 1.546,
+      "step": 1585
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.521427695737968e-05,
+      "loss": 1.3498,
+      "step": 1586
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.520610734106707e-05,
+      "loss": 1.5932,
+      "step": 1587
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.519793110870154e-05,
+      "loss": 1.5662,
+      "step": 1588
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.51897482614797e-05,
+      "loss": 1.7507,
+      "step": 1589
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.518155880059913e-05,
+      "loss": 1.6261,
+      "step": 1590
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.517336272725839e-05,
+      "loss": 1.557,
+      "step": 1591
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.516516004265701e-05,
+      "loss": 1.5885,
+      "step": 1592
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.515695074799546e-05,
+      "loss": 1.5898,
+      "step": 1593
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.514873484447523e-05,
+      "loss": 1.6235,
+      "step": 1594
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.51405123332987e-05,
+      "loss": 1.6347,
+      "step": 1595
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.513228321566927e-05,
+      "loss": 1.5499,
+      "step": 1596
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.512404749279131e-05,
+      "loss": 1.4102,
+      "step": 1597
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.511580516587015e-05,
+      "loss": 1.5332,
+      "step": 1598
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.510755623611204e-05,
+      "loss": 1.6196,
+      "step": 1599
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.50993007047243e-05,
+      "loss": 1.7217,
+      "step": 1600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.509103857291509e-05,
+      "loss": 1.5746,
+      "step": 1601
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.508276984189364e-05,
+      "loss": 1.668,
+      "step": 1602
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.507449451287007e-05,
+      "loss": 1.5718,
+      "step": 1603
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.506621258705553e-05,
+      "loss": 1.4438,
+      "step": 1604
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.505792406566209e-05,
+      "loss": 1.7008,
+      "step": 1605
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.50496289499028e-05,
+      "loss": 1.6191,
+      "step": 1606
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.504132724099168e-05,
+      "loss": 1.6759,
+      "step": 1607
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.503301894014372e-05,
+      "loss": 1.577,
+      "step": 1608
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.502470404857487e-05,
+      "loss": 1.5488,
+      "step": 1609
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.501638256750201e-05,
+      "loss": 1.5889,
+      "step": 1610
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.500805449814303e-05,
+      "loss": 1.453,
+      "step": 1611
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.499971984171675e-05,
+      "loss": 1.5023,
+      "step": 1612
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.499137859944303e-05,
+      "loss": 1.6093,
+      "step": 1613
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.498303077254258e-05,
+      "loss": 1.6005,
+      "step": 1614
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.497467636223713e-05,
+      "loss": 1.7097,
+      "step": 1615
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.496631536974942e-05,
+      "loss": 1.5685,
+      "step": 1616
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.495794779630308e-05,
+      "loss": 1.5833,
+      "step": 1617
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.494957364312271e-05,
+      "loss": 1.5767,
+      "step": 1618
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.494119291143392e-05,
+      "loss": 1.6225,
+      "step": 1619
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.493280560246324e-05,
+      "loss": 1.5895,
+      "step": 1620
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.49244117174382e-05,
+      "loss": 1.6579,
+      "step": 1621
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.491601125758725e-05,
+      "loss": 1.6149,
+      "step": 1622
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.490760422413984e-05,
+      "loss": 1.6356,
+      "step": 1623
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.489919061832635e-05,
+      "loss": 1.4747,
+      "step": 1624
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.489077044137816e-05,
+      "loss": 1.6415,
+      "step": 1625
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.488234369452753e-05,
+      "loss": 1.6253,
+      "step": 1626
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.487391037900781e-05,
+      "loss": 1.6299,
+      "step": 1627
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.486547049605322e-05,
+      "loss": 1.6305,
+      "step": 1628
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.485702404689893e-05,
+      "loss": 1.6338,
+      "step": 1629
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.484857103278116e-05,
+      "loss": 1.659,
+      "step": 1630
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.484011145493699e-05,
+      "loss": 1.4947,
+      "step": 1631
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.483164531460453e-05,
+      "loss": 1.5661,
+      "step": 1632
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.48231726130228e-05,
+      "loss": 1.7011,
+      "step": 1633
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.481469335143183e-05,
+      "loss": 1.6416,
+      "step": 1634
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.48062075310726e-05,
+      "loss": 1.4251,
+      "step": 1635
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.479771515318699e-05,
+      "loss": 1.6646,
+      "step": 1636
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.47892162190179e-05,
+      "loss": 1.6368,
+      "step": 1637
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.478071072980919e-05,
+      "loss": 1.6166,
+      "step": 1638
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.477219868680566e-05,
+      "loss": 1.4668,
+      "step": 1639
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.476368009125309e-05,
+      "loss": 1.5764,
+      "step": 1640
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.475515494439817e-05,
+      "loss": 1.5841,
+      "step": 1641
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.47466232474886e-05,
+      "loss": 1.5056,
+      "step": 1642
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.473808500177304e-05,
+      "loss": 1.5819,
+      "step": 1643
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.472954020850103e-05,
+      "loss": 1.6851,
+      "step": 1644
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.472098886892318e-05,
+      "loss": 1.5675,
+      "step": 1645
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.471243098429099e-05,
+      "loss": 1.522,
+      "step": 1646
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.470386655585693e-05,
+      "loss": 1.6143,
+      "step": 1647
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.469529558487442e-05,
+      "loss": 1.568,
+      "step": 1648
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.468671807259786e-05,
+      "loss": 1.5671,
+      "step": 1649
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.46781340202826e-05,
+      "loss": 1.6596,
+      "step": 1650
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.466954342918494e-05,
+      "loss": 1.4605,
+      "step": 1651
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.466094630056213e-05,
+      "loss": 1.5303,
+      "step": 1652
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.465234263567239e-05,
+      "loss": 1.6209,
+      "step": 1653
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.46437324357749e-05,
+      "loss": 1.6363,
+      "step": 1654
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.463511570212977e-05,
+      "loss": 1.521,
+      "step": 1655
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.462649243599813e-05,
+      "loss": 1.5366,
+      "step": 1656
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.461786263864196e-05,
+      "loss": 1.5532,
+      "step": 1657
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.46092263113243e-05,
+      "loss": 1.6021,
+      "step": 1658
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.460058345530909e-05,
+      "loss": 1.7048,
+      "step": 1659
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.459193407186124e-05,
+      "loss": 1.572,
+      "step": 1660
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.458327816224663e-05,
+      "loss": 1.5842,
+      "step": 1661
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.457461572773204e-05,
+      "loss": 1.5886,
+      "step": 1662
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.456594676958528e-05,
+      "loss": 1.5546,
+      "step": 1663
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.455727128907506e-05,
+      "loss": 1.5556,
+      "step": 1664
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.454858928747108e-05,
+      "loss": 1.6725,
+      "step": 1665
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.453990076604396e-05,
+      "loss": 1.6577,
+      "step": 1666
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.453120572606529e-05,
+      "loss": 1.5928,
+      "step": 1667
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.452250416880764e-05,
+      "loss": 1.4823,
+      "step": 1668
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.451379609554447e-05,
+      "loss": 1.6587,
+      "step": 1669
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.450508150755029e-05,
+      "loss": 1.5771,
+      "step": 1670
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.449636040610044e-05,
+      "loss": 1.6467,
+      "step": 1671
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.448763279247135e-05,
+      "loss": 1.6068,
+      "step": 1672
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.447889866794026e-05,
+      "loss": 1.6466,
+      "step": 1673
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.447015803378549e-05,
+      "loss": 1.6138,
+      "step": 1674
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.446141089128626e-05,
+      "loss": 1.5773,
+      "step": 1675
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.44526572417227e-05,
+      "loss": 1.6426,
+      "step": 1676
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.444389708637597e-05,
+      "loss": 1.686,
+      "step": 1677
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.443513042652811e-05,
+      "loss": 1.7147,
+      "step": 1678
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.44263572634622e-05,
+      "loss": 1.49,
+      "step": 1679
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.441757759846217e-05,
+      "loss": 1.638,
+      "step": 1680
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.440879143281298e-05,
+      "loss": 1.5984,
+      "step": 1681
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.439999876780051e-05,
+      "loss": 1.4381,
+      "step": 1682
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.439119960471157e-05,
+      "loss": 1.5036,
+      "step": 1683
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.438239394483397e-05,
+      "loss": 1.5423,
+      "step": 1684
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.437358178945643e-05,
+      "loss": 1.6065,
+      "step": 1685
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.436476313986863e-05,
+      "loss": 1.3642,
+      "step": 1686
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.435593799736125e-05,
+      "loss": 1.5226,
+      "step": 1687
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.434710636322584e-05,
+      "loss": 1.4318,
+      "step": 1688
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.433826823875493e-05,
+      "loss": 1.6577,
+      "step": 1689
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.432942362524202e-05,
+      "loss": 1.5995,
+      "step": 1690
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.432057252398154e-05,
+      "loss": 1.487,
+      "step": 1691
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.431171493626889e-05,
+      "loss": 1.6431,
+      "step": 1692
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.43028508634004e-05,
+      "loss": 1.6135,
+      "step": 1693
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.429398030667336e-05,
+      "loss": 1.4981,
+      "step": 1694
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.428510326738597e-05,
+      "loss": 1.465,
+      "step": 1695
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.427621974683746e-05,
+      "loss": 1.6302,
+      "step": 1696
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.426732974632792e-05,
+      "loss": 1.7001,
+      "step": 1697
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.425843326715847e-05,
+      "loss": 1.4926,
+      "step": 1698
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.424953031063109e-05,
+      "loss": 1.7597,
+      "step": 1699
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.424062087804878e-05,
+      "loss": 1.693,
+      "step": 1700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.423170497071546e-05,
+      "loss": 1.508,
+      "step": 1701
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.4222782589936e-05,
+      "loss": 1.7507,
+      "step": 1702
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.421385373701624e-05,
+      "loss": 1.5483,
+      "step": 1703
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.420491841326292e-05,
+      "loss": 1.568,
+      "step": 1704
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.419597661998374e-05,
+      "loss": 1.6771,
+      "step": 1705
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.418702835848738e-05,
+      "loss": 1.6415,
+      "step": 1706
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.417807363008344e-05,
+      "loss": 1.6318,
+      "step": 1707
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.416911243608248e-05,
+      "loss": 1.5299,
+      "step": 1708
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.4160144777796e-05,
+      "loss": 1.6709,
+      "step": 1709
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.415117065653643e-05,
+      "loss": 1.6211,
+      "step": 1710
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.414219007361716e-05,
+      "loss": 1.5981,
+      "step": 1711
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.413320303035254e-05,
+      "loss": 1.6127,
+      "step": 1712
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.412420952805782e-05,
+      "loss": 1.529,
+      "step": 1713
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.411520956804926e-05,
+      "loss": 1.6223,
+      "step": 1714
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.410620315164405e-05,
+      "loss": 1.6113,
+      "step": 1715
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.409719028016026e-05,
+      "loss": 1.6089,
+      "step": 1716
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.408817095491698e-05,
+      "loss": 1.6042,
+      "step": 1717
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.407914517723419e-05,
+      "loss": 1.5653,
+      "step": 1718
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.407011294843286e-05,
+      "loss": 1.5734,
+      "step": 1719
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.406107426983488e-05,
+      "loss": 1.6204,
+      "step": 1720
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.40520291427631e-05,
+      "loss": 1.6609,
+      "step": 1721
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.404297756854129e-05,
+      "loss": 1.6813,
+      "step": 1722
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.403391954849417e-05,
+      "loss": 1.5482,
+      "step": 1723
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.402485508394742e-05,
+      "loss": 1.6346,
+      "step": 1724
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.401578417622766e-05,
+      "loss": 1.6112,
+      "step": 1725
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.400670682666242e-05,
+      "loss": 1.5998,
+      "step": 1726
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.399762303658021e-05,
+      "loss": 1.6174,
+      "step": 1727
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.398853280731049e-05,
+      "loss": 1.6006,
+      "step": 1728
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.397943614018361e-05,
+      "loss": 1.3272,
+      "step": 1729
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.397033303653091e-05,
+      "loss": 1.5399,
+      "step": 1730
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.396122349768468e-05,
+      "loss": 1.6138,
+      "step": 1731
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.39521075249781e-05,
+      "loss": 1.6066,
+      "step": 1732
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.39429851197453e-05,
+      "loss": 1.4683,
+      "step": 1733
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.393385628332143e-05,
+      "loss": 1.1139,
+      "step": 1734
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.392472101704249e-05,
+      "loss": 1.4526,
+      "step": 1735
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.391557932224545e-05,
+      "loss": 1.6505,
+      "step": 1736
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.390643120026823e-05,
+      "loss": 1.6997,
+      "step": 1737
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.38972766524497e-05,
+      "loss": 1.4938,
+      "step": 1738
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.388811568012964e-05,
+      "loss": 1.6384,
+      "step": 1739
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.387894828464879e-05,
+      "loss": 1.5022,
+      "step": 1740
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.386977446734883e-05,
+      "loss": 1.7306,
+      "step": 1741
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.386059422957238e-05,
+      "loss": 1.5912,
+      "step": 1742
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.385140757266299e-05,
+      "loss": 1.6467,
+      "step": 1743
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.384221449796515e-05,
+      "loss": 1.5745,
+      "step": 1744
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.38330150068243e-05,
+      "loss": 1.5015,
+      "step": 1745
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.382380910058681e-05,
+      "loss": 1.724,
+      "step": 1746
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.38145967806e-05,
+      "loss": 1.6366,
+      "step": 1747
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.380537804821215e-05,
+      "loss": 1.6276,
+      "step": 1748
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.379615290477237e-05,
+      "loss": 1.7825,
+      "step": 1749
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.378692135163088e-05,
+      "loss": 1.6079,
+      "step": 1750
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.377768339013867e-05,
+      "loss": 1.5349,
+      "step": 1751
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.376843902164779e-05,
+      "loss": 1.4339,
+      "step": 1752
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.375918824751116e-05,
+      "loss": 1.6071,
+      "step": 1753
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.374993106908267e-05,
+      "loss": 1.6153,
+      "step": 1754
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.374066748771715e-05,
+      "loss": 1.6436,
+      "step": 1755
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.373139750477034e-05,
+      "loss": 1.6311,
+      "step": 1756
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.372212112159891e-05,
+      "loss": 1.6068,
+      "step": 1757
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.371283833956052e-05,
+      "loss": 1.5967,
+      "step": 1758
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.370354916001372e-05,
+      "loss": 1.5979,
+      "step": 1759
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.369425358431799e-05,
+      "loss": 1.6427,
+      "step": 1760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.368495161383379e-05,
+      "loss": 1.6272,
+      "step": 1761
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.367564324992251e-05,
+      "loss": 1.5099,
+      "step": 1762
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.366632849394642e-05,
+      "loss": 1.5941,
+      "step": 1763
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.365700734726879e-05,
+      "loss": 1.4367,
+      "step": 1764
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.364767981125378e-05,
+      "loss": 1.4775,
+      "step": 1765
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.36383458872665e-05,
+      "loss": 1.5853,
+      "step": 1766
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.362900557667303e-05,
+      "loss": 1.7506,
+      "step": 1767
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.361965888084032e-05,
+      "loss": 1.7116,
+      "step": 1768
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.36103058011363e-05,
+      "loss": 1.6953,
+      "step": 1769
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.360094633892981e-05,
+      "loss": 1.5271,
+      "step": 1770
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.359158049559068e-05,
+      "loss": 1.5136,
+      "step": 1771
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.358220827248959e-05,
+      "loss": 1.6036,
+      "step": 1772
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.357282967099818e-05,
+      "loss": 1.515,
+      "step": 1773
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.356344469248908e-05,
+      "loss": 1.6468,
+      "step": 1774
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.355405333833577e-05,
+      "loss": 1.6302,
+      "step": 1775
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.354465560991275e-05,
+      "loss": 1.6223,
+      "step": 1776
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.353525150859535e-05,
+      "loss": 1.6347,
+      "step": 1777
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.352584103575994e-05,
+      "loss": 1.7429,
+      "step": 1778
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.351642419278374e-05,
+      "loss": 1.7043,
+      "step": 1779
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.350700098104497e-05,
+      "loss": 1.6001,
+      "step": 1780
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.349757140192269e-05,
+      "loss": 1.2592,
+      "step": 1781
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.3488135456797e-05,
+      "loss": 1.6733,
+      "step": 1782
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.347869314704883e-05,
+      "loss": 1.6513,
+      "step": 1783
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.346924447406013e-05,
+      "loss": 1.6297,
+      "step": 1784
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.345978943921375e-05,
+      "loss": 1.6668,
+      "step": 1785
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.345032804389345e-05,
+      "loss": 1.6349,
+      "step": 1786
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.34408602894839e-05,
+      "loss": 1.4704,
+      "step": 1787
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.343138617737077e-05,
+      "loss": 1.5588,
+      "step": 1788
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.342190570894062e-05,
+      "loss": 1.4783,
+      "step": 1789
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.341241888558097e-05,
+      "loss": 1.5099,
+      "step": 1790
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.34029257086802e-05,
+      "loss": 1.683,
+      "step": 1791
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.33934261796277e-05,
+      "loss": 1.6885,
+      "step": 1792
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.338392029981373e-05,
+      "loss": 1.5135,
+      "step": 1793
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.337440807062951e-05,
+      "loss": 1.4994,
+      "step": 1794
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.336488949346721e-05,
+      "loss": 1.3849,
+      "step": 1795
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.335536456971988e-05,
+      "loss": 1.5084,
+      "step": 1796
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.334583330078152e-05,
+      "loss": 1.5919,
+      "step": 1797
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.333629568804707e-05,
+      "loss": 1.5345,
+      "step": 1798
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.332675173291238e-05,
+      "loss": 1.4617,
+      "step": 1799
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.331720143677424e-05,
+      "loss": 1.6592,
+      "step": 1800
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.330764480103038e-05,
+      "loss": 1.6224,
+      "step": 1801
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.329808182707943e-05,
+      "loss": 1.605,
+      "step": 1802
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.328851251632096e-05,
+      "loss": 1.6394,
+      "step": 1803
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.327893687015547e-05,
+      "loss": 1.6394,
+      "step": 1804
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.326935488998439e-05,
+      "loss": 1.6577,
+      "step": 1805
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.325976657721004e-05,
+      "loss": 1.567,
+      "step": 1806
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.325017193323576e-05,
+      "loss": 1.635,
+      "step": 1807
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.32405709594657e-05,
+      "loss": 1.6449,
+      "step": 1808
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.323096365730502e-05,
+      "loss": 1.6435,
+      "step": 1809
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.322135002815978e-05,
+      "loss": 1.6497,
+      "step": 1810
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.321173007343695e-05,
+      "loss": 1.6536,
+      "step": 1811
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.320210379454446e-05,
+      "loss": 1.6432,
+      "step": 1812
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.319247119289111e-05,
+      "loss": 1.5835,
+      "step": 1813
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.318283226988669e-05,
+      "loss": 1.618,
+      "step": 1814
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.317318702694189e-05,
+      "loss": 1.7328,
+      "step": 1815
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.31635354654683e-05,
+      "loss": 1.6271,
+      "step": 1816
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.315387758687847e-05,
+      "loss": 1.7174,
+      "step": 1817
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.314421339258586e-05,
+      "loss": 1.6199,
+      "step": 1818
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.313454288400485e-05,
+      "loss": 1.668,
+      "step": 1819
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.312486606255073e-05,
+      "loss": 1.5604,
+      "step": 1820
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.311518292963977e-05,
+      "loss": 1.5865,
+      "step": 1821
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.31054934866891e-05,
+      "loss": 1.634,
+      "step": 1822
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.309579773511683e-05,
+      "loss": 1.538,
+      "step": 1823
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.308609567634193e-05,
+      "loss": 1.6913,
+      "step": 1824
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.307638731178432e-05,
+      "loss": 1.7259,
+      "step": 1825
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.306667264286489e-05,
+      "loss": 1.6389,
+      "step": 1826
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.305695167100537e-05,
+      "loss": 1.5613,
+      "step": 1827
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.304722439762847e-05,
+      "loss": 1.6491,
+      "step": 1828
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.303749082415783e-05,
+      "loss": 1.685,
+      "step": 1829
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.302775095201794e-05,
+      "loss": 1.7105,
+      "step": 1830
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.30180047826343e-05,
+      "loss": 1.7074,
+      "step": 1831
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.300825231743329e-05,
+      "loss": 1.4704,
+      "step": 1832
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.299849355784219e-05,
+      "loss": 1.5592,
+      "step": 1833
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.298872850528924e-05,
+      "loss": 1.6333,
+      "step": 1834
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.297895716120357e-05,
+      "loss": 1.4719,
+      "step": 1835
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.296917952701527e-05,
+      "loss": 1.606,
+      "step": 1836
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.295939560415533e-05,
+      "loss": 1.6142,
+      "step": 1837
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.294960539405561e-05,
+      "loss": 1.6185,
+      "step": 1838
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.2939808898149e-05,
+      "loss": 1.5692,
+      "step": 1839
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.29300061178692e-05,
+      "loss": 1.5496,
+      "step": 1840
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.29201970546509e-05,
+      "loss": 1.5028,
+      "step": 1841
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.291038170992969e-05,
+      "loss": 1.631,
+      "step": 1842
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.290056008514205e-05,
+      "loss": 1.5304,
+      "step": 1843
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.289073218172543e-05,
+      "loss": 1.4099,
+      "step": 1844
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.288089800111816e-05,
+      "loss": 1.3558,
+      "step": 1845
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.287105754475952e-05,
+      "loss": 1.6214,
+      "step": 1846
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.286121081408967e-05,
+      "loss": 1.7352,
+      "step": 1847
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.285135781054973e-05,
+      "loss": 1.6236,
+      "step": 1848
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.28414985355817e-05,
+      "loss": 1.6529,
+      "step": 1849
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.283163299062852e-05,
+      "loss": 1.7619,
+      "step": 1850
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.282176117713404e-05,
+      "loss": 1.7334,
+      "step": 1851
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.281188309654305e-05,
+      "loss": 1.4643,
+      "step": 1852
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.28019987503012e-05,
+      "loss": 1.452,
+      "step": 1853
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.279210813985513e-05,
+      "loss": 1.4752,
+      "step": 1854
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.278221126665235e-05,
+      "loss": 1.6766,
+      "step": 1855
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.277230813214129e-05,
+      "loss": 1.6488,
+      "step": 1856
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.276239873777132e-05,
+      "loss": 1.5876,
+      "step": 1857
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.275248308499269e-05,
+      "loss": 1.7294,
+      "step": 1858
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.274256117525659e-05,
+      "loss": 1.6021,
+      "step": 1859
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.273263301001515e-05,
+      "loss": 1.7077,
+      "step": 1860
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.272269859072136e-05,
+      "loss": 1.6483,
+      "step": 1861
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.271275791882914e-05,
+      "loss": 1.627,
+      "step": 1862
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.270281099579338e-05,
+      "loss": 1.6402,
+      "step": 1863
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.26928578230698e-05,
+      "loss": 1.6373,
+      "step": 1864
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.26828984021151e-05,
+      "loss": 1.4891,
+      "step": 1865
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.267293273438689e-05,
+      "loss": 1.6191,
+      "step": 1866
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.266296082134365e-05,
+      "loss": 1.7333,
+      "step": 1867
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.265298266444479e-05,
+      "loss": 1.6441,
+      "step": 1868
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.264299826515066e-05,
+      "loss": 1.6653,
+      "step": 1869
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.263300762492252e-05,
+      "loss": 1.5359,
+      "step": 1870
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.26230107452225e-05,
+      "loss": 1.5912,
+      "step": 1871
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.261300762751371e-05,
+      "loss": 1.5156,
+      "step": 1872
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.260299827326011e-05,
+      "loss": 1.7088,
+      "step": 1873
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.259298268392662e-05,
+      "loss": 1.664,
+      "step": 1874
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.258296086097903e-05,
+      "loss": 1.595,
+      "step": 1875
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.257293280588408e-05,
+      "loss": 1.7385,
+      "step": 1876
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.25628985201094e-05,
+      "loss": 1.5868,
+      "step": 1877
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.255285800512354e-05,
+      "loss": 1.6841,
+      "step": 1878
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.254281126239596e-05,
+      "loss": 1.622,
+      "step": 1879
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.253275829339704e-05,
+      "loss": 1.6164,
+      "step": 1880
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.252269909959806e-05,
+      "loss": 1.6703,
+      "step": 1881
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.251263368247119e-05,
+      "loss": 1.6745,
+      "step": 1882
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.250256204348957e-05,
+      "loss": 1.4541,
+      "step": 1883
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.24924841841272e-05,
+      "loss": 1.5208,
+      "step": 1884
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.248240010585901e-05,
+      "loss": 1.5859,
+      "step": 1885
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.247230981016082e-05,
+      "loss": 1.6472,
+      "step": 1886
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.24622132985094e-05,
+      "loss": 1.7629,
+      "step": 1887
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.245211057238236e-05,
+      "loss": 1.3103,
+      "step": 1888
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.244200163325834e-05,
+      "loss": 1.4802,
+      "step": 1889
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.243188648261676e-05,
+      "loss": 1.6049,
+      "step": 1890
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.2421765121938e-05,
+      "loss": 1.6761,
+      "step": 1891
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.241163755270339e-05,
+      "loss": 1.6449,
+      "step": 1892
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.24015037763951e-05,
+      "loss": 1.3679,
+      "step": 1893
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.239136379449624e-05,
+      "loss": 1.6377,
+      "step": 1894
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.238121760849086e-05,
+      "loss": 1.6706,
+      "step": 1895
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.237106521986383e-05,
+      "loss": 1.5468,
+      "step": 1896
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.236090663010103e-05,
+      "loss": 1.6595,
+      "step": 1897
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.235074184068919e-05,
+      "loss": 1.648,
+      "step": 1898
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.234057085311596e-05,
+      "loss": 1.6672,
+      "step": 1899
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.233039366886987e-05,
+      "loss": 1.5517,
+      "step": 1900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.232021028944043e-05,
+      "loss": 1.6905,
+      "step": 1901
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.231002071631795e-05,
+      "loss": 1.5508,
+      "step": 1902
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.229982495099374e-05,
+      "loss": 1.6424,
+      "step": 1903
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.228962299495999e-05,
+      "loss": 1.6107,
+      "step": 1904
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.227941484970976e-05,
+      "loss": 1.6776,
+      "step": 1905
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.226920051673709e-05,
+      "loss": 1.6529,
+      "step": 1906
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.225897999753683e-05,
+      "loss": 1.5583,
+      "step": 1907
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.224875329360478e-05,
+      "loss": 1.6153,
+      "step": 1908
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.22385204064377e-05,
+      "loss": 1.6779,
+      "step": 1909
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.222828133753316e-05,
+      "loss": 1.5562,
+      "step": 1910
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.221803608838971e-05,
+      "loss": 1.6027,
+      "step": 1911
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.220778466050674e-05,
+      "loss": 1.5791,
+      "step": 1912
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.219752705538463e-05,
+      "loss": 1.6304,
+      "step": 1913
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.218726327452455e-05,
+      "loss": 1.6367,
+      "step": 1914
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.217699331942868e-05,
+      "loss": 1.5799,
+      "step": 1915
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.216671719160004e-05,
+      "loss": 1.7253,
+      "step": 1916
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.215643489254258e-05,
+      "loss": 1.5965,
+      "step": 1917
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.214614642376113e-05,
+      "loss": 1.5632,
+      "step": 1918
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.213585178676147e-05,
+      "loss": 1.597,
+      "step": 1919
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.212555098305023e-05,
+      "loss": 1.5072,
+      "step": 1920
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.211524401413496e-05,
+      "loss": 1.7013,
+      "step": 1921
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.210493088152414e-05,
+      "loss": 1.5079,
+      "step": 1922
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.209461158672709e-05,
+      "loss": 1.6546,
+      "step": 1923
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.208428613125413e-05,
+      "loss": 1.5825,
+      "step": 1924
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.207395451661637e-05,
+      "loss": 1.7352,
+      "step": 1925
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.206361674432589e-05,
+      "loss": 1.6466,
+      "step": 1926
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.205327281589566e-05,
+      "loss": 1.6462,
+      "step": 1927
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.204292273283952e-05,
+      "loss": 1.6637,
+      "step": 1928
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.203256649667229e-05,
+      "loss": 1.5941,
+      "step": 1929
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.202220410890958e-05,
+      "loss": 1.6268,
+      "step": 1930
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.2011835571068e-05,
+      "loss": 1.3896,
+      "step": 1931
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.200146088466499e-05,
+      "loss": 1.6436,
+      "step": 1932
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.199108005121893e-05,
+      "loss": 1.8396,
+      "step": 1933
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.198069307224908e-05,
+      "loss": 1.3659,
+      "step": 1934
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.197029994927561e-05,
+      "loss": 1.4792,
+      "step": 1935
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.195990068381959e-05,
+      "loss": 1.5764,
+      "step": 1936
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.194949527740298e-05,
+      "loss": 1.5208,
+      "step": 1937
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.193908373154862e-05,
+      "loss": 1.6335,
+      "step": 1938
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.192866604778032e-05,
+      "loss": 1.5416,
+      "step": 1939
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.191824222762272e-05,
+      "loss": 1.6897,
+      "step": 1940
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.190781227260137e-05,
+      "loss": 1.6412,
+      "step": 1941
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.189737618424272e-05,
+      "loss": 1.7224,
+      "step": 1942
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.188693396407412e-05,
+      "loss": 1.5964,
+      "step": 1943
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.187648561362387e-05,
+      "loss": 1.566,
+      "step": 1944
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.186603113442105e-05,
+      "loss": 1.4068,
+      "step": 1945
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.185557052799576e-05,
+      "loss": 1.2991,
+      "step": 1946
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.18451037958789e-05,
+      "loss": 1.6991,
+      "step": 1947
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.183463093960234e-05,
+      "loss": 1.6607,
+      "step": 1948
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.18241519606988e-05,
+      "loss": 1.6079,
+      "step": 1949
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.181366686070192e-05,
+      "loss": 1.7494,
+      "step": 1950
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.180317564114621e-05,
+      "loss": 1.682,
+      "step": 1951
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.17926783035671e-05,
+      "loss": 1.6165,
+      "step": 1952
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.178217484950092e-05,
+      "loss": 1.5153,
+      "step": 1953
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.177166528048488e-05,
+      "loss": 1.4798,
+      "step": 1954
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.176114959805706e-05,
+      "loss": 1.5633,
+      "step": 1955
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.17506278037565e-05,
+      "loss": 1.6239,
+      "step": 1956
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.174009989912307e-05,
+      "loss": 1.6417,
+      "step": 1957
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.172956588569758e-05,
+      "loss": 1.5221,
+      "step": 1958
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.171902576502169e-05,
+      "loss": 1.6352,
+      "step": 1959
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.170847953863802e-05,
+      "loss": 1.5941,
+      "step": 1960
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.169792720809001e-05,
+      "loss": 1.6508,
+      "step": 1961
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.168736877492202e-05,
+      "loss": 1.5777,
+      "step": 1962
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.167680424067934e-05,
+      "loss": 1.6124,
+      "step": 1963
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.16662336069081e-05,
+      "loss": 1.4952,
+      "step": 1964
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.165565687515533e-05,
+      "loss": 1.6113,
+      "step": 1965
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.164507404696903e-05,
+      "loss": 1.6424,
+      "step": 1966
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.163448512389796e-05,
+      "loss": 1.4426,
+      "step": 1967
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.162389010749186e-05,
+      "loss": 1.6134,
+      "step": 1968
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.161328899930137e-05,
+      "loss": 1.6619,
+      "step": 1969
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.160268180087797e-05,
+      "loss": 1.6909,
+      "step": 1970
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.159206851377407e-05,
+      "loss": 1.5979,
+      "step": 1971
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.158144913954294e-05,
+      "loss": 1.6319,
+      "step": 1972
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.157082367973879e-05,
+      "loss": 1.6069,
+      "step": 1973
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.156019213591664e-05,
+      "loss": 1.6582,
+      "step": 1974
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.154955450963248e-05,
+      "loss": 1.5267,
+      "step": 1975
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.153891080244317e-05,
+      "loss": 1.6452,
+      "step": 1976
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.152826101590642e-05,
+      "loss": 1.6661,
+      "step": 1977
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.151760515158089e-05,
+      "loss": 1.6201,
+      "step": 1978
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.150694321102606e-05,
+      "loss": 1.5046,
+      "step": 1979
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.149627519580238e-05,
+      "loss": 1.6225,
+      "step": 1980
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.148560110747111e-05,
+      "loss": 1.5066,
+      "step": 1981
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.147492094759447e-05,
+      "loss": 1.6342,
+      "step": 1982
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.146423471773551e-05,
+      "loss": 1.5068,
+      "step": 1983
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.14535424194582e-05,
+      "loss": 1.6619,
+      "step": 1984
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.144284405432739e-05,
+      "loss": 1.6465,
+      "step": 1985
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.143213962390882e-05,
+      "loss": 1.5579,
+      "step": 1986
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.142142912976912e-05,
+      "loss": 1.4997,
+      "step": 1987
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.141071257347579e-05,
+      "loss": 1.6216,
+      "step": 1988
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.139998995659724e-05,
+      "loss": 1.6895,
+      "step": 1989
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.138926128070275e-05,
+      "loss": 1.6132,
+      "step": 1990
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.137852654736252e-05,
+      "loss": 1.7302,
+      "step": 1991
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.136778575814759e-05,
+      "loss": 1.5162,
+      "step": 1992
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.135703891462991e-05,
+      "loss": 1.5398,
+      "step": 1993
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.134628601838231e-05,
+      "loss": 1.4154,
+      "step": 1994
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.133552707097852e-05,
+      "loss": 1.4754,
+      "step": 1995
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.132476207399314e-05,
+      "loss": 1.4432,
+      "step": 1996
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.131399102900167e-05,
+      "loss": 1.5176,
+      "step": 1997
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.130321393758045e-05,
+      "loss": 1.5256,
+      "step": 1998
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.12924308013068e-05,
+      "loss": 1.7559,
+      "step": 1999
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.128164162175881e-05,
+      "loss": 1.6137,
+      "step": 2000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.127084640051552e-05,
+      "loss": 1.5982,
+      "step": 2001
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.126004513915687e-05,
+      "loss": 1.6259,
+      "step": 2002
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.124923783926364e-05,
+      "loss": 1.6524,
+      "step": 2003
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.123842450241753e-05,
+      "loss": 1.6223,
+      "step": 2004
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.122760513020107e-05,
+      "loss": 1.603,
+      "step": 2005
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.121677972419773e-05,
+      "loss": 1.6476,
+      "step": 2006
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.120594828599185e-05,
+      "loss": 1.5563,
+      "step": 2007
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.119511081716863e-05,
+      "loss": 1.6236,
+      "step": 2008
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.118426731931416e-05,
+      "loss": 1.5902,
+      "step": 2009
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.117341779401544e-05,
+      "loss": 1.5638,
+      "step": 2010
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.116256224286032e-05,
+      "loss": 1.6365,
+      "step": 2011
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.115170066743754e-05,
+      "loss": 1.5414,
+      "step": 2012
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.114083306933672e-05,
+      "loss": 1.5906,
+      "step": 2013
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.11299594501484e-05,
+      "loss": 1.5751,
+      "step": 2014
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.111907981146392e-05,
+      "loss": 1.5433,
+      "step": 2015
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.110819415487559e-05,
+      "loss": 1.6454,
+      "step": 2016
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.109730248197654e-05,
+      "loss": 1.6788,
+      "step": 2017
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.108640479436079e-05,
+      "loss": 1.6451,
+      "step": 2018
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.107550109362328e-05,
+      "loss": 1.6085,
+      "step": 2019
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.106459138135977e-05,
+      "loss": 1.6409,
+      "step": 2020
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.105367565916693e-05,
+      "loss": 1.5935,
+      "step": 2021
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.104275392864235e-05,
+      "loss": 1.6565,
+      "step": 2022
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.103182619138441e-05,
+      "loss": 1.5995,
+      "step": 2023
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.102089244899244e-05,
+      "loss": 1.5717,
+      "step": 2024
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.100995270306662e-05,
+      "loss": 1.6027,
+      "step": 2025
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.099900695520803e-05,
+      "loss": 1.506,
+      "step": 2026
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.098805520701861e-05,
+      "loss": 1.5909,
+      "step": 2027
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.097709746010115e-05,
+      "loss": 1.6227,
+      "step": 2028
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.09661337160594e-05,
+      "loss": 1.5235,
+      "step": 2029
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.09551639764979e-05,
+      "loss": 1.5737,
+      "step": 2030
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.094418824302214e-05,
+      "loss": 1.3414,
+      "step": 2031
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.093320651723843e-05,
+      "loss": 1.4304,
+      "step": 2032
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.092221880075396e-05,
+      "loss": 1.6586,
+      "step": 2033
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.091122509517684e-05,
+      "loss": 1.5779,
+      "step": 2034
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.090022540211604e-05,
+      "loss": 1.5057,
+      "step": 2035
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.088921972318138e-05,
+      "loss": 1.6731,
+      "step": 2036
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.08782080599836e-05,
+      "loss": 1.6109,
+      "step": 2037
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.086719041413424e-05,
+      "loss": 1.6609,
+      "step": 2038
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.085616678724584e-05,
+      "loss": 1.6233,
+      "step": 2039
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.084513718093168e-05,
+      "loss": 1.548,
+      "step": 2040
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.0834101596806e-05,
+      "loss": 1.6346,
+      "step": 2041
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.08230600364839e-05,
+      "loss": 1.5896,
+      "step": 2042
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.081201250158134e-05,
+      "loss": 1.6157,
+      "step": 2043
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.080095899371515e-05,
+      "loss": 1.4762,
+      "step": 2044
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.078989951450305e-05,
+      "loss": 1.5484,
+      "step": 2045
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.077883406556364e-05,
+      "loss": 1.4192,
+      "step": 2046
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.076776264851637e-05,
+      "loss": 1.6141,
+      "step": 2047
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.075668526498158e-05,
+      "loss": 1.5946,
+      "step": 2048
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.074560191658047e-05,
+      "loss": 1.6072,
+      "step": 2049
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.073451260493515e-05,
+      "loss": 1.6946,
+      "step": 2050
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.072341733166854e-05,
+      "loss": 1.6165,
+      "step": 2051
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.07123160984045e-05,
+      "loss": 1.573,
+      "step": 2052
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.07012089067677e-05,
+      "loss": 1.6772,
+      "step": 2053
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.069009575838371e-05,
+      "loss": 1.4928,
+      "step": 2054
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.067897665487903e-05,
+      "loss": 1.5977,
+      "step": 2055
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.066785159788091e-05,
+      "loss": 1.6517,
+      "step": 2056
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.065672058901755e-05,
+      "loss": 1.6535,
+      "step": 2057
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.064558362991805e-05,
+      "loss": 1.6179,
+      "step": 2058
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.063444072221226e-05,
+      "loss": 1.5441,
+      "step": 2059
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.062329186753104e-05,
+      "loss": 1.6566,
+      "step": 2060
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.061213706750606e-05,
+      "loss": 1.6221,
+      "step": 2061
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.060097632376984e-05,
+      "loss": 1.5442,
+      "step": 2062
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.05898096379558e-05,
+      "loss": 1.46,
+      "step": 2063
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.057863701169819e-05,
+      "loss": 1.3983,
+      "step": 2064
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.05674584466322e-05,
+      "loss": 1.4577,
+      "step": 2065
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.05562739443938e-05,
+      "loss": 1.476,
+      "step": 2066
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.054508350661992e-05,
+      "loss": 1.5647,
+      "step": 2067
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.053388713494827e-05,
+      "loss": 1.5162,
+      "step": 2068
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.052268483101755e-05,
+      "loss": 1.6189,
+      "step": 2069
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.051147659646717e-05,
+      "loss": 1.6529,
+      "step": 2070
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.050026243293753e-05,
+      "loss": 1.5199,
+      "step": 2071
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.048904234206985e-05,
+      "loss": 1.6548,
+      "step": 2072
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.047781632550622e-05,
+      "loss": 1.6049,
+      "step": 2073
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.046658438488962e-05,
+      "loss": 1.6833,
+      "step": 2074
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.045534652186386e-05,
+      "loss": 1.6182,
+      "step": 2075
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.044410273807364e-05,
+      "loss": 1.6096,
+      "step": 2076
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.043285303516454e-05,
+      "loss": 1.6761,
+      "step": 2077
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.042159741478295e-05,
+      "loss": 1.6269,
+      "step": 2078
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.04103358785762e-05,
+      "loss": 1.4712,
+      "step": 2079
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.03990684281924e-05,
+      "loss": 1.3792,
+      "step": 2080
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.038779506528066e-05,
+      "loss": 1.5523,
+      "step": 2081
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.037651579149082e-05,
+      "loss": 1.7664,
+      "step": 2082
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.036523060847362e-05,
+      "loss": 1.6293,
+      "step": 2083
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.035393951788072e-05,
+      "loss": 1.6035,
+      "step": 2084
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.03426425213646e-05,
+      "loss": 1.5156,
+      "step": 2085
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.033133962057858e-05,
+      "loss": 1.237,
+      "step": 2086
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.03200308171769e-05,
+      "loss": 1.5548,
+      "step": 2087
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.030871611281461e-05,
+      "loss": 1.605,
+      "step": 2088
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.029739550914771e-05,
+      "loss": 1.5853,
+      "step": 2089
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.028606900783295e-05,
+      "loss": 1.4755,
+      "step": 2090
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.027473661052801e-05,
+      "loss": 1.4618,
+      "step": 2091
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.026339831889143e-05,
+      "loss": 1.5655,
+      "step": 2092
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.025205413458262e-05,
+      "loss": 1.7503,
+      "step": 2093
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.02407040592618e-05,
+      "loss": 1.4464,
+      "step": 2094
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.022934809459012e-05,
+      "loss": 1.4989,
+      "step": 2095
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.021798624222952e-05,
+      "loss": 1.5262,
+      "step": 2096
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.02066185038429e-05,
+      "loss": 1.6275,
+      "step": 2097
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.019524488109392e-05,
+      "loss": 1.6958,
+      "step": 2098
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.018386537564716e-05,
+      "loss": 1.4849,
+      "step": 2099
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.017247998916804e-05,
+      "loss": 1.6781,
+      "step": 2100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.016108872332286e-05,
+      "loss": 1.4063,
+      "step": 2101
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.014969157977872e-05,
+      "loss": 1.621,
+      "step": 2102
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.01382885602037e-05,
+      "loss": 1.5644,
+      "step": 2103
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.012687966626662e-05,
+      "loss": 1.6376,
+      "step": 2104
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.011546489963722e-05,
+      "loss": 1.6295,
+      "step": 2105
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.010404426198609e-05,
+      "loss": 1.6663,
+      "step": 2106
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.009261775498465e-05,
+      "loss": 1.5835,
+      "step": 2107
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.008118538030523e-05,
+      "loss": 1.5371,
+      "step": 2108
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.006974713962099e-05,
+      "loss": 1.6527,
+      "step": 2109
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.005830303460594e-05,
+      "loss": 1.5971,
+      "step": 2110
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.004685306693497e-05,
+      "loss": 1.5782,
+      "step": 2111
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.003539723828383e-05,
+      "loss": 1.5175,
+      "step": 2112
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.002393555032909e-05,
+      "loss": 1.561,
+      "step": 2113
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.00124680047482e-05,
+      "loss": 1.5174,
+      "step": 2114
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.000099460321952e-05,
+      "loss": 1.609,
+      "step": 2115
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.998951534742216e-05,
+      "loss": 1.569,
+      "step": 2116
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.997803023903617e-05,
+      "loss": 1.4965,
+      "step": 2117
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.996653927974243e-05,
+      "loss": 1.4809,
+      "step": 2118
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.995504247122268e-05,
+      "loss": 1.572,
+      "step": 2119
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.994353981515951e-05,
+      "loss": 1.4795,
+      "step": 2120
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.993203131323635e-05,
+      "loss": 1.4481,
+      "step": 2121
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.992051696713755e-05,
+      "loss": 1.5539,
+      "step": 2122
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.990899677854823e-05,
+      "loss": 1.5974,
+      "step": 2123
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.989747074915442e-05,
+      "loss": 1.6486,
+      "step": 2124
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.988593888064299e-05,
+      "loss": 1.5328,
+      "step": 2125
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.987440117470165e-05,
+      "loss": 1.5325,
+      "step": 2126
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.986285763301899e-05,
+      "loss": 1.583,
+      "step": 2127
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.985130825728445e-05,
+      "loss": 1.5079,
+      "step": 2128
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.983975304918832e-05,
+      "loss": 1.5883,
+      "step": 2129
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.982819201042172e-05,
+      "loss": 1.5539,
+      "step": 2130
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.981662514267666e-05,
+      "loss": 1.6057,
+      "step": 2131
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.980505244764598e-05,
+      "loss": 1.6686,
+      "step": 2132
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.979347392702337e-05,
+      "loss": 1.3948,
+      "step": 2133
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.978188958250342e-05,
+      "loss": 1.3904,
+      "step": 2134
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.977029941578148e-05,
+      "loss": 1.4618,
+      "step": 2135
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.975870342855385e-05,
+      "loss": 1.6688,
+      "step": 2136
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.974710162251762e-05,
+      "loss": 1.6437,
+      "step": 2137
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.973549399937074e-05,
+      "loss": 1.5794,
+      "step": 2138
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.972388056081205e-05,
+      "loss": 1.5861,
+      "step": 2139
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.971226130854118e-05,
+      "loss": 1.5281,
+      "step": 2140
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.970063624425868e-05,
+      "loss": 1.4307,
+      "step": 2141
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.968900536966588e-05,
+      "loss": 1.6187,
+      "step": 2142
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.967736868646501e-05,
+      "loss": 1.3983,
+      "step": 2143
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.966572619635913e-05,
+      "loss": 1.585,
+      "step": 2144
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.965407790105215e-05,
+      "loss": 1.6419,
+      "step": 2145
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.964242380224885e-05,
+      "loss": 1.5428,
+      "step": 2146
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.963076390165483e-05,
+      "loss": 1.6717,
+      "step": 2147
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.961909820097656e-05,
+      "loss": 1.548,
+      "step": 2148
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.960742670192135e-05,
+      "loss": 1.5706,
+      "step": 2149
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.959574940619735e-05,
+      "loss": 1.6867,
+      "step": 2150
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.958406631551357e-05,
+      "loss": 1.5268,
+      "step": 2151
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.957237743157986e-05,
+      "loss": 1.6904,
+      "step": 2152
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.956068275610696e-05,
+      "loss": 1.6984,
+      "step": 2153
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.95489822908064e-05,
+      "loss": 1.6462,
+      "step": 2154
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.953727603739054e-05,
+      "loss": 1.4066,
+      "step": 2155
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.95255639975727e-05,
+      "loss": 1.5357,
+      "step": 2156
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.951384617306693e-05,
+      "loss": 1.564,
+      "step": 2157
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.950212256558817e-05,
+      "loss": 1.645,
+      "step": 2158
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.94903931768522e-05,
+      "loss": 1.6243,
+      "step": 2159
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.947865800857567e-05,
+      "loss": 1.6805,
+      "step": 2160
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.946691706247605e-05,
+      "loss": 1.599,
+      "step": 2161
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.945517034027168e-05,
+      "loss": 1.5898,
+      "step": 2162
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.944341784368168e-05,
+      "loss": 1.604,
+      "step": 2163
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.943165957442614e-05,
+      "loss": 1.715,
+      "step": 2164
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.941989553422586e-05,
+      "loss": 1.6069,
+      "step": 2165
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.940812572480255e-05,
+      "loss": 1.5952,
+      "step": 2166
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.939635014787877e-05,
+      "loss": 1.6132,
+      "step": 2167
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.938456880517791e-05,
+      "loss": 1.5391,
+      "step": 2168
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.937278169842422e-05,
+      "loss": 1.6083,
+      "step": 2169
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.936098882934275e-05,
+      "loss": 1.4471,
+      "step": 2170
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.934919019965944e-05,
+      "loss": 1.6886,
+      "step": 2171
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.933738581110106e-05,
+      "loss": 1.5385,
+      "step": 2172
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.932557566539521e-05,
+      "loss": 1.5395,
+      "step": 2173
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.931375976427034e-05,
+      "loss": 1.6025,
+      "step": 2174
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.930193810945575e-05,
+      "loss": 1.595,
+      "step": 2175
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.929011070268157e-05,
+      "loss": 1.606,
+      "step": 2176
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.927827754567878e-05,
+      "loss": 1.6629,
+      "step": 2177
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.92664386401792e-05,
+      "loss": 1.6119,
+      "step": 2178
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.925459398791549e-05,
+      "loss": 1.615,
+      "step": 2179
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.924274359062115e-05,
+      "loss": 1.4251,
+      "step": 2180
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.92308874500305e-05,
+      "loss": 1.5379,
+      "step": 2181
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.921902556787877e-05,
+      "loss": 1.6228,
+      "step": 2182
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.920715794590195e-05,
+      "loss": 1.5667,
+      "step": 2183
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.919528458583693e-05,
+      "loss": 1.4675,
+      "step": 2184
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.918340548942137e-05,
+      "loss": 1.5532,
+      "step": 2185
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.917152065839384e-05,
+      "loss": 1.5524,
+      "step": 2186
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.915963009449373e-05,
+      "loss": 1.6022,
+      "step": 2187
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.914773379946124e-05,
+      "loss": 1.6177,
+      "step": 2188
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.913583177503744e-05,
+      "loss": 1.4263,
+      "step": 2189
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.912392402296422e-05,
+      "loss": 1.638,
+      "step": 2190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.911201054498433e-05,
+      "loss": 1.5888,
+      "step": 2191
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.910009134284131e-05,
+      "loss": 1.5696,
+      "step": 2192
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.908816641827963e-05,
+      "loss": 1.3888,
+      "step": 2193
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.907623577304449e-05,
+      "loss": 1.6871,
+      "step": 2194
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.9064299408882e-05,
+      "loss": 1.4844,
+      "step": 2195
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.905235732753907e-05,
+      "loss": 1.6119,
+      "step": 2196
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.904040953076348e-05,
+      "loss": 1.4362,
+      "step": 2197
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.902845602030381e-05,
+      "loss": 1.4757,
+      "step": 2198
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.90164967979095e-05,
+      "loss": 1.5472,
+      "step": 2199
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.900453186533082e-05,
+      "loss": 1.6726,
+      "step": 2200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.899256122431885e-05,
+      "loss": 1.5938,
+      "step": 2201
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.898058487662557e-05,
+      "loss": 1.7595,
+      "step": 2202
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.896860282400373e-05,
+      "loss": 1.6871,
+      "step": 2203
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.895661506820696e-05,
+      "loss": 1.5842,
+      "step": 2204
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.894462161098967e-05,
+      "loss": 1.6817,
+      "step": 2205
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.893262245410719e-05,
+      "loss": 1.6892,
+      "step": 2206
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.892061759931558e-05,
+      "loss": 1.6403,
+      "step": 2207
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.890860704837181e-05,
+      "loss": 1.5814,
+      "step": 2208
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.889659080303369e-05,
+      "loss": 1.6573,
+      "step": 2209
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.888456886505978e-05,
+      "loss": 1.5419,
+      "step": 2210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.887254123620958e-05,
+      "loss": 1.3807,
+      "step": 2211
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.886050791824331e-05,
+      "loss": 1.5376,
+      "step": 2212
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.884846891292215e-05,
+      "loss": 1.5325,
+      "step": 2213
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.883642422200801e-05,
+      "loss": 1.6573,
+      "step": 2214
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.882437384726368e-05,
+      "loss": 1.5753,
+      "step": 2215
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.881231779045274e-05,
+      "loss": 1.5537,
+      "step": 2216
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.880025605333968e-05,
+      "loss": 1.6361,
+      "step": 2217
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.878818863768972e-05,
+      "loss": 1.5324,
+      "step": 2218
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.8776115545269e-05,
+      "loss": 1.4019,
+      "step": 2219
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.876403677784443e-05,
+      "loss": 1.6732,
+      "step": 2220
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.875195233718381e-05,
+      "loss": 1.662,
+      "step": 2221
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.873986222505571e-05,
+      "loss": 1.6324,
+      "step": 2222
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.872776644322954e-05,
+      "loss": 1.5951,
+      "step": 2223
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.871566499347559e-05,
+      "loss": 1.6403,
+      "step": 2224
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.87035578775649e-05,
+      "loss": 1.6506,
+      "step": 2225
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.869144509726943e-05,
+      "loss": 1.4073,
+      "step": 2226
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.867932665436192e-05,
+      "loss": 1.6158,
+      "step": 2227
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.86672025506159e-05,
+      "loss": 1.6303,
+      "step": 2228
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.865507278780581e-05,
+      "loss": 1.4887,
+      "step": 2229
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.864293736770685e-05,
+      "loss": 1.4505,
+      "step": 2230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.863079629209508e-05,
+      "loss": 1.6157,
+      "step": 2231
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.861864956274742e-05,
+      "loss": 1.6274,
+      "step": 2232
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.860649718144154e-05,
+      "loss": 1.4844,
+      "step": 2233
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.859433914995599e-05,
+      "loss": 1.5635,
+      "step": 2234
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.858217547007013e-05,
+      "loss": 1.2655,
+      "step": 2235
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.857000614356416e-05,
+      "loss": 1.6062,
+      "step": 2236
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.855783117221912e-05,
+      "loss": 1.5,
+      "step": 2237
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.854565055781682e-05,
+      "loss": 1.5801,
+      "step": 2238
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.853346430213994e-05,
+      "loss": 1.5671,
+      "step": 2239
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.852127240697198e-05,
+      "loss": 1.6844,
+      "step": 2240
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.850907487409725e-05,
+      "loss": 1.514,
+      "step": 2241
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.849687170530092e-05,
+      "loss": 1.5991,
+      "step": 2242
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.848466290236896e-05,
+      "loss": 1.4802,
+      "step": 2243
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.847244846708813e-05,
+      "loss": 1.6907,
+      "step": 2244
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.84602284012461e-05,
+      "loss": 1.4677,
+      "step": 2245
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.844800270663126e-05,
+      "loss": 1.5627,
+      "step": 2246
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.843577138503292e-05,
+      "loss": 1.6187,
+      "step": 2247
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.842353443824116e-05,
+      "loss": 1.6382,
+      "step": 2248
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.841129186804688e-05,
+      "loss": 1.6699,
+      "step": 2249
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.839904367624184e-05,
+      "loss": 1.6475,
+      "step": 2250
+    },
+    {
+      "epoch": 0.26,
+      "eval_loss": 1.6286474466323853,
+      "eval_runtime": 4348.7031,
+      "eval_samples_per_second": 2.814,
+      "eval_steps_per_second": 0.352,
+      "step": 2250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.83867898646186e-05,
+      "loss": 1.5459,
+      "step": 2251
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.837453043497053e-05,
+      "loss": 1.6688,
+      "step": 2252
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.836226538909183e-05,
+      "loss": 1.6312,
+      "step": 2253
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.834999472877754e-05,
+      "loss": 1.5708,
+      "step": 2254
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.83377184558235e-05,
+      "loss": 1.6298,
+      "step": 2255
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.832543657202639e-05,
+      "loss": 1.5527,
+      "step": 2256
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.831314907918371e-05,
+      "loss": 1.4176,
+      "step": 2257
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.830085597909375e-05,
+      "loss": 1.5383,
+      "step": 2258
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.828855727355565e-05,
+      "loss": 1.5771,
+      "step": 2259
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.827625296436936e-05,
+      "loss": 1.5447,
+      "step": 2260
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.826394305333568e-05,
+      "loss": 1.5804,
+      "step": 2261
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.825162754225616e-05,
+      "loss": 1.6187,
+      "step": 2262
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.823930643293326e-05,
+      "loss": 1.4857,
+      "step": 2263
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.822697972717017e-05,
+      "loss": 1.7057,
+      "step": 2264
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.821464742677096e-05,
+      "loss": 1.6305,
+      "step": 2265
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.82023095335405e-05,
+      "loss": 1.6114,
+      "step": 2266
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.818996604928449e-05,
+      "loss": 1.5736,
+      "step": 2267
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.817761697580942e-05,
+      "loss": 1.557,
+      "step": 2268
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.816526231492261e-05,
+      "loss": 1.5177,
+      "step": 2269
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.815290206843223e-05,
+      "loss": 1.7145,
+      "step": 2270
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.814053623814721e-05,
+      "loss": 1.6045,
+      "step": 2271
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.812816482587733e-05,
+      "loss": 1.6436,
+      "step": 2272
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.811578783343322e-05,
+      "loss": 1.7009,
+      "step": 2273
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.810340526262627e-05,
+      "loss": 1.5501,
+      "step": 2274
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.809101711526868e-05,
+      "loss": 1.564,
+      "step": 2275
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.807862339317351e-05,
+      "loss": 1.3244,
+      "step": 2276
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.806622409815463e-05,
+      "loss": 1.6489,
+      "step": 2277
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.805381923202673e-05,
+      "loss": 1.4644,
+      "step": 2278
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.804140879660527e-05,
+      "loss": 1.7241,
+      "step": 2279
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.802899279370658e-05,
+      "loss": 1.5642,
+      "step": 2280
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.801657122514774e-05,
+      "loss": 1.6117,
+      "step": 2281
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.800414409274673e-05,
+      "loss": 1.4563,
+      "step": 2282
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.799171139832227e-05,
+      "loss": 1.4156,
+      "step": 2283
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.797927314369393e-05,
+      "loss": 1.5493,
+      "step": 2284
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.79668293306821e-05,
+      "loss": 1.6805,
+      "step": 2285
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.795437996110797e-05,
+      "loss": 1.4872,
+      "step": 2286
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.794192503679353e-05,
+      "loss": 1.5498,
+      "step": 2287
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.792946455956159e-05,
+      "loss": 1.5269,
+      "step": 2288
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.791699853123578e-05,
+      "loss": 1.6564,
+      "step": 2289
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.790452695364057e-05,
+      "loss": 1.5151,
+      "step": 2290
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.78920498286012e-05,
+      "loss": 1.5897,
+      "step": 2291
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.787956715794371e-05,
+      "loss": 1.6703,
+      "step": 2292
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.786707894349502e-05,
+      "loss": 1.6104,
+      "step": 2293
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.785458518708278e-05,
+      "loss": 1.5157,
+      "step": 2294
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.78420858905355e-05,
+      "loss": 1.5806,
+      "step": 2295
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.782958105568253e-05,
+      "loss": 1.5169,
+      "step": 2296
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.781707068435394e-05,
+      "loss": 1.5362,
+      "step": 2297
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.780455477838068e-05,
+      "loss": 1.5204,
+      "step": 2298
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.779203333959449e-05,
+      "loss": 1.6266,
+      "step": 2299
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.77795063698279e-05,
+      "loss": 1.6369,
+      "step": 2300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.776697387091433e-05,
+      "loss": 1.6282,
+      "step": 2301
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.775443584468788e-05,
+      "loss": 1.4768,
+      "step": 2302
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.774189229298357e-05,
+      "loss": 1.574,
+      "step": 2303
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.77293432176372e-05,
+      "loss": 1.5219,
+      "step": 2304
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.771678862048532e-05,
+      "loss": 1.5769,
+      "step": 2305
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.770422850336537e-05,
+      "loss": 1.6131,
+      "step": 2306
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.769166286811554e-05,
+      "loss": 1.3888,
+      "step": 2307
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.767909171657487e-05,
+      "loss": 1.5095,
+      "step": 2308
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.766651505058317e-05,
+      "loss": 1.6539,
+      "step": 2309
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.765393287198107e-05,
+      "loss": 1.5908,
+      "step": 2310
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.764134518261004e-05,
+      "loss": 1.583,
+      "step": 2311
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.762875198431231e-05,
+      "loss": 1.4994,
+      "step": 2312
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.761615327893091e-05,
+      "loss": 1.662,
+      "step": 2313
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.760354906830973e-05,
+      "loss": 1.5936,
+      "step": 2314
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.759093935429344e-05,
+      "loss": 1.3991,
+      "step": 2315
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.757832413872746e-05,
+      "loss": 1.6661,
+      "step": 2316
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.756570342345813e-05,
+      "loss": 1.5155,
+      "step": 2317
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.75530772103325e-05,
+      "loss": 1.5629,
+      "step": 2318
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.754044550119844e-05,
+      "loss": 1.6567,
+      "step": 2319
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.752780829790466e-05,
+      "loss": 1.4948,
+      "step": 2320
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.751516560230066e-05,
+      "loss": 1.5584,
+      "step": 2321
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.750251741623672e-05,
+      "loss": 1.6497,
+      "step": 2322
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.748986374156395e-05,
+      "loss": 1.6495,
+      "step": 2323
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.747720458013425e-05,
+      "loss": 1.6753,
+      "step": 2324
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.746453993380034e-05,
+      "loss": 1.5388,
+      "step": 2325
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.745186980441571e-05,
+      "loss": 1.5979,
+      "step": 2326
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.743919419383468e-05,
+      "loss": 1.5121,
+      "step": 2327
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.742651310391237e-05,
+      "loss": 1.5785,
+      "step": 2328
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.74138265365047e-05,
+      "loss": 1.551,
+      "step": 2329
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.740113449346838e-05,
+      "loss": 1.466,
+      "step": 2330
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.738843697666092e-05,
+      "loss": 1.6816,
+      "step": 2331
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.737573398794065e-05,
+      "loss": 1.5925,
+      "step": 2332
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.736302552916671e-05,
+      "loss": 1.5268,
+      "step": 2333
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.7350311602199e-05,
+      "loss": 1.595,
+      "step": 2334
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.733759220889823e-05,
+      "loss": 1.2528,
+      "step": 2335
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.732486735112596e-05,
+      "loss": 1.33,
+      "step": 2336
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.731213703074449e-05,
+      "loss": 1.4589,
+      "step": 2337
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.729940124961693e-05,
+      "loss": 1.527,
+      "step": 2338
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.728666000960723e-05,
+      "loss": 1.5494,
+      "step": 2339
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.727391331258008e-05,
+      "loss": 1.3953,
+      "step": 2340
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.726116116040101e-05,
+      "loss": 1.6251,
+      "step": 2341
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.724840355493635e-05,
+      "loss": 1.3943,
+      "step": 2342
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.723564049805322e-05,
+      "loss": 1.3994,
+      "step": 2343
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.72228719916195e-05,
+      "loss": 1.7292,
+      "step": 2344
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.721009803750391e-05,
+      "loss": 1.5074,
+      "step": 2345
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.719731863757596e-05,
+      "loss": 1.5117,
+      "step": 2346
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.718453379370597e-05,
+      "loss": 1.7877,
+      "step": 2347
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.717174350776502e-05,
+      "loss": 1.6312,
+      "step": 2348
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.715894778162503e-05,
+      "loss": 1.6438,
+      "step": 2349
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.714614661715868e-05,
+      "loss": 1.8213,
+      "step": 2350
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.713334001623946e-05,
+      "loss": 1.5042,
+      "step": 2351
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.712052798074167e-05,
+      "loss": 1.5859,
+      "step": 2352
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.710771051254035e-05,
+      "loss": 1.6712,
+      "step": 2353
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.709488761351141e-05,
+      "loss": 1.5142,
+      "step": 2354
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.708205928553153e-05,
+      "loss": 1.6731,
+      "step": 2355
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.706922553047814e-05,
+      "loss": 1.6571,
+      "step": 2356
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.705638635022952e-05,
+      "loss": 1.5225,
+      "step": 2357
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.704354174666474e-05,
+      "loss": 1.5228,
+      "step": 2358
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.703069172166363e-05,
+      "loss": 1.5552,
+      "step": 2359
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.701783627710682e-05,
+      "loss": 1.5299,
+      "step": 2360
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.700497541487575e-05,
+      "loss": 1.6537,
+      "step": 2361
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.699210913685267e-05,
+      "loss": 1.4561,
+      "step": 2362
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.697923744492056e-05,
+      "loss": 1.6473,
+      "step": 2363
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.696636034096327e-05,
+      "loss": 1.6352,
+      "step": 2364
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.695347782686536e-05,
+      "loss": 1.5228,
+      "step": 2365
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.694058990451227e-05,
+      "loss": 1.6566,
+      "step": 2366
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.692769657579016e-05,
+      "loss": 1.609,
+      "step": 2367
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.691479784258603e-05,
+      "loss": 1.547,
+      "step": 2368
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.690189370678762e-05,
+      "loss": 1.6334,
+      "step": 2369
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.688898417028353e-05,
+      "loss": 1.3739,
+      "step": 2370
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.687606923496307e-05,
+      "loss": 1.5474,
+      "step": 2371
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.686314890271641e-05,
+      "loss": 1.6385,
+      "step": 2372
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.685022317543446e-05,
+      "loss": 1.6981,
+      "step": 2373
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.683729205500893e-05,
+      "loss": 1.4068,
+      "step": 2374
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.682435554333237e-05,
+      "loss": 1.617,
+      "step": 2375
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.681141364229806e-05,
+      "loss": 1.6107,
+      "step": 2376
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.679846635380006e-05,
+      "loss": 1.564,
+      "step": 2377
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.678551367973327e-05,
+      "loss": 1.6197,
+      "step": 2378
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.677255562199336e-05,
+      "loss": 1.6191,
+      "step": 2379
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.675959218247678e-05,
+      "loss": 1.516,
+      "step": 2380
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.674662336308074e-05,
+      "loss": 1.5156,
+      "step": 2381
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.673364916570333e-05,
+      "loss": 1.465,
+      "step": 2382
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.672066959224329e-05,
+      "loss": 1.5912,
+      "step": 2383
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.670768464460026e-05,
+      "loss": 1.4358,
+      "step": 2384
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.669469432467461e-05,
+      "loss": 1.4435,
+      "step": 2385
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.668169863436755e-05,
+      "loss": 1.3741,
+      "step": 2386
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.666869757558098e-05,
+      "loss": 1.5602,
+      "step": 2387
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.66556911502177e-05,
+      "loss": 1.4253,
+      "step": 2388
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.664267936018122e-05,
+      "loss": 1.4764,
+      "step": 2389
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.662966220737584e-05,
+      "loss": 1.5198,
+      "step": 2390
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.661663969370668e-05,
+      "loss": 1.6749,
+      "step": 2391
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.660361182107962e-05,
+      "loss": 1.5656,
+      "step": 2392
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.659057859140132e-05,
+      "loss": 1.6143,
+      "step": 2393
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.657754000657926e-05,
+      "loss": 1.5136,
+      "step": 2394
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.656449606852163e-05,
+      "loss": 1.5201,
+      "step": 2395
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.655144677913751e-05,
+      "loss": 1.5355,
+      "step": 2396
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.653839214033666e-05,
+      "loss": 1.6087,
+      "step": 2397
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.652533215402968e-05,
+      "loss": 1.3194,
+      "step": 2398
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.651226682212794e-05,
+      "loss": 1.5665,
+      "step": 2399
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.649919614654359e-05,
+      "loss": 1.5398,
+      "step": 2400
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.648612012918955e-05,
+      "loss": 1.5182,
+      "step": 2401
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.647303877197956e-05,
+      "loss": 1.6191,
+      "step": 2402
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.645995207682811e-05,
+      "loss": 1.5961,
+      "step": 2403
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.644686004565047e-05,
+      "loss": 1.6044,
+      "step": 2404
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.643376268036269e-05,
+      "loss": 1.596,
+      "step": 2405
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.642065998288165e-05,
+      "loss": 1.5903,
+      "step": 2406
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.640755195512493e-05,
+      "loss": 1.6502,
+      "step": 2407
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.639443859901094e-05,
+      "loss": 1.56,
+      "step": 2408
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.638131991645886e-05,
+      "loss": 1.6733,
+      "step": 2409
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.636819590938866e-05,
+      "loss": 1.5946,
+      "step": 2410
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.635506657972106e-05,
+      "loss": 1.6268,
+      "step": 2411
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.63419319293776e-05,
+      "loss": 1.6108,
+      "step": 2412
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.632879196028056e-05,
+      "loss": 1.58,
+      "step": 2413
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.631564667435304e-05,
+      "loss": 1.7549,
+      "step": 2414
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.630249607351885e-05,
+      "loss": 1.6011,
+      "step": 2415
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.628934015970265e-05,
+      "loss": 1.5036,
+      "step": 2416
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.627617893482984e-05,
+      "loss": 1.5748,
+      "step": 2417
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.626301240082662e-05,
+      "loss": 1.4272,
+      "step": 2418
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.624984055961993e-05,
+      "loss": 1.6305,
+      "step": 2419
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.623666341313753e-05,
+      "loss": 1.4796,
+      "step": 2420
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.622348096330792e-05,
+      "loss": 1.5698,
+      "step": 2421
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.62102932120604e-05,
+      "loss": 1.4555,
+      "step": 2422
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.619710016132505e-05,
+      "loss": 1.6317,
+      "step": 2423
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.618390181303268e-05,
+      "loss": 1.5854,
+      "step": 2424
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.617069816911495e-05,
+      "loss": 1.6532,
+      "step": 2425
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.615748923150422e-05,
+      "loss": 1.5009,
+      "step": 2426
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.614427500213368e-05,
+      "loss": 1.5779,
+      "step": 2427
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.613105548293725e-05,
+      "loss": 1.5407,
+      "step": 2428
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.611783067584968e-05,
+      "loss": 1.5995,
+      "step": 2429
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.610460058280644e-05,
+      "loss": 1.4646,
+      "step": 2430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.609136520574379e-05,
+      "loss": 1.5052,
+      "step": 2431
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.607812454659878e-05,
+      "loss": 1.6449,
+      "step": 2432
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.606487860730923e-05,
+      "loss": 1.5248,
+      "step": 2433
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.605162738981369e-05,
+      "loss": 1.635,
+      "step": 2434
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.603837089605156e-05,
+      "loss": 1.3225,
+      "step": 2435
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.602510912796294e-05,
+      "loss": 1.487,
+      "step": 2436
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.601184208748874e-05,
+      "loss": 1.4995,
+      "step": 2437
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.599856977657061e-05,
+      "loss": 1.5991,
+      "step": 2438
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.598529219715103e-05,
+      "loss": 1.6999,
+      "step": 2439
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.597200935117319e-05,
+      "loss": 1.4952,
+      "step": 2440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.595872124058111e-05,
+      "loss": 1.6712,
+      "step": 2441
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.59454278673195e-05,
+      "loss": 1.7363,
+      "step": 2442
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.593212923333389e-05,
+      "loss": 1.4979,
+      "step": 2443
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.591882534057061e-05,
+      "loss": 1.2759,
+      "step": 2444
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.590551619097671e-05,
+      "loss": 1.5868,
+      "step": 2445
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.589220178650001e-05,
+      "loss": 1.4396,
+      "step": 2446
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.587888212908913e-05,
+      "loss": 1.5313,
+      "step": 2447
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.586555722069345e-05,
+      "loss": 1.4991,
+      "step": 2448
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.585222706326308e-05,
+      "loss": 1.4812,
+      "step": 2449
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.583889165874897e-05,
+      "loss": 1.6353,
+      "step": 2450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.582555100910276e-05,
+      "loss": 1.5598,
+      "step": 2451
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.581220511627692e-05,
+      "loss": 1.6578,
+      "step": 2452
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.579885398222466e-05,
+      "loss": 1.6065,
+      "step": 2453
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.578549760889995e-05,
+      "loss": 1.6152,
+      "step": 2454
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.577213599825753e-05,
+      "loss": 1.5364,
+      "step": 2455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.575876915225294e-05,
+      "loss": 1.4352,
+      "step": 2456
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.574539707284243e-05,
+      "loss": 1.4567,
+      "step": 2457
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.573201976198308e-05,
+      "loss": 1.4442,
+      "step": 2458
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.571863722163265e-05,
+      "loss": 1.6706,
+      "step": 2459
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.570524945374976e-05,
+      "loss": 1.733,
+      "step": 2460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.569185646029373e-05,
+      "loss": 1.636,
+      "step": 2461
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.567845824322469e-05,
+      "loss": 1.6171,
+      "step": 2462
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.566505480450347e-05,
+      "loss": 1.5412,
+      "step": 2463
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.565164614609172e-05,
+      "loss": 1.5823,
+      "step": 2464
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.563823226995186e-05,
+      "loss": 1.5677,
+      "step": 2465
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.562481317804703e-05,
+      "loss": 1.6727,
+      "step": 2466
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.561138887234117e-05,
+      "loss": 1.6363,
+      "step": 2467
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.559795935479896e-05,
+      "loss": 1.5377,
+      "step": 2468
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.558452462738583e-05,
+      "loss": 1.5777,
+      "step": 2469
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.557108469206803e-05,
+      "loss": 1.6272,
+      "step": 2470
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.555763955081253e-05,
+      "loss": 1.5088,
+      "step": 2471
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.554418920558705e-05,
+      "loss": 1.5807,
+      "step": 2472
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.553073365836009e-05,
+      "loss": 1.5634,
+      "step": 2473
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.551727291110092e-05,
+      "loss": 1.5033,
+      "step": 2474
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.550380696577957e-05,
+      "loss": 1.5824,
+      "step": 2475
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.549033582436682e-05,
+      "loss": 1.6674,
+      "step": 2476
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.54768594888342e-05,
+      "loss": 1.5787,
+      "step": 2477
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.546337796115401e-05,
+      "loss": 1.4957,
+      "step": 2478
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.544989124329932e-05,
+      "loss": 1.5243,
+      "step": 2479
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.543639933724398e-05,
+      "loss": 1.4083,
+      "step": 2480
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.542290224496252e-05,
+      "loss": 1.4917,
+      "step": 2481
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.540939996843033e-05,
+      "loss": 1.6077,
+      "step": 2482
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.539589250962348e-05,
+      "loss": 1.2914,
+      "step": 2483
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.538237987051882e-05,
+      "loss": 1.568,
+      "step": 2484
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.5368862053094e-05,
+      "loss": 1.5693,
+      "step": 2485
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.535533905932738e-05,
+      "loss": 1.4563,
+      "step": 2486
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.534181089119808e-05,
+      "loss": 1.4166,
+      "step": 2487
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.5328277550686e-05,
+      "loss": 1.5894,
+      "step": 2488
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.531473903977178e-05,
+      "loss": 1.4965,
+      "step": 2489
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.530119536043684e-05,
+      "loss": 1.6603,
+      "step": 2490
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.52876465146633e-05,
+      "loss": 1.6943,
+      "step": 2491
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.527409250443412e-05,
+      "loss": 1.5464,
+      "step": 2492
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.526053333173294e-05,
+      "loss": 1.6389,
+      "step": 2493
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.524696899854421e-05,
+      "loss": 1.5764,
+      "step": 2494
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.52333995068531e-05,
+      "loss": 1.4158,
+      "step": 2495
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.521982485864554e-05,
+      "loss": 1.476,
+      "step": 2496
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.520624505590823e-05,
+      "loss": 1.6888,
+      "step": 2497
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.519266010062861e-05,
+      "loss": 1.5701,
+      "step": 2498
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.517906999479489e-05,
+      "loss": 1.5486,
+      "step": 2499
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.516547474039602e-05,
+      "loss": 1.5176,
+      "step": 2500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.51518743394217e-05,
+      "loss": 1.6582,
+      "step": 2501
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.51382687938624e-05,
+      "loss": 1.6416,
+      "step": 2502
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.512465810570933e-05,
+      "loss": 1.5859,
+      "step": 2503
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.511104227695446e-05,
+      "loss": 1.3408,
+      "step": 2504
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.509742130959049e-05,
+      "loss": 1.6996,
+      "step": 2505
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.508379520561091e-05,
+      "loss": 1.5663,
+      "step": 2506
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.507016396700994e-05,
+      "loss": 1.5822,
+      "step": 2507
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.505652759578255e-05,
+      "loss": 1.6582,
+      "step": 2508
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.504288609392448e-05,
+      "loss": 1.6082,
+      "step": 2509
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.502923946343217e-05,
+      "loss": 1.4989,
+      "step": 2510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.501558770630288e-05,
+      "loss": 1.5944,
+      "step": 2511
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.500193082453457e-05,
+      "loss": 1.5776,
+      "step": 2512
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.498826882012597e-05,
+      "loss": 1.4808,
+      "step": 2513
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.497460169507657e-05,
+      "loss": 1.5625,
+      "step": 2514
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.496092945138656e-05,
+      "loss": 1.6738,
+      "step": 2515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.494725209105694e-05,
+      "loss": 1.5598,
+      "step": 2516
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.493356961608945e-05,
+      "loss": 1.4825,
+      "step": 2517
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.491988202848654e-05,
+      "loss": 1.503,
+      "step": 2518
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.490618933025143e-05,
+      "loss": 1.5613,
+      "step": 2519
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.48924915233881e-05,
+      "loss": 1.5416,
+      "step": 2520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.487878860990124e-05,
+      "loss": 1.6086,
+      "step": 2521
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.486508059179635e-05,
+      "loss": 1.6542,
+      "step": 2522
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.485136747107963e-05,
+      "loss": 1.6615,
+      "step": 2523
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.483764924975801e-05,
+      "loss": 1.5103,
+      "step": 2524
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.482392592983923e-05,
+      "loss": 1.6626,
+      "step": 2525
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.481019751333172e-05,
+      "loss": 1.5744,
+      "step": 2526
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.479646400224469e-05,
+      "loss": 1.6554,
+      "step": 2527
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.478272539858807e-05,
+      "loss": 1.6737,
+      "step": 2528
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.476898170437254e-05,
+      "loss": 1.6335,
+      "step": 2529
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.475523292160954e-05,
+      "loss": 1.3483,
+      "step": 2530
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.474147905231127e-05,
+      "loss": 1.4,
+      "step": 2531
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.47277200984906e-05,
+      "loss": 1.5227,
+      "step": 2532
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.471395606216124e-05,
+      "loss": 1.6722,
+      "step": 2533
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.470018694533759e-05,
+      "loss": 1.5392,
+      "step": 2534
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.468641275003479e-05,
+      "loss": 1.4814,
+      "step": 2535
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.467263347826873e-05,
+      "loss": 1.5074,
+      "step": 2536
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.465884913205608e-05,
+      "loss": 1.604,
+      "step": 2537
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.46450597134142e-05,
+      "loss": 1.4162,
+      "step": 2538
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.463126522436121e-05,
+      "loss": 1.6018,
+      "step": 2539
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.461746566691599e-05,
+      "loss": 1.4607,
+      "step": 2540
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.460366104309814e-05,
+      "loss": 1.5399,
+      "step": 2541
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.458985135492799e-05,
+      "loss": 1.56,
+      "step": 2542
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.457603660442667e-05,
+      "loss": 1.5781,
+      "step": 2543
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.456221679361597e-05,
+      "loss": 1.5786,
+      "step": 2544
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.45483919245185e-05,
+      "loss": 1.4562,
+      "step": 2545
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.453456199915755e-05,
+      "loss": 1.5554,
+      "step": 2546
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.452072701955718e-05,
+      "loss": 1.3862,
+      "step": 2547
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.450688698774219e-05,
+      "loss": 1.5884,
+      "step": 2548
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.449304190573808e-05,
+      "loss": 1.5384,
+      "step": 2549
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.447919177557115e-05,
+      "loss": 1.6739,
+      "step": 2550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.446533659926838e-05,
+      "loss": 1.5912,
+      "step": 2551
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.445147637885755e-05,
+      "loss": 1.5575,
+      "step": 2552
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.443761111636714e-05,
+      "loss": 1.6533,
+      "step": 2553
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.442374081382637e-05,
+      "loss": 1.4653,
+      "step": 2554
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.44098654732652e-05,
+      "loss": 1.6029,
+      "step": 2555
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.439598509671432e-05,
+      "loss": 1.5336,
+      "step": 2556
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.438209968620517e-05,
+      "loss": 1.5462,
+      "step": 2557
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.436820924376991e-05,
+      "loss": 1.6418,
+      "step": 2558
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.43543137714415e-05,
+      "loss": 1.649,
+      "step": 2559
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.434041327125353e-05,
+      "loss": 1.6095,
+      "step": 2560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.43265077452404e-05,
+      "loss": 1.3874,
+      "step": 2561
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.431259719543723e-05,
+      "loss": 1.6867,
+      "step": 2562
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.429868162387986e-05,
+      "loss": 1.6695,
+      "step": 2563
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.42847610326049e-05,
+      "loss": 1.6489,
+      "step": 2564
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.427083542364966e-05,
+      "loss": 1.5642,
+      "step": 2565
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.425690479905217e-05,
+      "loss": 1.6925,
+      "step": 2566
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.424296916085126e-05,
+      "loss": 1.5882,
+      "step": 2567
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.422902851108643e-05,
+      "loss": 1.6967,
+      "step": 2568
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.421508285179795e-05,
+      "loss": 1.7063,
+      "step": 2569
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.420113218502679e-05,
+      "loss": 1.5386,
+      "step": 2570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.41871765128147e-05,
+      "loss": 1.5464,
+      "step": 2571
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.417321583720412e-05,
+      "loss": 1.5904,
+      "step": 2572
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.415925016023823e-05,
+      "loss": 1.5444,
+      "step": 2573
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.414527948396099e-05,
+      "loss": 1.6608,
+      "step": 2574
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.413130381041698e-05,
+      "loss": 1.6392,
+      "step": 2575
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.411732314165164e-05,
+      "loss": 1.525,
+      "step": 2576
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.410333747971106e-05,
+      "loss": 1.5545,
+      "step": 2577
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.408934682664211e-05,
+      "loss": 1.614,
+      "step": 2578
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.407535118449235e-05,
+      "loss": 1.5815,
+      "step": 2579
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.406135055531006e-05,
+      "loss": 1.5211,
+      "step": 2580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.40473449411443e-05,
+      "loss": 1.3513,
+      "step": 2581
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.403333434404485e-05,
+      "loss": 1.4765,
+      "step": 2582
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.401931876606217e-05,
+      "loss": 1.5859,
+      "step": 2583
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.400529820924752e-05,
+      "loss": 1.2604,
+      "step": 2584
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.399127267565281e-05,
+      "loss": 1.5552,
+      "step": 2585
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.397724216733076e-05,
+      "loss": 1.6567,
+      "step": 2586
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.396320668633475e-05,
+      "loss": 1.5836,
+      "step": 2587
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.394916623471892e-05,
+      "loss": 1.5725,
+      "step": 2588
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.393512081453814e-05,
+      "loss": 1.6426,
+      "step": 2589
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.3921070427848e-05,
+      "loss": 1.4281,
+      "step": 2590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.390701507670483e-05,
+      "loss": 1.6412,
+      "step": 2591
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.389295476316565e-05,
+      "loss": 1.6018,
+      "step": 2592
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.387888948928824e-05,
+      "loss": 1.5184,
+      "step": 2593
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.386481925713108e-05,
+      "loss": 1.4448,
+      "step": 2594
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.385074406875343e-05,
+      "loss": 1.3026,
+      "step": 2595
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.383666392621521e-05,
+      "loss": 1.5705,
+      "step": 2596
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.38225788315771e-05,
+      "loss": 1.4996,
+      "step": 2597
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.380848878690049e-05,
+      "loss": 1.4611,
+      "step": 2598
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.37943937942475e-05,
+      "loss": 1.4984,
+      "step": 2599
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.378029385568099e-05,
+      "loss": 1.621,
+      "step": 2600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.376618897326452e-05,
+      "loss": 1.6466,
+      "step": 2601
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.375207914906239e-05,
+      "loss": 1.5943,
+      "step": 2602
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.37379643851396e-05,
+      "loss": 1.5819,
+      "step": 2603
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.37238446835619e-05,
+      "loss": 1.6039,
+      "step": 2604
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.370972004639575e-05,
+      "loss": 1.5643,
+      "step": 2605
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.369559047570834e-05,
+      "loss": 1.6528,
+      "step": 2606
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.368145597356758e-05,
+      "loss": 1.5868,
+      "step": 2607
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.366731654204207e-05,
+      "loss": 1.532,
+      "step": 2608
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.365317218320119e-05,
+      "loss": 1.6197,
+      "step": 2609
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.363902289911498e-05,
+      "loss": 1.6667,
+      "step": 2610
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.362486869185427e-05,
+      "loss": 1.5762,
+      "step": 2611
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.361070956349056e-05,
+      "loss": 1.5925,
+      "step": 2612
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.359654551609606e-05,
+      "loss": 1.5742,
+      "step": 2613
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.358237655174373e-05,
+      "loss": 1.6345,
+      "step": 2614
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.356820267250725e-05,
+      "loss": 1.7128,
+      "step": 2615
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.355402388046101e-05,
+      "loss": 1.5178,
+      "step": 2616
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.353984017768013e-05,
+      "loss": 1.5248,
+      "step": 2617
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.352565156624041e-05,
+      "loss": 1.5323,
+      "step": 2618
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.351145804821842e-05,
+      "loss": 1.5233,
+      "step": 2619
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.349725962569142e-05,
+      "loss": 1.5032,
+      "step": 2620
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.348305630073739e-05,
+      "loss": 1.7162,
+      "step": 2621
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.346884807543503e-05,
+      "loss": 1.5578,
+      "step": 2622
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.345463495186377e-05,
+      "loss": 1.503,
+      "step": 2623
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.344041693210371e-05,
+      "loss": 1.5399,
+      "step": 2624
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.342619401823575e-05,
+      "loss": 1.7063,
+      "step": 2625
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.341196621234142e-05,
+      "loss": 1.4115,
+      "step": 2626
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.3397733516503e-05,
+      "loss": 1.6224,
+      "step": 2627
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.338349593280353e-05,
+      "loss": 1.6001,
+      "step": 2628
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.33692534633267e-05,
+      "loss": 1.5643,
+      "step": 2629
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.335500611015691e-05,
+      "loss": 1.5835,
+      "step": 2630
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.334075387537934e-05,
+      "loss": 1.5296,
+      "step": 2631
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.332649676107986e-05,
+      "loss": 1.5315,
+      "step": 2632
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.3312234769345e-05,
+      "loss": 1.5668,
+      "step": 2633
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.329796790226208e-05,
+      "loss": 1.636,
+      "step": 2634
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.328369616191908e-05,
+      "loss": 1.5828,
+      "step": 2635
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.326941955040473e-05,
+      "loss": 1.3474,
+      "step": 2636
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.325513806980844e-05,
+      "loss": 1.3169,
+      "step": 2637
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.324085172222036e-05,
+      "loss": 1.4526,
+      "step": 2638
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.322656050973133e-05,
+      "loss": 1.4867,
+      "step": 2639
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.321226443443293e-05,
+      "loss": 1.5994,
+      "step": 2640
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.319796349841743e-05,
+      "loss": 1.5685,
+      "step": 2641
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.318365770377779e-05,
+      "loss": 1.6975,
+      "step": 2642
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.316934705260774e-05,
+      "loss": 1.5014,
+      "step": 2643
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.315503154700168e-05,
+      "loss": 1.3894,
+      "step": 2644
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.31407111890547e-05,
+      "loss": 1.2705,
+      "step": 2645
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.312638598086269e-05,
+      "loss": 1.4589,
+      "step": 2646
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.311205592452212e-05,
+      "loss": 1.4747,
+      "step": 2647
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.309772102213029e-05,
+      "loss": 1.5943,
+      "step": 2648
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.308338127578512e-05,
+      "loss": 1.6516,
+      "step": 2649
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.30690366875853e-05,
+      "loss": 1.6158,
+      "step": 2650
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.30546872596302e-05,
+      "loss": 1.6101,
+      "step": 2651
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.304033299401988e-05,
+      "loss": 1.7399,
+      "step": 2652
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.302597389285515e-05,
+      "loss": 1.5691,
+      "step": 2653
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.301160995823753e-05,
+      "loss": 1.5318,
+      "step": 2654
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.299724119226919e-05,
+      "loss": 1.6493,
+      "step": 2655
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.298286759705306e-05,
+      "loss": 1.5964,
+      "step": 2656
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.296848917469277e-05,
+      "loss": 1.5248,
+      "step": 2657
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.295410592729262e-05,
+      "loss": 1.7002,
+      "step": 2658
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.293971785695766e-05,
+      "loss": 1.5183,
+      "step": 2659
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.292532496579364e-05,
+      "loss": 1.4444,
+      "step": 2660
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.291092725590696e-05,
+      "loss": 1.6235,
+      "step": 2661
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.289652472940481e-05,
+      "loss": 1.6223,
+      "step": 2662
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.288211738839505e-05,
+      "loss": 1.7104,
+      "step": 2663
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.286770523498622e-05,
+      "loss": 1.6346,
+      "step": 2664
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.285328827128755e-05,
+      "loss": 1.6547,
+      "step": 2665
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.283886649940906e-05,
+      "loss": 1.5589,
+      "step": 2666
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.28244399214614e-05,
+      "loss": 1.5605,
+      "step": 2667
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.281000853955596e-05,
+      "loss": 1.6166,
+      "step": 2668
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.27955723558048e-05,
+      "loss": 1.6795,
+      "step": 2669
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.278113137232069e-05,
+      "loss": 1.5237,
+      "step": 2670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.276668559121713e-05,
+      "loss": 1.5827,
+      "step": 2671
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.275223501460829e-05,
+      "loss": 1.6492,
+      "step": 2672
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.273777964460908e-05,
+      "loss": 1.6585,
+      "step": 2673
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.272331948333507e-05,
+      "loss": 1.5918,
+      "step": 2674
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.270885453290254e-05,
+      "loss": 1.4459,
+      "step": 2675
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.26943847954285e-05,
+      "loss": 1.6084,
+      "step": 2676
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.267991027303063e-05,
+      "loss": 1.6505,
+      "step": 2677
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.266543096782735e-05,
+      "loss": 1.4718,
+      "step": 2678
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.26509468819377e-05,
+      "loss": 1.4242,
+      "step": 2679
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.26364580174815e-05,
+      "loss": 1.3987,
+      "step": 2680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.262196437657924e-05,
+      "loss": 1.6573,
+      "step": 2681
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.260746596135211e-05,
+      "loss": 1.6876,
+      "step": 2682
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.259296277392197e-05,
+      "loss": 1.5641,
+      "step": 2683
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.257845481641144e-05,
+      "loss": 1.4995,
+      "step": 2684
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.25639420909438e-05,
+      "loss": 1.5782,
+      "step": 2685
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.254942459964303e-05,
+      "loss": 1.5421,
+      "step": 2686
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.253490234463379e-05,
+      "loss": 1.4309,
+      "step": 2687
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.252037532804149e-05,
+      "loss": 1.4833,
+      "step": 2688
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.250584355199216e-05,
+      "loss": 1.3971,
+      "step": 2689
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.249130701861261e-05,
+      "loss": 1.5271,
+      "step": 2690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.247676573003029e-05,
+      "loss": 1.4527,
+      "step": 2691
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.246221968837335e-05,
+      "loss": 1.5183,
+      "step": 2692
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.244766889577067e-05,
+      "loss": 1.6105,
+      "step": 2693
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.24331133543518e-05,
+      "loss": 1.6362,
+      "step": 2694
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.241855306624697e-05,
+      "loss": 1.3814,
+      "step": 2695
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.240398803358716e-05,
+      "loss": 1.617,
+      "step": 2696
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.238941825850396e-05,
+      "loss": 1.5928,
+      "step": 2697
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.237484374312973e-05,
+      "loss": 1.5776,
+      "step": 2698
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.23602644895975e-05,
+      "loss": 1.5144,
+      "step": 2699
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.234568050004097e-05,
+      "loss": 1.5388,
+      "step": 2700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.233109177659457e-05,
+      "loss": 1.5931,
+      "step": 2701
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.231649832139338e-05,
+      "loss": 1.5273,
+      "step": 2702
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.230190013657321e-05,
+      "loss": 1.6165,
+      "step": 2703
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.228729722427055e-05,
+      "loss": 1.5831,
+      "step": 2704
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.22726895866226e-05,
+      "loss": 1.4537,
+      "step": 2705
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.22580772257672e-05,
+      "loss": 1.5485,
+      "step": 2706
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.224346014384294e-05,
+      "loss": 1.5237,
+      "step": 2707
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.222883834298905e-05,
+      "loss": 1.505,
+      "step": 2708
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.221421182534551e-05,
+      "loss": 1.6821,
+      "step": 2709
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.219958059305293e-05,
+      "loss": 1.5274,
+      "step": 2710
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.218494464825263e-05,
+      "loss": 1.6046,
+      "step": 2711
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.217030399308665e-05,
+      "loss": 1.5694,
+      "step": 2712
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.215565862969766e-05,
+      "loss": 1.6081,
+      "step": 2713
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.214100856022909e-05,
+      "loss": 1.695,
+      "step": 2714
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.212635378682501e-05,
+      "loss": 1.5833,
+      "step": 2715
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.211169431163019e-05,
+      "loss": 1.6125,
+      "step": 2716
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.209703013679009e-05,
+      "loss": 1.5396,
+      "step": 2717
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.208236126445084e-05,
+      "loss": 1.6617,
+      "step": 2718
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.20676876967593e-05,
+      "loss": 1.5831,
+      "step": 2719
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.205300943586299e-05,
+      "loss": 1.5687,
+      "step": 2720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.20383264839101e-05,
+      "loss": 1.6107,
+      "step": 2721
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.202363884304956e-05,
+      "loss": 1.6161,
+      "step": 2722
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.20089465154309e-05,
+      "loss": 1.5731,
+      "step": 2723
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.199424950320442e-05,
+      "loss": 1.6922,
+      "step": 2724
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.197954780852109e-05,
+      "loss": 1.6449,
+      "step": 2725
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.19648414335325e-05,
+      "loss": 1.5326,
+      "step": 2726
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.195013038039101e-05,
+      "loss": 1.6006,
+      "step": 2727
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.193541465124961e-05,
+      "loss": 1.61,
+      "step": 2728
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.1920694248262e-05,
+      "loss": 1.5563,
+      "step": 2729
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.190596917358257e-05,
+      "loss": 1.254,
+      "step": 2730
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.189123942936637e-05,
+      "loss": 1.4714,
+      "step": 2731
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.187650501776911e-05,
+      "loss": 1.5535,
+      "step": 2732
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.186176594094727e-05,
+      "loss": 1.6986,
+      "step": 2733
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.184702220105793e-05,
+      "loss": 1.3327,
+      "step": 2734
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.183227380025889e-05,
+      "loss": 1.3808,
+      "step": 2735
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.18175207407086e-05,
+      "loss": 1.4911,
+      "step": 2736
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.180276302456626e-05,
+      "loss": 1.4613,
+      "step": 2737
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.178800065399168e-05,
+      "loss": 1.6693,
+      "step": 2738
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.177323363114539e-05,
+      "loss": 1.6532,
+      "step": 2739
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.175846195818857e-05,
+      "loss": 1.5687,
+      "step": 2740
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.17436856372831e-05,
+      "loss": 1.6764,
+      "step": 2741
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.172890467059156e-05,
+      "loss": 1.326,
+      "step": 2742
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.171411906027718e-05,
+      "loss": 1.2682,
+      "step": 2743
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.169932880850388e-05,
+      "loss": 1.7491,
+      "step": 2744
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.168453391743626e-05,
+      "loss": 1.3788,
+      "step": 2745
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.16697343892396e-05,
+      "loss": 1.413,
+      "step": 2746
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.165493022607987e-05,
+      "loss": 1.6076,
+      "step": 2747
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.164012143012364e-05,
+      "loss": 1.5937,
+      "step": 2748
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.162530800353832e-05,
+      "loss": 1.7381,
+      "step": 2749
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.161048994849182e-05,
+      "loss": 1.7102,
+      "step": 2750
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.159566726715286e-05,
+      "loss": 1.7128,
+      "step": 2751
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.158083996169075e-05,
+      "loss": 1.588,
+      "step": 2752
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.156600803427554e-05,
+      "loss": 1.6785,
+      "step": 2753
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.155117148707792e-05,
+      "loss": 1.6136,
+      "step": 2754
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.153633032226925e-05,
+      "loss": 1.4843,
+      "step": 2755
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.152148454202159e-05,
+      "loss": 1.6086,
+      "step": 2756
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.150663414850766e-05,
+      "loss": 1.5487,
+      "step": 2757
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.149177914390087e-05,
+      "loss": 1.644,
+      "step": 2758
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.147691953037528e-05,
+      "loss": 1.4757,
+      "step": 2759
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.146205531010567e-05,
+      "loss": 1.7137,
+      "step": 2760
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.144718648526744e-05,
+      "loss": 1.575,
+      "step": 2761
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.143231305803669e-05,
+      "loss": 1.5488,
+      "step": 2762
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.14174350305902e-05,
+      "loss": 1.4937,
+      "step": 2763
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.140255240510541e-05,
+      "loss": 1.5799,
+      "step": 2764
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.138766518376045e-05,
+      "loss": 1.6581,
+      "step": 2765
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.137277336873408e-05,
+      "loss": 1.6556,
+      "step": 2766
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.135787696220578e-05,
+      "loss": 1.6338,
+      "step": 2767
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.13429759663557e-05,
+      "loss": 1.5437,
+      "step": 2768
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.132807038336464e-05,
+      "loss": 1.6679,
+      "step": 2769
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.131316021541407e-05,
+      "loss": 1.5804,
+      "step": 2770
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.129824546468614e-05,
+      "loss": 1.4422,
+      "step": 2771
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.128332613336367e-05,
+      "loss": 1.5915,
+      "step": 2772
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.126840222363014e-05,
+      "loss": 1.5008,
+      "step": 2773
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.125347373766974e-05,
+      "loss": 1.5485,
+      "step": 2774
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.123854067766728e-05,
+      "loss": 1.6966,
+      "step": 2775
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.122360304580824e-05,
+      "loss": 1.6043,
+      "step": 2776
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.120866084427882e-05,
+      "loss": 1.6595,
+      "step": 2777
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.119371407526584e-05,
+      "loss": 1.562,
+      "step": 2778
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.117876274095682e-05,
+      "loss": 1.4526,
+      "step": 2779
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.116380684353992e-05,
+      "loss": 1.5157,
+      "step": 2780
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.114884638520399e-05,
+      "loss": 1.4689,
+      "step": 2781
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.113388136813853e-05,
+      "loss": 1.461,
+      "step": 2782
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.111891179453373e-05,
+      "loss": 1.4502,
+      "step": 2783
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.110393766658042e-05,
+      "loss": 1.4907,
+      "step": 2784
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.108895898647013e-05,
+      "loss": 1.5731,
+      "step": 2785
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.107397575639501e-05,
+      "loss": 1.3103,
+      "step": 2786
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.105898797854791e-05,
+      "loss": 1.4353,
+      "step": 2787
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.104399565512236e-05,
+      "loss": 1.6277,
+      "step": 2788
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.102899878831248e-05,
+      "loss": 1.6392,
+      "step": 2789
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.101399738031316e-05,
+      "loss": 1.5771,
+      "step": 2790
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.099899143331988e-05,
+      "loss": 1.5402,
+      "step": 2791
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.098398094952882e-05,
+      "loss": 1.5919,
+      "step": 2792
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.096896593113678e-05,
+      "loss": 1.6496,
+      "step": 2793
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.095394638034129e-05,
+      "loss": 1.5891,
+      "step": 2794
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.093892229934047e-05,
+      "loss": 1.5147,
+      "step": 2795
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.092389369033317e-05,
+      "loss": 1.4977,
+      "step": 2796
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.090886055551887e-05,
+      "loss": 1.5067,
+      "step": 2797
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.089382289709771e-05,
+      "loss": 1.5098,
+      "step": 2798
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.08787807172705e-05,
+      "loss": 1.7086,
+      "step": 2799
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.086373401823873e-05,
+      "loss": 1.6645,
+      "step": 2800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.084868280220448e-05,
+      "loss": 1.5653,
+      "step": 2801
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.083362707137057e-05,
+      "loss": 1.5906,
+      "step": 2802
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.081856682794047e-05,
+      "loss": 1.5106,
+      "step": 2803
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.080350207411825e-05,
+      "loss": 1.7211,
+      "step": 2804
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.078843281210873e-05,
+      "loss": 1.5982,
+      "step": 2805
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.077335904411731e-05,
+      "loss": 1.5755,
+      "step": 2806
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.07582807723501e-05,
+      "loss": 1.5109,
+      "step": 2807
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.074319799901384e-05,
+      "loss": 1.5752,
+      "step": 2808
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.072811072631594e-05,
+      "loss": 1.6428,
+      "step": 2809
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.071301895646449e-05,
+      "loss": 1.4762,
+      "step": 2810
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.069792269166819e-05,
+      "loss": 1.6976,
+      "step": 2811
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.068282193413646e-05,
+      "loss": 1.5276,
+      "step": 2812
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.066771668607929e-05,
+      "loss": 1.471,
+      "step": 2813
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.065260694970743e-05,
+      "loss": 1.5599,
+      "step": 2814
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.063749272723218e-05,
+      "loss": 1.6635,
+      "step": 2815
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.06223740208656e-05,
+      "loss": 1.481,
+      "step": 2816
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.060725083282037e-05,
+      "loss": 1.6193,
+      "step": 2817
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.059212316530977e-05,
+      "loss": 1.5771,
+      "step": 2818
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.057699102054782e-05,
+      "loss": 1.6578,
+      "step": 2819
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.056185440074913e-05,
+      "loss": 1.4487,
+      "step": 2820
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.054671330812899e-05,
+      "loss": 1.4644,
+      "step": 2821
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.053156774490337e-05,
+      "loss": 1.7203,
+      "step": 2822
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.051641771328884e-05,
+      "loss": 1.5985,
+      "step": 2823
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.050126321550268e-05,
+      "loss": 1.6519,
+      "step": 2824
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.048610425376278e-05,
+      "loss": 1.5972,
+      "step": 2825
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.047094083028771e-05,
+      "loss": 1.7101,
+      "step": 2826
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.045577294729666e-05,
+      "loss": 1.7522,
+      "step": 2827
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.044060060700956e-05,
+      "loss": 1.5394,
+      "step": 2828
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.042542381164685e-05,
+      "loss": 1.4224,
+      "step": 2829
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.041024256342975e-05,
+      "loss": 1.6069,
+      "step": 2830
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.039505686458007e-05,
+      "loss": 1.5832,
+      "step": 2831
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.037986671732027e-05,
+      "loss": 1.6719,
+      "step": 2832
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.036467212387349e-05,
+      "loss": 1.5205,
+      "step": 2833
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.03494730864635e-05,
+      "loss": 1.4847,
+      "step": 2834
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.033426960731472e-05,
+      "loss": 1.35,
+      "step": 2835
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.031906168865225e-05,
+      "loss": 1.512,
+      "step": 2836
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.03038493327018e-05,
+      "loss": 1.6916,
+      "step": 2837
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.028863254168975e-05,
+      "loss": 1.4412,
+      "step": 2838
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.02734113178431e-05,
+      "loss": 1.6927,
+      "step": 2839
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.025818566338956e-05,
+      "loss": 1.5465,
+      "step": 2840
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.024295558055743e-05,
+      "loss": 1.5962,
+      "step": 2841
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.022772107157567e-05,
+      "loss": 1.5155,
+      "step": 2842
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.021248213867391e-05,
+      "loss": 1.6112,
+      "step": 2843
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.019723878408242e-05,
+      "loss": 1.5464,
+      "step": 2844
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.018199101003212e-05,
+      "loss": 1.5108,
+      "step": 2845
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.016673881875455e-05,
+      "loss": 1.5329,
+      "step": 2846
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.015148221248193e-05,
+      "loss": 1.7141,
+      "step": 2847
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.013622119344708e-05,
+      "loss": 1.5166,
+      "step": 2848
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.012095576388354e-05,
+      "loss": 1.6326,
+      "step": 2849
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.01056859260254e-05,
+      "loss": 1.5188,
+      "step": 2850
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.009041168210751e-05,
+      "loss": 1.5933,
+      "step": 2851
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.007513303436526e-05,
+      "loss": 1.6387,
+      "step": 2852
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.005984998503474e-05,
+      "loss": 1.6113,
+      "step": 2853
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.004456253635266e-05,
+      "loss": 1.6195,
+      "step": 2854
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.002927069055641e-05,
+      "loss": 1.6281,
+      "step": 2855
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.001397444988399e-05,
+      "loss": 1.5965,
+      "step": 2856
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.999867381657402e-05,
+      "loss": 1.6039,
+      "step": 2857
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.998336879286584e-05,
+      "loss": 1.5913,
+      "step": 2858
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.996805938099936e-05,
+      "loss": 1.5816,
+      "step": 2859
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.995274558321517e-05,
+      "loss": 1.5088,
+      "step": 2860
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.99374274017545e-05,
+      "loss": 1.5315,
+      "step": 2861
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.99221048388592e-05,
+      "loss": 1.7029,
+      "step": 2862
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.990677789677177e-05,
+      "loss": 1.5962,
+      "step": 2863
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.989144657773536e-05,
+      "loss": 1.5777,
+      "step": 2864
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.987611088399377e-05,
+      "loss": 1.5809,
+      "step": 2865
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.98607708177914e-05,
+      "loss": 1.6216,
+      "step": 2866
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.984542638137333e-05,
+      "loss": 1.6119,
+      "step": 2867
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.983007757698527e-05,
+      "loss": 1.5721,
+      "step": 2868
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.981472440687354e-05,
+      "loss": 1.5932,
+      "step": 2869
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.979936687328517e-05,
+      "loss": 1.6687,
+      "step": 2870
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.978400497846775e-05,
+      "loss": 1.644,
+      "step": 2871
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.976863872466954e-05,
+      "loss": 1.5229,
+      "step": 2872
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.975326811413943e-05,
+      "loss": 1.621,
+      "step": 2873
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.973789314912699e-05,
+      "loss": 1.7339,
+      "step": 2874
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.972251383188236e-05,
+      "loss": 1.6638,
+      "step": 2875
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.970713016465637e-05,
+      "loss": 1.5736,
+      "step": 2876
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.969174214970048e-05,
+      "loss": 1.5883,
+      "step": 2877
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.967634978926674e-05,
+      "loss": 1.6785,
+      "step": 2878
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.966095308560788e-05,
+      "loss": 1.5803,
+      "step": 2879
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.964555204097726e-05,
+      "loss": 1.618,
+      "step": 2880
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.963014665762887e-05,
+      "loss": 1.4508,
+      "step": 2881
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.961473693781733e-05,
+      "loss": 1.1769,
+      "step": 2882
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.959932288379793e-05,
+      "loss": 1.6116,
+      "step": 2883
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.958390449782652e-05,
+      "loss": 1.5833,
+      "step": 2884
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.956848178215963e-05,
+      "loss": 1.5444,
+      "step": 2885
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.955305473905446e-05,
+      "loss": 1.4335,
+      "step": 2886
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.953762337076878e-05,
+      "loss": 1.6126,
+      "step": 2887
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.952218767956104e-05,
+      "loss": 1.6439,
+      "step": 2888
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.950674766769025e-05,
+      "loss": 1.4667,
+      "step": 2889
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.949130333741615e-05,
+      "loss": 1.3694,
+      "step": 2890
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.947585469099905e-05,
+      "loss": 1.4395,
+      "step": 2891
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.946040173069992e-05,
+      "loss": 1.5723,
+      "step": 2892
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.944494445878034e-05,
+      "loss": 1.459,
+      "step": 2893
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.942948287750251e-05,
+      "loss": 1.5407,
+      "step": 2894
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.941401698912931e-05,
+      "loss": 1.5343,
+      "step": 2895
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.93985467959242e-05,
+      "loss": 1.5584,
+      "step": 2896
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.93830723001513e-05,
+      "loss": 1.4444,
+      "step": 2897
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.936759350407534e-05,
+      "loss": 1.7949,
+      "step": 2898
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.93521104099617e-05,
+      "loss": 1.6497,
+      "step": 2899
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.933662302007638e-05,
+      "loss": 1.4367,
+      "step": 2900
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.932113133668599e-05,
+      "loss": 1.6085,
+      "step": 2901
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.930563536205782e-05,
+      "loss": 1.627,
+      "step": 2902
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.92901350984597e-05,
+      "loss": 1.5266,
+      "step": 2903
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.92746305481602e-05,
+      "loss": 1.7022,
+      "step": 2904
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.925912171342842e-05,
+      "loss": 1.5519,
+      "step": 2905
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.924360859653413e-05,
+      "loss": 1.5863,
+      "step": 2906
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.922809119974774e-05,
+      "loss": 1.5312,
+      "step": 2907
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.921256952534026e-05,
+      "loss": 1.5567,
+      "step": 2908
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.919704357558333e-05,
+      "loss": 1.5911,
+      "step": 2909
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.918151335274924e-05,
+      "loss": 1.5677,
+      "step": 2910
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.916597885911084e-05,
+      "loss": 1.6926,
+      "step": 2911
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.915044009694169e-05,
+      "loss": 1.6915,
+      "step": 2912
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.913489706851594e-05,
+      "loss": 1.6759,
+      "step": 2913
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.911934977610833e-05,
+      "loss": 1.6094,
+      "step": 2914
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.910379822199427e-05,
+      "loss": 1.6348,
+      "step": 2915
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.908824240844978e-05,
+      "loss": 1.6636,
+      "step": 2916
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.90726823377515e-05,
+      "loss": 1.6509,
+      "step": 2917
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.90571180121767e-05,
+      "loss": 1.632,
+      "step": 2918
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.904154943400326e-05,
+      "loss": 1.5499,
+      "step": 2919
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.902597660550967e-05,
+      "loss": 1.4311,
+      "step": 2920
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.901039952897511e-05,
+      "loss": 1.4824,
+      "step": 2921
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.899481820667929e-05,
+      "loss": 1.6954,
+      "step": 2922
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.89792326409026e-05,
+      "loss": 1.4956,
+      "step": 2923
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.896364283392603e-05,
+      "loss": 1.6132,
+      "step": 2924
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.894804878803121e-05,
+      "loss": 1.6487,
+      "step": 2925
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.893245050550036e-05,
+      "loss": 1.5185,
+      "step": 2926
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.891684798861634e-05,
+      "loss": 1.408,
+      "step": 2927
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.890124123966263e-05,
+      "loss": 1.5958,
+      "step": 2928
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.888563026092333e-05,
+      "loss": 1.5865,
+      "step": 2929
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.887001505468315e-05,
+      "loss": 1.6528,
+      "step": 2930
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.885439562322743e-05,
+      "loss": 1.5459,
+      "step": 2931
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.88387719688421e-05,
+      "loss": 1.6532,
+      "step": 2932
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.882314409381376e-05,
+      "loss": 1.4366,
+      "step": 2933
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.880751200042957e-05,
+      "loss": 1.6146,
+      "step": 2934
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.879187569097737e-05,
+      "loss": 1.3138,
+      "step": 2935
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.877623516774554e-05,
+      "loss": 1.3837,
+      "step": 2936
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.876059043302315e-05,
+      "loss": 1.4119,
+      "step": 2937
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.874494148909985e-05,
+      "loss": 1.3481,
+      "step": 2938
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.87292883382659e-05,
+      "loss": 1.5732,
+      "step": 2939
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.871363098281217e-05,
+      "loss": 1.4905,
+      "step": 2940
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.869796942503021e-05,
+      "loss": 1.5153,
+      "step": 2941
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.868230366721209e-05,
+      "loss": 1.5528,
+      "step": 2942
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.866663371165059e-05,
+      "loss": 1.5024,
+      "step": 2943
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.865095956063901e-05,
+      "loss": 1.2599,
+      "step": 2944
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.863528121647133e-05,
+      "loss": 1.5023,
+      "step": 2945
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.861959868144214e-05,
+      "loss": 1.44,
+      "step": 2946
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.860391195784659e-05,
+      "loss": 1.5606,
+      "step": 2947
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.858822104798051e-05,
+      "loss": 1.4254,
+      "step": 2948
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.857252595414031e-05,
+      "loss": 1.5657,
+      "step": 2949
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.8556826678623e-05,
+      "loss": 1.6297,
+      "step": 2950
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.854112322372623e-05,
+      "loss": 1.6493,
+      "step": 2951
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.852541559174824e-05,
+      "loss": 1.6139,
+      "step": 2952
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.850970378498789e-05,
+      "loss": 1.5902,
+      "step": 2953
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.849398780574468e-05,
+      "loss": 1.6175,
+      "step": 2954
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.847826765631867e-05,
+      "loss": 1.5032,
+      "step": 2955
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.846254333901052e-05,
+      "loss": 1.6043,
+      "step": 2956
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.84468148561216e-05,
+      "loss": 1.6248,
+      "step": 2957
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.843108220995376e-05,
+      "loss": 1.6699,
+      "step": 2958
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.841534540280954e-05,
+      "loss": 1.538,
+      "step": 2959
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.839960443699211e-05,
+      "loss": 1.4557,
+      "step": 2960
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.838385931480514e-05,
+      "loss": 1.4161,
+      "step": 2961
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.836811003855305e-05,
+      "loss": 1.5783,
+      "step": 2962
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.835235661054072e-05,
+      "loss": 1.6837,
+      "step": 2963
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.833659903307377e-05,
+      "loss": 1.5593,
+      "step": 2964
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.832083730845835e-05,
+      "loss": 1.6067,
+      "step": 2965
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.830507143900123e-05,
+      "loss": 1.6361,
+      "step": 2966
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.828930142700981e-05,
+      "loss": 1.5115,
+      "step": 2967
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.827352727479205e-05,
+      "loss": 1.5629,
+      "step": 2968
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.825774898465658e-05,
+      "loss": 1.5096,
+      "step": 2969
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.82419665589126e-05,
+      "loss": 1.6315,
+      "step": 2970
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.82261799998699e-05,
+      "loss": 1.5161,
+      "step": 2971
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.821038930983888e-05,
+      "loss": 1.5885,
+      "step": 2972
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.819459449113058e-05,
+      "loss": 1.6364,
+      "step": 2973
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.817879554605661e-05,
+      "loss": 1.6515,
+      "step": 2974
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.81629924769292e-05,
+      "loss": 1.517,
+      "step": 2975
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.814718528606117e-05,
+      "loss": 1.5537,
+      "step": 2976
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.813137397576595e-05,
+      "loss": 1.5097,
+      "step": 2977
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.81155585483576e-05,
+      "loss": 1.542,
+      "step": 2978
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.80997390061507e-05,
+      "loss": 1.5971,
+      "step": 2979
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.808391535146055e-05,
+      "loss": 1.4477,
+      "step": 2980
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.806808758660295e-05,
+      "loss": 1.6044,
+      "step": 2981
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.805225571389439e-05,
+      "loss": 1.4331,
+      "step": 2982
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.803641973565184e-05,
+      "loss": 1.5784,
+      "step": 2983
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.8020579654193e-05,
+      "loss": 1.4603,
+      "step": 2984
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.800473547183611e-05,
+      "loss": 1.155,
+      "step": 2985
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.798888719089998e-05,
+      "loss": 1.5432,
+      "step": 2986
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.79730348137041e-05,
+      "loss": 1.4819,
+      "step": 2987
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.795717834256848e-05,
+      "loss": 1.5941,
+      "step": 2988
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.794131777981378e-05,
+      "loss": 1.5539,
+      "step": 2989
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.792545312776123e-05,
+      "loss": 1.5617,
+      "step": 2990
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.790958438873269e-05,
+      "loss": 1.5641,
+      "step": 2991
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.789371156505058e-05,
+      "loss": 1.5069,
+      "step": 2992
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.787783465903796e-05,
+      "loss": 1.6479,
+      "step": 2993
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.786195367301842e-05,
+      "loss": 1.4045,
+      "step": 2994
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.784606860931625e-05,
+      "loss": 1.546,
+      "step": 2995
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.783017947025622e-05,
+      "loss": 1.4995,
+      "step": 2996
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.781428625816379e-05,
+      "loss": 1.4831,
+      "step": 2997
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.779838897536498e-05,
+      "loss": 1.5297,
+      "step": 2998
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.778248762418639e-05,
+      "loss": 1.4517,
+      "step": 2999
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.776658220695525e-05,
+      "loss": 1.6647,
+      "step": 3000
+    },
+    {
+      "epoch": 0.35,
+      "eval_loss": 1.6164995431900024,
+      "eval_runtime": 4348.0653,
+      "eval_samples_per_second": 2.815,
+      "eval_steps_per_second": 0.352,
+      "step": 3000
+    }
+  ],
+  "max_steps": 8645,
+  "num_train_epochs": 1,
+  "total_flos": 1.545737794194899e+18,
+  "trial_name": null,
+  "trial_params": null
+}