diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -2,6684 +2,3204 @@
   "best_metric": null,
   "best_model_checkpoint": null,
   "epoch": 10.0,
-  "global_step": 1110,
+  "global_step": 530,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.01,
-      "learning_rate": 0.00019981981981981984,
-      "loss": 9.3363,
+      "epoch": 0.02,
+      "learning_rate": 4.9905660377358493e-05,
+      "loss": 9.8227,
       "step": 1
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 0.00019963963963963967,
-      "loss": 5.0479,
+      "epoch": 0.04,
+      "learning_rate": 4.9811320754716985e-05,
+      "loss": 5.3632,
       "step": 2
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 0.00019945945945945947,
-      "loss": 4.3977,
+      "epoch": 0.06,
+      "learning_rate": 4.9716981132075476e-05,
+      "loss": 4.5065,
       "step": 3
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 0.0001992792792792793,
-      "loss": 3.9062,
+      "epoch": 0.08,
+      "learning_rate": 4.962264150943397e-05,
+      "loss": 3.6359,
       "step": 4
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 0.00019909909909909912,
-      "loss": 3.6201,
+      "epoch": 0.09,
+      "learning_rate": 4.952830188679246e-05,
+      "loss": 3.3431,
       "step": 5
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 0.00019891891891891895,
-      "loss": 3.2993,
+      "epoch": 0.11,
+      "learning_rate": 4.943396226415095e-05,
+      "loss": 3.0381,
       "step": 6
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 0.00019873873873873873,
-      "loss": 2.7671,
+      "epoch": 0.13,
+      "learning_rate": 4.933962264150943e-05,
+      "loss": 3.0548,
       "step": 7
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 0.00019855855855855856,
-      "loss": 2.8142,
+      "epoch": 0.15,
+      "learning_rate": 4.9245283018867924e-05,
+      "loss": 2.7514,
       "step": 8
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 0.00019837837837837838,
-      "loss": 2.6968,
+      "epoch": 0.17,
+      "learning_rate": 4.9150943396226415e-05,
+      "loss": 2.8738,
       "step": 9
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 0.0001981981981981982,
-      "loss": 2.3751,
+      "epoch": 0.19,
+      "learning_rate": 4.9056603773584906e-05,
+      "loss": 2.5376,
       "step": 10
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 0.000198018018018018,
-      "loss": 2.2539,
+      "epoch": 0.21,
+      "learning_rate": 4.89622641509434e-05,
+      "loss": 2.4207,
       "step": 11
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 0.00019783783783783784,
-      "loss": 2.2155,
+      "epoch": 0.23,
+      "learning_rate": 4.886792452830189e-05,
+      "loss": 2.4471,
       "step": 12
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 0.00019765765765765767,
-      "loss": 2.1593,
+      "epoch": 0.25,
+      "learning_rate": 4.877358490566038e-05,
+      "loss": 2.3879,
       "step": 13
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 0.0001974774774774775,
-      "loss": 2.1809,
+      "epoch": 0.26,
+      "learning_rate": 4.867924528301887e-05,
+      "loss": 2.328,
       "step": 14
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 0.0001972972972972973,
-      "loss": 2.4524,
+      "epoch": 0.28,
+      "learning_rate": 4.858490566037736e-05,
+      "loss": 2.1208,
       "step": 15
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 0.00019711711711711713,
-      "loss": 2.1628,
+      "epoch": 0.3,
+      "learning_rate": 4.849056603773585e-05,
+      "loss": 2.2425,
       "step": 16
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 0.00019693693693693696,
-      "loss": 2.1451,
+      "epoch": 0.32,
+      "learning_rate": 4.8396226415094344e-05,
+      "loss": 2.2906,
       "step": 17
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 0.00019675675675675678,
-      "loss": 2.1149,
+      "epoch": 0.34,
+      "learning_rate": 4.8301886792452835e-05,
+      "loss": 2.1438,
       "step": 18
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 0.00019657657657657659,
-      "loss": 1.8829,
+      "epoch": 0.36,
+      "learning_rate": 4.8207547169811326e-05,
+      "loss": 2.2857,
       "step": 19
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 0.0001963963963963964,
-      "loss": 1.752,
+      "epoch": 0.38,
+      "learning_rate": 4.811320754716982e-05,
+      "loss": 2.1923,
       "step": 20
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 0.00019621621621621622,
-      "loss": 1.9402,
+      "epoch": 0.4,
+      "learning_rate": 4.80188679245283e-05,
+      "loss": 2.2866,
       "step": 21
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 0.00019603603603603604,
-      "loss": 1.915,
+      "epoch": 0.42,
+      "learning_rate": 4.792452830188679e-05,
+      "loss": 2.174,
       "step": 22
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 0.00019585585585585587,
-      "loss": 1.5993,
+      "epoch": 0.43,
+      "learning_rate": 4.7830188679245284e-05,
+      "loss": 2.2098,
       "step": 23
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 0.00019567567567567567,
-      "loss": 1.8401,
+      "epoch": 0.45,
+      "learning_rate": 4.7735849056603775e-05,
+      "loss": 1.9968,
       "step": 24
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 0.0001954954954954955,
-      "loss": 2.0089,
+      "epoch": 0.47,
+      "learning_rate": 4.7641509433962266e-05,
+      "loss": 2.0575,
       "step": 25
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 0.00019531531531531533,
-      "loss": 1.9848,
+      "epoch": 0.49,
+      "learning_rate": 4.754716981132076e-05,
+      "loss": 2.0604,
       "step": 26
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 0.00019513513513513516,
-      "loss": 1.6511,
+      "epoch": 0.51,
+      "learning_rate": 4.745283018867925e-05,
+      "loss": 2.2124,
       "step": 27
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 0.00019495495495495496,
-      "loss": 1.6407,
+      "epoch": 0.53,
+      "learning_rate": 4.735849056603774e-05,
+      "loss": 2.002,
       "step": 28
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 0.0001947747747747748,
-      "loss": 1.8594,
+      "epoch": 0.55,
+      "learning_rate": 4.726415094339623e-05,
+      "loss": 1.9484,
       "step": 29
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 0.00019459459459459462,
-      "loss": 1.8043,
+      "epoch": 0.57,
+      "learning_rate": 4.716981132075472e-05,
+      "loss": 1.9954,
       "step": 30
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 0.00019441441441441442,
-      "loss": 1.6376,
+      "epoch": 0.58,
+      "learning_rate": 4.707547169811321e-05,
+      "loss": 1.917,
       "step": 31
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 0.00019423423423423425,
-      "loss": 1.6761,
+      "epoch": 0.6,
+      "learning_rate": 4.6981132075471704e-05,
+      "loss": 2.0466,
       "step": 32
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 0.00019405405405405405,
-      "loss": 1.5476,
+      "epoch": 0.62,
+      "learning_rate": 4.6886792452830195e-05,
+      "loss": 1.8827,
       "step": 33
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 0.00019387387387387388,
-      "loss": 1.6939,
+      "epoch": 0.64,
+      "learning_rate": 4.679245283018868e-05,
+      "loss": 2.1251,
       "step": 34
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 0.0001936936936936937,
-      "loss": 1.7951,
+      "epoch": 0.66,
+      "learning_rate": 4.669811320754717e-05,
+      "loss": 2.1592,
       "step": 35
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 0.00019351351351351353,
-      "loss": 1.8343,
+      "epoch": 0.68,
+      "learning_rate": 4.660377358490566e-05,
+      "loss": 2.2133,
       "step": 36
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 0.00019333333333333333,
-      "loss": 1.8248,
+      "epoch": 0.7,
+      "learning_rate": 4.650943396226415e-05,
+      "loss": 2.1562,
       "step": 37
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 0.00019315315315315316,
-      "loss": 1.7202,
+      "epoch": 0.72,
+      "learning_rate": 4.641509433962264e-05,
+      "loss": 1.9496,
       "step": 38
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 0.000192972972972973,
-      "loss": 1.7938,
+      "epoch": 0.74,
+      "learning_rate": 4.6320754716981134e-05,
+      "loss": 2.0698,
       "step": 39
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 0.00019279279279279282,
-      "loss": 1.7348,
+      "epoch": 0.75,
+      "learning_rate": 4.6226415094339625e-05,
+      "loss": 2.0431,
       "step": 40
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 0.00019261261261261262,
-      "loss": 1.6271,
+      "epoch": 0.77,
+      "learning_rate": 4.6132075471698117e-05,
+      "loss": 2.0038,
       "step": 41
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 0.00019243243243243245,
-      "loss": 1.5305,
+      "epoch": 0.79,
+      "learning_rate": 4.603773584905661e-05,
+      "loss": 2.0959,
       "step": 42
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 0.00019225225225225225,
-      "loss": 1.6801,
+      "epoch": 0.81,
+      "learning_rate": 4.59433962264151e-05,
+      "loss": 1.9298,
       "step": 43
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 0.00019207207207207208,
-      "loss": 1.6942,
+      "epoch": 0.83,
+      "learning_rate": 4.584905660377359e-05,
+      "loss": 1.9915,
       "step": 44
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 0.0001918918918918919,
-      "loss": 1.7218,
+      "epoch": 0.85,
+      "learning_rate": 4.575471698113208e-05,
+      "loss": 1.8004,
       "step": 45
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 0.0001917117117117117,
-      "loss": 1.6827,
+      "epoch": 0.87,
+      "learning_rate": 4.566037735849057e-05,
+      "loss": 1.9061,
       "step": 46
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 0.00019153153153153154,
-      "loss": 1.4275,
+      "epoch": 0.89,
+      "learning_rate": 4.556603773584906e-05,
+      "loss": 1.8904,
       "step": 47
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 0.00019135135135135137,
-      "loss": 1.605,
+      "epoch": 0.91,
+      "learning_rate": 4.547169811320755e-05,
+      "loss": 2.0976,
       "step": 48
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 0.0001911711711711712,
-      "loss": 1.6862,
+      "epoch": 0.92,
+      "learning_rate": 4.537735849056604e-05,
+      "loss": 1.7822,
       "step": 49
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 0.000190990990990991,
-      "loss": 1.529,
+      "epoch": 0.94,
+      "learning_rate": 4.528301886792453e-05,
+      "loss": 1.8883,
       "step": 50
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 0.00019081081081081082,
-      "loss": 1.6986,
+      "epoch": 0.96,
+      "learning_rate": 4.518867924528302e-05,
+      "loss": 1.6874,
       "step": 51
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 0.00019063063063063065,
-      "loss": 1.6081,
+      "epoch": 0.98,
+      "learning_rate": 4.509433962264151e-05,
+      "loss": 1.7476,
       "step": 52
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 0.00019045045045045048,
-      "loss": 1.6748,
+      "epoch": 1.0,
+      "learning_rate": 4.5e-05,
+      "loss": 1.6606,
       "step": 53
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 0.00019027027027027028,
-      "loss": 1.3963,
+      "epoch": 1.02,
+      "learning_rate": 4.4905660377358494e-05,
+      "loss": 1.701,
       "step": 54
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 0.00019009009009009008,
-      "loss": 1.6614,
+      "epoch": 1.04,
+      "learning_rate": 4.4811320754716985e-05,
+      "loss": 1.6084,
       "step": 55
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 0.0001899099099099099,
-      "loss": 1.4796,
+      "epoch": 1.06,
+      "learning_rate": 4.4716981132075476e-05,
+      "loss": 1.6217,
       "step": 56
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 0.00018972972972972974,
-      "loss": 1.4965,
+      "epoch": 1.08,
+      "learning_rate": 4.462264150943397e-05,
+      "loss": 1.5683,
       "step": 57
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 0.00018954954954954957,
-      "loss": 1.5498,
+      "epoch": 1.09,
+      "learning_rate": 4.452830188679246e-05,
+      "loss": 1.6248,
       "step": 58
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 0.00018936936936936937,
-      "loss": 1.5678,
+      "epoch": 1.11,
+      "learning_rate": 4.443396226415095e-05,
+      "loss": 1.6011,
       "step": 59
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 0.0001891891891891892,
-      "loss": 1.4824,
+      "epoch": 1.13,
+      "learning_rate": 4.433962264150944e-05,
+      "loss": 1.6206,
       "step": 60
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 0.00018900900900900903,
-      "loss": 1.4765,
+      "epoch": 1.15,
+      "learning_rate": 4.4245283018867925e-05,
+      "loss": 1.5233,
       "step": 61
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 0.00018882882882882885,
-      "loss": 1.5106,
+      "epoch": 1.17,
+      "learning_rate": 4.4150943396226416e-05,
+      "loss": 1.6248,
       "step": 62
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 0.00018864864864864866,
-      "loss": 1.5217,
+      "epoch": 1.19,
+      "learning_rate": 4.405660377358491e-05,
+      "loss": 1.5203,
       "step": 63
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 0.00018846846846846848,
-      "loss": 1.5239,
+      "epoch": 1.21,
+      "learning_rate": 4.39622641509434e-05,
+      "loss": 1.4521,
       "step": 64
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 0.0001882882882882883,
-      "loss": 1.5392,
+      "epoch": 1.23,
+      "learning_rate": 4.386792452830189e-05,
+      "loss": 1.4511,
       "step": 65
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 0.00018810810810810814,
-      "loss": 1.4498,
+      "epoch": 1.25,
+      "learning_rate": 4.377358490566038e-05,
+      "loss": 1.4902,
       "step": 66
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 0.00018792792792792791,
-      "loss": 1.4997,
+      "epoch": 1.26,
+      "learning_rate": 4.367924528301887e-05,
+      "loss": 1.6299,
       "step": 67
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 0.00018774774774774774,
-      "loss": 1.4124,
+      "epoch": 1.28,
+      "learning_rate": 4.358490566037736e-05,
+      "loss": 1.4494,
       "step": 68
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 0.00018756756756756757,
-      "loss": 1.5549,
+      "epoch": 1.3,
+      "learning_rate": 4.3490566037735853e-05,
+      "loss": 1.4915,
       "step": 69
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 0.0001873873873873874,
-      "loss": 1.431,
+      "epoch": 1.32,
+      "learning_rate": 4.3396226415094345e-05,
+      "loss": 1.5363,
       "step": 70
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 0.0001872072072072072,
-      "loss": 1.3728,
+      "epoch": 1.34,
+      "learning_rate": 4.3301886792452836e-05,
+      "loss": 1.4763,
       "step": 71
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 0.00018702702702702703,
-      "loss": 1.5376,
+      "epoch": 1.36,
+      "learning_rate": 4.320754716981133e-05,
+      "loss": 1.4828,
       "step": 72
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 0.00018684684684684686,
-      "loss": 1.6378,
+      "epoch": 1.38,
+      "learning_rate": 4.311320754716982e-05,
+      "loss": 1.4703,
       "step": 73
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 0.0001866666666666667,
-      "loss": 1.4646,
+      "epoch": 1.4,
+      "learning_rate": 4.301886792452831e-05,
+      "loss": 1.3921,
       "step": 74
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 0.0001864864864864865,
-      "loss": 1.2362,
+      "epoch": 1.42,
+      "learning_rate": 4.292452830188679e-05,
+      "loss": 1.5084,
       "step": 75
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 0.00018630630630630632,
-      "loss": 1.3428,
+      "epoch": 1.43,
+      "learning_rate": 4.2830188679245284e-05,
+      "loss": 1.4489,
       "step": 76
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 0.00018612612612612614,
-      "loss": 1.4464,
+      "epoch": 1.45,
+      "learning_rate": 4.2735849056603775e-05,
+      "loss": 1.479,
       "step": 77
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 0.00018594594594594597,
-      "loss": 1.4192,
+      "epoch": 1.47,
+      "learning_rate": 4.2641509433962266e-05,
+      "loss": 1.4791,
       "step": 78
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 0.00018576576576576577,
-      "loss": 1.4174,
+      "epoch": 1.49,
+      "learning_rate": 4.254716981132076e-05,
+      "loss": 1.3819,
       "step": 79
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 0.00018558558558558558,
-      "loss": 1.5149,
+      "epoch": 1.51,
+      "learning_rate": 4.245283018867925e-05,
+      "loss": 1.4143,
       "step": 80
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 0.0001854054054054054,
-      "loss": 1.504,
+      "epoch": 1.53,
+      "learning_rate": 4.235849056603774e-05,
+      "loss": 1.4606,
       "step": 81
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 0.00018522522522522523,
-      "loss": 1.4452,
+      "epoch": 1.55,
+      "learning_rate": 4.226415094339623e-05,
+      "loss": 1.4457,
       "step": 82
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 0.00018504504504504506,
-      "loss": 1.5763,
+      "epoch": 1.57,
+      "learning_rate": 4.216981132075472e-05,
+      "loss": 1.3819,
       "step": 83
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 0.00018486486486486486,
-      "loss": 1.4288,
+      "epoch": 1.58,
+      "learning_rate": 4.207547169811321e-05,
+      "loss": 1.3984,
       "step": 84
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 0.0001846846846846847,
-      "loss": 1.5056,
+      "epoch": 1.6,
+      "learning_rate": 4.1981132075471704e-05,
+      "loss": 1.4533,
       "step": 85
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 0.00018450450450450452,
-      "loss": 1.3584,
+      "epoch": 1.62,
+      "learning_rate": 4.1886792452830195e-05,
+      "loss": 1.4731,
       "step": 86
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 0.00018432432432432435,
-      "loss": 1.2679,
+      "epoch": 1.64,
+      "learning_rate": 4.1792452830188686e-05,
+      "loss": 1.4723,
       "step": 87
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 0.00018414414414414415,
-      "loss": 1.2421,
+      "epoch": 1.66,
+      "learning_rate": 4.169811320754717e-05,
+      "loss": 1.4317,
       "step": 88
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 0.00018396396396396398,
-      "loss": 1.2359,
+      "epoch": 1.68,
+      "learning_rate": 4.160377358490566e-05,
+      "loss": 1.441,
       "step": 89
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 0.0001837837837837838,
-      "loss": 1.2961,
+      "epoch": 1.7,
+      "learning_rate": 4.150943396226415e-05,
+      "loss": 1.3809,
       "step": 90
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 0.0001836036036036036,
-      "loss": 1.5433,
+      "epoch": 1.72,
+      "learning_rate": 4.1415094339622644e-05,
+      "loss": 1.2942,
       "step": 91
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 0.00018342342342342343,
-      "loss": 1.3349,
+      "epoch": 1.74,
+      "learning_rate": 4.1320754716981135e-05,
+      "loss": 1.4409,
       "step": 92
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 0.00018324324324324324,
-      "loss": 1.3233,
+      "epoch": 1.75,
+      "learning_rate": 4.1226415094339626e-05,
+      "loss": 1.5428,
       "step": 93
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 0.00018306306306306306,
-      "loss": 1.4935,
+      "epoch": 1.77,
+      "learning_rate": 4.113207547169812e-05,
+      "loss": 1.3872,
       "step": 94
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 0.0001828828828828829,
-      "loss": 1.4808,
+      "epoch": 1.79,
+      "learning_rate": 4.103773584905661e-05,
+      "loss": 1.4422,
       "step": 95
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 0.00018270270270270272,
-      "loss": 1.3105,
+      "epoch": 1.81,
+      "learning_rate": 4.09433962264151e-05,
+      "loss": 1.5333,
       "step": 96
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 0.00018252252252252252,
-      "loss": 1.6545,
+      "epoch": 1.83,
+      "learning_rate": 4.084905660377359e-05,
+      "loss": 1.229,
       "step": 97
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 0.00018234234234234235,
-      "loss": 1.5198,
+      "epoch": 1.85,
+      "learning_rate": 4.075471698113208e-05,
+      "loss": 1.3927,
       "step": 98
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 0.00018216216216216218,
-      "loss": 1.2939,
+      "epoch": 1.87,
+      "learning_rate": 4.066037735849057e-05,
+      "loss": 1.2784,
       "step": 99
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 0.000181981981981982,
-      "loss": 1.4136,
+      "epoch": 1.89,
+      "learning_rate": 4.0566037735849064e-05,
+      "loss": 1.3527,
       "step": 100
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 0.0001818018018018018,
-      "loss": 1.2831,
+      "epoch": 1.91,
+      "learning_rate": 4.047169811320755e-05,
+      "loss": 1.422,
       "step": 101
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 0.00018162162162162164,
-      "loss": 1.2908,
+      "epoch": 1.92,
+      "learning_rate": 4.037735849056604e-05,
+      "loss": 1.3688,
       "step": 102
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 0.00018144144144144144,
-      "loss": 1.2553,
+      "epoch": 1.94,
+      "learning_rate": 4.028301886792453e-05,
+      "loss": 1.3664,
       "step": 103
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 0.00018126126126126127,
-      "loss": 1.2482,
+      "epoch": 1.96,
+      "learning_rate": 4.018867924528302e-05,
+      "loss": 1.3683,
       "step": 104
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 0.0001810810810810811,
-      "loss": 1.3024,
+      "epoch": 1.98,
+      "learning_rate": 4.009433962264151e-05,
+      "loss": 1.3948,
       "step": 105
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 0.0001809009009009009,
-      "loss": 1.3831,
+      "epoch": 2.0,
+      "learning_rate": 4e-05,
+      "loss": 1.258,
       "step": 106
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 0.00018072072072072072,
-      "loss": 1.2204,
+      "epoch": 2.02,
+      "learning_rate": 3.9905660377358494e-05,
+      "loss": 1.3721,
       "step": 107
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 0.00018054054054054055,
-      "loss": 1.193,
+      "epoch": 2.04,
+      "learning_rate": 3.9811320754716985e-05,
+      "loss": 1.2634,
       "step": 108
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 0.00018036036036036038,
-      "loss": 1.2178,
+      "epoch": 2.06,
+      "learning_rate": 3.9716981132075477e-05,
+      "loss": 1.3369,
       "step": 109
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 0.00018018018018018018,
-      "loss": 1.3477,
+      "epoch": 2.08,
+      "learning_rate": 3.962264150943397e-05,
+      "loss": 1.2034,
       "step": 110
     },
     {
-      "epoch": 1.0,
-      "learning_rate": 0.00018,
-      "loss": 1.848,
+      "epoch": 2.09,
+      "learning_rate": 3.952830188679246e-05,
+      "loss": 1.2868,
       "step": 111
     },
     {
-      "epoch": 1.01,
-      "learning_rate": 0.00017981981981981984,
-      "loss": 1.0224,
+      "epoch": 2.11,
+      "learning_rate": 3.943396226415095e-05,
+      "loss": 1.2322,
       "step": 112
     },
     {
-      "epoch": 1.02,
-      "learning_rate": 0.00017963963963963967,
-      "loss": 1.0316,
+      "epoch": 2.13,
+      "learning_rate": 3.933962264150944e-05,
+      "loss": 1.2472,
       "step": 113
     },
     {
-      "epoch": 1.03,
-      "learning_rate": 0.00017945945945945947,
-      "loss": 0.9298,
+      "epoch": 2.15,
+      "learning_rate": 3.924528301886793e-05,
+      "loss": 1.1504,
       "step": 114
     },
     {
-      "epoch": 1.04,
-      "learning_rate": 0.00017927927927927927,
-      "loss": 0.9607,
+      "epoch": 2.17,
+      "learning_rate": 3.9150943396226416e-05,
+      "loss": 1.2511,
       "step": 115
     },
     {
-      "epoch": 1.05,
-      "learning_rate": 0.0001790990990990991,
-      "loss": 0.9851,
+      "epoch": 2.19,
+      "learning_rate": 3.905660377358491e-05,
+      "loss": 1.1804,
       "step": 116
     },
     {
-      "epoch": 1.05,
-      "learning_rate": 0.00017891891891891893,
-      "loss": 1.1184,
+      "epoch": 2.21,
+      "learning_rate": 3.89622641509434e-05,
+      "loss": 1.0368,
       "step": 117
     },
     {
-      "epoch": 1.06,
-      "learning_rate": 0.00017873873873873876,
-      "loss": 1.1724,
+      "epoch": 2.23,
+      "learning_rate": 3.886792452830189e-05,
+      "loss": 1.168,
       "step": 118
     },
     {
-      "epoch": 1.07,
-      "learning_rate": 0.00017855855855855856,
-      "loss": 0.9032,
+      "epoch": 2.25,
+      "learning_rate": 3.877358490566038e-05,
+      "loss": 1.2403,
       "step": 119
     },
     {
-      "epoch": 1.08,
-      "learning_rate": 0.00017837837837837839,
-      "loss": 1.0655,
+      "epoch": 2.26,
+      "learning_rate": 3.867924528301887e-05,
+      "loss": 1.2023,
       "step": 120
     },
     {
-      "epoch": 1.09,
-      "learning_rate": 0.00017819819819819821,
-      "loss": 0.9477,
+      "epoch": 2.28,
+      "learning_rate": 3.858490566037736e-05,
+      "loss": 1.2519,
       "step": 121
     },
     {
-      "epoch": 1.1,
-      "learning_rate": 0.00017801801801801804,
-      "loss": 1.0931,
+      "epoch": 2.3,
+      "learning_rate": 3.8490566037735854e-05,
+      "loss": 1.1993,
       "step": 122
     },
     {
-      "epoch": 1.11,
-      "learning_rate": 0.00017783783783783784,
-      "loss": 1.1222,
+      "epoch": 2.32,
+      "learning_rate": 3.8396226415094345e-05,
+      "loss": 1.1572,
       "step": 123
     },
     {
-      "epoch": 1.12,
-      "learning_rate": 0.00017765765765765767,
-      "loss": 1.0831,
+      "epoch": 2.34,
+      "learning_rate": 3.8301886792452836e-05,
+      "loss": 1.2227,
       "step": 124
     },
     {
-      "epoch": 1.13,
-      "learning_rate": 0.0001774774774774775,
-      "loss": 0.996,
+      "epoch": 2.36,
+      "learning_rate": 3.820754716981133e-05,
+      "loss": 1.1286,
       "step": 125
     },
     {
-      "epoch": 1.14,
-      "learning_rate": 0.00017729729729729733,
-      "loss": 1.1083,
+      "epoch": 2.38,
+      "learning_rate": 3.811320754716982e-05,
+      "loss": 1.2267,
       "step": 126
     },
     {
-      "epoch": 1.14,
-      "learning_rate": 0.0001771171171171171,
-      "loss": 1.0783,
+      "epoch": 2.4,
+      "learning_rate": 3.801886792452831e-05,
+      "loss": 1.1177,
       "step": 127
     },
     {
-      "epoch": 1.15,
-      "learning_rate": 0.00017693693693693693,
-      "loss": 1.0126,
+      "epoch": 2.42,
+      "learning_rate": 3.7924528301886794e-05,
+      "loss": 1.2352,
       "step": 128
     },
     {
-      "epoch": 1.16,
-      "learning_rate": 0.00017675675675675676,
-      "loss": 1.0566,
+      "epoch": 2.43,
+      "learning_rate": 3.7830188679245285e-05,
+      "loss": 1.1686,
       "step": 129
     },
     {
-      "epoch": 1.17,
-      "learning_rate": 0.0001765765765765766,
-      "loss": 1.1383,
+      "epoch": 2.45,
+      "learning_rate": 3.7735849056603776e-05,
+      "loss": 1.1913,
       "step": 130
     },
     {
-      "epoch": 1.18,
-      "learning_rate": 0.0001763963963963964,
-      "loss": 1.0675,
+      "epoch": 2.47,
+      "learning_rate": 3.764150943396227e-05,
+      "loss": 1.1315,
       "step": 131
     },
     {
-      "epoch": 1.19,
-      "learning_rate": 0.00017621621621621622,
-      "loss": 0.9634,
+      "epoch": 2.49,
+      "learning_rate": 3.754716981132076e-05,
+      "loss": 1.1151,
       "step": 132
     },
     {
-      "epoch": 1.2,
-      "learning_rate": 0.00017603603603603605,
-      "loss": 0.9781,
+      "epoch": 2.51,
+      "learning_rate": 3.745283018867924e-05,
+      "loss": 1.1532,
       "step": 133
     },
     {
-      "epoch": 1.21,
-      "learning_rate": 0.00017585585585585587,
-      "loss": 1.0291,
+      "epoch": 2.53,
+      "learning_rate": 3.735849056603773e-05,
+      "loss": 1.1501,
       "step": 134
     },
     {
-      "epoch": 1.22,
-      "learning_rate": 0.00017567567567567568,
-      "loss": 1.0779,
+      "epoch": 2.55,
+      "learning_rate": 3.7264150943396224e-05,
+      "loss": 1.161,
       "step": 135
     },
     {
-      "epoch": 1.23,
-      "learning_rate": 0.0001754954954954955,
-      "loss": 0.946,
+      "epoch": 2.57,
+      "learning_rate": 3.7169811320754716e-05,
+      "loss": 1.1592,
       "step": 136
     },
     {
-      "epoch": 1.23,
-      "learning_rate": 0.00017531531531531533,
-      "loss": 1.1033,
+      "epoch": 2.58,
+      "learning_rate": 3.7075471698113207e-05,
+      "loss": 1.1969,
       "step": 137
     },
     {
-      "epoch": 1.24,
-      "learning_rate": 0.00017513513513513516,
-      "loss": 1.1328,
+      "epoch": 2.6,
+      "learning_rate": 3.69811320754717e-05,
+      "loss": 1.1805,
       "step": 138
     },
     {
-      "epoch": 1.25,
-      "learning_rate": 0.00017495495495495496,
-      "loss": 1.014,
+      "epoch": 2.62,
+      "learning_rate": 3.688679245283019e-05,
+      "loss": 1.0553,
       "step": 139
     },
     {
-      "epoch": 1.26,
-      "learning_rate": 0.00017477477477477476,
-      "loss": 1.0098,
+      "epoch": 2.64,
+      "learning_rate": 3.679245283018868e-05,
+      "loss": 1.1072,
       "step": 140
     },
     {
-      "epoch": 1.27,
-      "learning_rate": 0.0001745945945945946,
-      "loss": 1.0973,
+      "epoch": 2.66,
+      "learning_rate": 3.669811320754717e-05,
+      "loss": 1.234,
       "step": 141
     },
     {
-      "epoch": 1.28,
-      "learning_rate": 0.00017441441441441442,
-      "loss": 1.0991,
+      "epoch": 2.68,
+      "learning_rate": 3.660377358490566e-05,
+      "loss": 1.1156,
       "step": 142
     },
     {
-      "epoch": 1.29,
-      "learning_rate": 0.00017423423423423425,
-      "loss": 0.9668,
+      "epoch": 2.7,
+      "learning_rate": 3.650943396226415e-05,
+      "loss": 1.1624,
       "step": 143
     },
     {
-      "epoch": 1.3,
-      "learning_rate": 0.00017405405405405405,
-      "loss": 0.8774,
+      "epoch": 2.72,
+      "learning_rate": 3.641509433962264e-05,
+      "loss": 1.1259,
       "step": 144
     },
     {
-      "epoch": 1.31,
-      "learning_rate": 0.00017387387387387388,
-      "loss": 1.0744,
+      "epoch": 2.74,
+      "learning_rate": 3.632075471698113e-05,
+      "loss": 1.1218,
       "step": 145
     },
     {
-      "epoch": 1.32,
-      "learning_rate": 0.0001736936936936937,
-      "loss": 0.8432,
+      "epoch": 2.75,
+      "learning_rate": 3.622641509433962e-05,
+      "loss": 1.133,
       "step": 146
     },
     {
-      "epoch": 1.32,
-      "learning_rate": 0.00017351351351351353,
-      "loss": 0.9951,
+      "epoch": 2.77,
+      "learning_rate": 3.613207547169811e-05,
+      "loss": 1.2087,
       "step": 147
     },
     {
-      "epoch": 1.33,
-      "learning_rate": 0.00017333333333333334,
-      "loss": 1.0802,
+      "epoch": 2.79,
+      "learning_rate": 3.60377358490566e-05,
+      "loss": 1.1558,
       "step": 148
     },
     {
-      "epoch": 1.34,
-      "learning_rate": 0.00017315315315315316,
-      "loss": 1.1801,
+      "epoch": 2.81,
+      "learning_rate": 3.594339622641509e-05,
+      "loss": 1.1518,
       "step": 149
     },
     {
-      "epoch": 1.35,
-      "learning_rate": 0.000172972972972973,
-      "loss": 0.9059,
+      "epoch": 2.83,
+      "learning_rate": 3.5849056603773584e-05,
+      "loss": 1.2106,
       "step": 150
     },
     {
-      "epoch": 1.36,
-      "learning_rate": 0.0001727927927927928,
-      "loss": 0.9663,
+      "epoch": 2.85,
+      "learning_rate": 3.5754716981132075e-05,
+      "loss": 1.0969,
       "step": 151
     },
     {
-      "epoch": 1.37,
-      "learning_rate": 0.00017261261261261262,
-      "loss": 1.1322,
+      "epoch": 2.87,
+      "learning_rate": 3.5660377358490566e-05,
+      "loss": 1.0236,
       "step": 152
     },
     {
-      "epoch": 1.38,
-      "learning_rate": 0.00017243243243243242,
-      "loss": 1.082,
+      "epoch": 2.89,
+      "learning_rate": 3.556603773584906e-05,
+      "loss": 1.0231,
       "step": 153
     },
     {
-      "epoch": 1.39,
-      "learning_rate": 0.00017225225225225225,
-      "loss": 0.9189,
+      "epoch": 2.91,
+      "learning_rate": 3.547169811320755e-05,
+      "loss": 1.0921,
       "step": 154
     },
     {
-      "epoch": 1.4,
-      "learning_rate": 0.00017207207207207208,
-      "loss": 1.0324,
+      "epoch": 2.92,
+      "learning_rate": 3.537735849056604e-05,
+      "loss": 1.1057,
       "step": 155
     },
     {
-      "epoch": 1.41,
-      "learning_rate": 0.0001718918918918919,
-      "loss": 0.9904,
+      "epoch": 2.94,
+      "learning_rate": 3.528301886792453e-05,
+      "loss": 1.121,
       "step": 156
     },
     {
-      "epoch": 1.41,
-      "learning_rate": 0.0001717117117117117,
-      "loss": 1.1135,
+      "epoch": 2.96,
+      "learning_rate": 3.518867924528302e-05,
+      "loss": 1.1201,
       "step": 157
     },
     {
-      "epoch": 1.42,
-      "learning_rate": 0.00017153153153153154,
-      "loss": 0.9649,
+      "epoch": 2.98,
+      "learning_rate": 3.5094339622641506e-05,
+      "loss": 1.0761,
       "step": 158
     },
     {
-      "epoch": 1.43,
-      "learning_rate": 0.00017135135135135137,
-      "loss": 1.0241,
+      "epoch": 3.0,
+      "learning_rate": 3.5e-05,
+      "loss": 1.1364,
       "step": 159
     },
     {
-      "epoch": 1.44,
-      "learning_rate": 0.0001711711711711712,
-      "loss": 0.8952,
+      "epoch": 3.02,
+      "learning_rate": 3.490566037735849e-05,
+      "loss": 1.0045,
       "step": 160
     },
     {
-      "epoch": 1.45,
-      "learning_rate": 0.000170990990990991,
-      "loss": 0.9294,
+      "epoch": 3.04,
+      "learning_rate": 3.481132075471698e-05,
+      "loss": 1.082,
       "step": 161
     },
     {
-      "epoch": 1.46,
-      "learning_rate": 0.00017081081081081083,
-      "loss": 1.0734,
+      "epoch": 3.06,
+      "learning_rate": 3.471698113207547e-05,
+      "loss": 1.0907,
       "step": 162
     },
     {
-      "epoch": 1.47,
-      "learning_rate": 0.00017063063063063063,
-      "loss": 1.075,
+      "epoch": 3.08,
+      "learning_rate": 3.462264150943396e-05,
+      "loss": 1.0209,
       "step": 163
     },
     {
-      "epoch": 1.48,
-      "learning_rate": 0.00017045045045045045,
-      "loss": 0.9994,
+      "epoch": 3.09,
+      "learning_rate": 3.452830188679245e-05,
+      "loss": 1.1183,
       "step": 164
     },
     {
-      "epoch": 1.49,
-      "learning_rate": 0.00017027027027027028,
-      "loss": 0.8595,
+      "epoch": 3.11,
+      "learning_rate": 3.4433962264150943e-05,
+      "loss": 1.0522,
       "step": 165
     },
     {
-      "epoch": 1.5,
-      "learning_rate": 0.00017009009009009008,
-      "loss": 0.9343,
+      "epoch": 3.13,
+      "learning_rate": 3.4339622641509435e-05,
+      "loss": 1.1326,
       "step": 166
     },
     {
-      "epoch": 1.5,
-      "learning_rate": 0.0001699099099099099,
-      "loss": 1.0496,
+      "epoch": 3.15,
+      "learning_rate": 3.4245283018867926e-05,
+      "loss": 1.0603,
       "step": 167
     },
     {
-      "epoch": 1.51,
-      "learning_rate": 0.00016972972972972974,
-      "loss": 0.9934,
+      "epoch": 3.17,
+      "learning_rate": 3.415094339622642e-05,
+      "loss": 1.0376,
       "step": 168
     },
     {
-      "epoch": 1.52,
-      "learning_rate": 0.00016954954954954957,
-      "loss": 0.9766,
+      "epoch": 3.19,
+      "learning_rate": 3.405660377358491e-05,
+      "loss": 0.9987,
       "step": 169
     },
     {
-      "epoch": 1.53,
-      "learning_rate": 0.00016936936936936937,
-      "loss": 0.8376,
+      "epoch": 3.21,
+      "learning_rate": 3.39622641509434e-05,
+      "loss": 0.959,
       "step": 170
     },
     {
-      "epoch": 1.54,
-      "learning_rate": 0.0001691891891891892,
-      "loss": 0.9374,
+      "epoch": 3.23,
+      "learning_rate": 3.386792452830188e-05,
+      "loss": 1.0101,
       "step": 171
     },
     {
-      "epoch": 1.55,
-      "learning_rate": 0.00016900900900900903,
-      "loss": 1.0476,
+      "epoch": 3.25,
+      "learning_rate": 3.3773584905660374e-05,
+      "loss": 0.9348,
       "step": 172
     },
     {
-      "epoch": 1.56,
-      "learning_rate": 0.00016882882882882886,
-      "loss": 0.9219,
+      "epoch": 3.26,
+      "learning_rate": 3.3679245283018865e-05,
+      "loss": 0.9486,
       "step": 173
     },
     {
-      "epoch": 1.57,
-      "learning_rate": 0.00016864864864864866,
-      "loss": 0.9586,
+      "epoch": 3.28,
+      "learning_rate": 3.3584905660377356e-05,
+      "loss": 1.0116,
       "step": 174
     },
     {
-      "epoch": 1.58,
-      "learning_rate": 0.00016846846846846846,
-      "loss": 1.0288,
+      "epoch": 3.3,
+      "learning_rate": 3.349056603773585e-05,
+      "loss": 0.9613,
       "step": 175
     },
     {
-      "epoch": 1.59,
-      "learning_rate": 0.0001682882882882883,
-      "loss": 0.9829,
+      "epoch": 3.32,
+      "learning_rate": 3.339622641509434e-05,
+      "loss": 0.9316,
       "step": 176
     },
     {
-      "epoch": 1.59,
-      "learning_rate": 0.00016810810810810812,
-      "loss": 0.8158,
+      "epoch": 3.34,
+      "learning_rate": 3.330188679245283e-05,
+      "loss": 0.9969,
       "step": 177
     },
     {
-      "epoch": 1.6,
-      "learning_rate": 0.00016792792792792794,
-      "loss": 1.0338,
+      "epoch": 3.36,
+      "learning_rate": 3.320754716981132e-05,
+      "loss": 0.8827,
       "step": 178
     },
     {
-      "epoch": 1.61,
-      "learning_rate": 0.00016774774774774775,
-      "loss": 1.0119,
+      "epoch": 3.38,
+      "learning_rate": 3.311320754716981e-05,
+      "loss": 1.0186,
       "step": 179
     },
     {
-      "epoch": 1.62,
-      "learning_rate": 0.00016756756756756757,
-      "loss": 0.7975,
+      "epoch": 3.4,
+      "learning_rate": 3.30188679245283e-05,
+      "loss": 0.914,
       "step": 180
     },
     {
-      "epoch": 1.63,
-      "learning_rate": 0.0001673873873873874,
-      "loss": 0.8601,
+      "epoch": 3.42,
+      "learning_rate": 3.2924528301886794e-05,
+      "loss": 0.9294,
       "step": 181
     },
     {
-      "epoch": 1.64,
-      "learning_rate": 0.00016720720720720723,
-      "loss": 1.0333,
+      "epoch": 3.43,
+      "learning_rate": 3.2830188679245285e-05,
+      "loss": 0.9551,
       "step": 182
     },
     {
-      "epoch": 1.65,
-      "learning_rate": 0.00016702702702702703,
-      "loss": 1.0862,
+      "epoch": 3.45,
+      "learning_rate": 3.2735849056603776e-05,
+      "loss": 1.013,
       "step": 183
     },
     {
-      "epoch": 1.66,
-      "learning_rate": 0.00016684684684684686,
-      "loss": 0.8784,
+      "epoch": 3.47,
+      "learning_rate": 3.264150943396227e-05,
+      "loss": 0.9576,
       "step": 184
     },
     {
-      "epoch": 1.67,
-      "learning_rate": 0.0001666666666666667,
-      "loss": 1.0995,
+      "epoch": 3.49,
+      "learning_rate": 3.254716981132075e-05,
+      "loss": 0.9847,
       "step": 185
     },
     {
-      "epoch": 1.68,
-      "learning_rate": 0.00016648648648648652,
-      "loss": 1.0452,
+      "epoch": 3.51,
+      "learning_rate": 3.245283018867924e-05,
+      "loss": 1.0326,
       "step": 186
     },
     {
-      "epoch": 1.68,
-      "learning_rate": 0.00016630630630630632,
-      "loss": 0.8971,
+      "epoch": 3.53,
+      "learning_rate": 3.2358490566037734e-05,
+      "loss": 0.9624,
       "step": 187
     },
     {
-      "epoch": 1.69,
-      "learning_rate": 0.00016612612612612612,
-      "loss": 0.9339,
+      "epoch": 3.55,
+      "learning_rate": 3.2264150943396225e-05,
+      "loss": 1.0267,
       "step": 188
     },
     {
-      "epoch": 1.7,
-      "learning_rate": 0.00016594594594594595,
-      "loss": 0.8219,
+      "epoch": 3.57,
+      "learning_rate": 3.2169811320754716e-05,
+      "loss": 1.039,
       "step": 189
     },
     {
-      "epoch": 1.71,
-      "learning_rate": 0.00016576576576576578,
-      "loss": 0.9829,
+      "epoch": 3.58,
+      "learning_rate": 3.207547169811321e-05,
+      "loss": 1.1197,
       "step": 190
     },
     {
-      "epoch": 1.72,
-      "learning_rate": 0.00016558558558558558,
-      "loss": 0.7705,
+      "epoch": 3.6,
+      "learning_rate": 3.19811320754717e-05,
+      "loss": 0.8363,
       "step": 191
     },
     {
-      "epoch": 1.73,
-      "learning_rate": 0.0001654054054054054,
-      "loss": 0.8442,
+      "epoch": 3.62,
+      "learning_rate": 3.188679245283019e-05,
+      "loss": 0.9551,
       "step": 192
     },
     {
-      "epoch": 1.74,
-      "learning_rate": 0.00016522522522522523,
-      "loss": 0.8804,
+      "epoch": 3.64,
+      "learning_rate": 3.179245283018868e-05,
+      "loss": 1.0216,
       "step": 193
     },
     {
-      "epoch": 1.75,
-      "learning_rate": 0.00016504504504504506,
-      "loss": 0.7614,
+      "epoch": 3.66,
+      "learning_rate": 3.169811320754717e-05,
+      "loss": 1.0711,
       "step": 194
     },
     {
-      "epoch": 1.76,
-      "learning_rate": 0.00016486486486486486,
-      "loss": 0.8259,
+      "epoch": 3.68,
+      "learning_rate": 3.160377358490566e-05,
+      "loss": 1.0057,
       "step": 195
     },
     {
-      "epoch": 1.77,
-      "learning_rate": 0.0001646846846846847,
-      "loss": 1.1247,
+      "epoch": 3.7,
+      "learning_rate": 3.1509433962264154e-05,
+      "loss": 0.9545,
       "step": 196
     },
     {
-      "epoch": 1.77,
-      "learning_rate": 0.00016450450450450452,
-      "loss": 0.8855,
+      "epoch": 3.72,
+      "learning_rate": 3.1415094339622645e-05,
+      "loss": 0.9554,
       "step": 197
     },
     {
-      "epoch": 1.78,
-      "learning_rate": 0.00016432432432432435,
-      "loss": 0.9202,
+      "epoch": 3.74,
+      "learning_rate": 3.132075471698113e-05,
+      "loss": 0.9069,
       "step": 198
     },
     {
-      "epoch": 1.79,
-      "learning_rate": 0.00016414414414414415,
-      "loss": 0.9203,
+      "epoch": 3.75,
+      "learning_rate": 3.122641509433962e-05,
+      "loss": 1.0332,
       "step": 199
     },
     {
-      "epoch": 1.8,
-      "learning_rate": 0.00016396396396396395,
-      "loss": 0.8537,
+      "epoch": 3.77,
+      "learning_rate": 3.113207547169811e-05,
+      "loss": 1.0513,
       "step": 200
     },
     {
-      "epoch": 1.81,
-      "learning_rate": 0.00016378378378378378,
-      "loss": 0.8746,
+      "epoch": 3.79,
+      "learning_rate": 3.10377358490566e-05,
+      "loss": 0.952,
       "step": 201
     },
     {
-      "epoch": 1.82,
-      "learning_rate": 0.0001636036036036036,
-      "loss": 1.0242,
+      "epoch": 3.81,
+      "learning_rate": 3.094339622641509e-05,
+      "loss": 1.0413,
       "step": 202
     },
     {
-      "epoch": 1.83,
-      "learning_rate": 0.00016342342342342344,
-      "loss": 0.8806,
+      "epoch": 3.83,
+      "learning_rate": 3.0849056603773584e-05,
+      "loss": 0.8725,
       "step": 203
     },
     {
-      "epoch": 1.84,
-      "learning_rate": 0.00016324324324324324,
-      "loss": 0.8827,
+      "epoch": 3.85,
+      "learning_rate": 3.0754716981132075e-05,
+      "loss": 1.0077,
       "step": 204
     },
     {
-      "epoch": 1.85,
-      "learning_rate": 0.00016306306306306307,
-      "loss": 0.9695,
+      "epoch": 3.87,
+      "learning_rate": 3.0660377358490567e-05,
+      "loss": 0.9643,
       "step": 205
     },
     {
-      "epoch": 1.86,
-      "learning_rate": 0.0001628828828828829,
-      "loss": 0.9495,
+      "epoch": 3.89,
+      "learning_rate": 3.056603773584906e-05,
+      "loss": 0.9741,
       "step": 206
     },
     {
-      "epoch": 1.86,
-      "learning_rate": 0.00016270270270270272,
-      "loss": 0.7122,
+      "epoch": 3.91,
+      "learning_rate": 3.047169811320755e-05,
+      "loss": 1.0225,
       "step": 207
     },
     {
-      "epoch": 1.87,
-      "learning_rate": 0.00016252252252252252,
-      "loss": 1.0111,
+      "epoch": 3.92,
+      "learning_rate": 3.0377358490566036e-05,
+      "loss": 1.0555,
       "step": 208
     },
     {
-      "epoch": 1.88,
-      "learning_rate": 0.00016234234234234235,
-      "loss": 1.1255,
+      "epoch": 3.94,
+      "learning_rate": 3.0283018867924528e-05,
+      "loss": 0.9697,
       "step": 209
     },
     {
-      "epoch": 1.89,
-      "learning_rate": 0.00016216216216216218,
-      "loss": 0.8789,
+      "epoch": 3.96,
+      "learning_rate": 3.018867924528302e-05,
+      "loss": 0.9397,
       "step": 210
     },
     {
-      "epoch": 1.9,
-      "learning_rate": 0.000161981981981982,
-      "loss": 0.9309,
+      "epoch": 3.98,
+      "learning_rate": 3.009433962264151e-05,
+      "loss": 0.9053,
       "step": 211
     },
     {
-      "epoch": 1.91,
-      "learning_rate": 0.0001618018018018018,
-      "loss": 0.9133,
+      "epoch": 4.0,
+      "learning_rate": 3e-05,
+      "loss": 1.1121,
       "step": 212
     },
     {
-      "epoch": 1.92,
-      "learning_rate": 0.0001616216216216216,
-      "loss": 0.9779,
+      "epoch": 4.02,
+      "learning_rate": 2.9905660377358492e-05,
+      "loss": 0.8519,
       "step": 213
     },
     {
-      "epoch": 1.93,
-      "learning_rate": 0.00016144144144144144,
-      "loss": 0.8261,
+      "epoch": 4.04,
+      "learning_rate": 2.9811320754716983e-05,
+      "loss": 0.967,
       "step": 214
     },
     {
-      "epoch": 1.94,
-      "learning_rate": 0.00016126126126126127,
-      "loss": 0.8621,
+      "epoch": 4.06,
+      "learning_rate": 2.971698113207547e-05,
+      "loss": 0.8814,
       "step": 215
     },
     {
-      "epoch": 1.95,
-      "learning_rate": 0.0001610810810810811,
-      "loss": 0.7617,
+      "epoch": 4.08,
+      "learning_rate": 2.9622641509433962e-05,
+      "loss": 1.0404,
       "step": 216
     },
     {
-      "epoch": 1.95,
-      "learning_rate": 0.0001609009009009009,
-      "loss": 0.9391,
+      "epoch": 4.09,
+      "learning_rate": 2.9528301886792453e-05,
+      "loss": 0.8551,
       "step": 217
     },
     {
-      "epoch": 1.96,
-      "learning_rate": 0.00016072072072072073,
-      "loss": 0.8437,
+      "epoch": 4.11,
+      "learning_rate": 2.9433962264150944e-05,
+      "loss": 0.8671,
       "step": 218
     },
     {
-      "epoch": 1.97,
-      "learning_rate": 0.00016054054054054056,
-      "loss": 0.9364,
+      "epoch": 4.13,
+      "learning_rate": 2.9339622641509435e-05,
+      "loss": 0.8446,
       "step": 219
     },
     {
-      "epoch": 1.98,
-      "learning_rate": 0.00016036036036036038,
-      "loss": 0.8519,
+      "epoch": 4.15,
+      "learning_rate": 2.9245283018867926e-05,
+      "loss": 0.8838,
       "step": 220
     },
     {
-      "epoch": 1.99,
-      "learning_rate": 0.00016018018018018018,
-      "loss": 0.8703,
+      "epoch": 4.17,
+      "learning_rate": 2.9150943396226417e-05,
+      "loss": 0.8685,
       "step": 221
     },
     {
-      "epoch": 2.0,
-      "learning_rate": 0.00016,
-      "loss": 0.5854,
+      "epoch": 4.19,
+      "learning_rate": 2.9056603773584905e-05,
+      "loss": 0.982,
       "step": 222
     },
     {
-      "epoch": 2.01,
-      "learning_rate": 0.00015981981981981984,
-      "loss": 0.6418,
+      "epoch": 4.21,
+      "learning_rate": 2.8962264150943396e-05,
+      "loss": 0.8266,
       "step": 223
     },
     {
-      "epoch": 2.02,
-      "learning_rate": 0.00015963963963963964,
-      "loss": 0.6642,
+      "epoch": 4.23,
+      "learning_rate": 2.8867924528301887e-05,
+      "loss": 0.8326,
       "step": 224
     },
     {
-      "epoch": 2.03,
-      "learning_rate": 0.00015945945945945947,
-      "loss": 0.7597,
+      "epoch": 4.25,
+      "learning_rate": 2.8773584905660378e-05,
+      "loss": 0.8903,
       "step": 225
     },
     {
-      "epoch": 2.04,
-      "learning_rate": 0.00015927927927927927,
-      "loss": 0.7035,
+      "epoch": 4.26,
+      "learning_rate": 2.867924528301887e-05,
+      "loss": 0.9631,
       "step": 226
     },
     {
-      "epoch": 2.05,
-      "learning_rate": 0.0001590990990990991,
-      "loss": 0.6098,
+      "epoch": 4.28,
+      "learning_rate": 2.858490566037736e-05,
+      "loss": 0.8748,
       "step": 227
     },
     {
-      "epoch": 2.05,
-      "learning_rate": 0.00015891891891891893,
-      "loss": 0.6118,
+      "epoch": 4.3,
+      "learning_rate": 2.8490566037735848e-05,
+      "loss": 0.8452,
       "step": 228
     },
     {
-      "epoch": 2.06,
-      "learning_rate": 0.00015873873873873876,
-      "loss": 0.6998,
+      "epoch": 4.32,
+      "learning_rate": 2.839622641509434e-05,
+      "loss": 0.9581,
       "step": 229
     },
     {
-      "epoch": 2.07,
-      "learning_rate": 0.00015855855855855856,
-      "loss": 0.6526,
+      "epoch": 4.34,
+      "learning_rate": 2.830188679245283e-05,
+      "loss": 0.8077,
       "step": 230
     },
     {
-      "epoch": 2.08,
-      "learning_rate": 0.0001583783783783784,
-      "loss": 0.6019,
+      "epoch": 4.36,
+      "learning_rate": 2.820754716981132e-05,
+      "loss": 0.8741,
       "step": 231
     },
     {
-      "epoch": 2.09,
-      "learning_rate": 0.00015819819819819822,
-      "loss": 0.5906,
+      "epoch": 4.38,
+      "learning_rate": 2.8113207547169812e-05,
+      "loss": 0.8932,
       "step": 232
     },
     {
-      "epoch": 2.1,
-      "learning_rate": 0.00015801801801801804,
-      "loss": 0.6253,
+      "epoch": 4.4,
+      "learning_rate": 2.8018867924528303e-05,
+      "loss": 0.7975,
       "step": 233
     },
     {
-      "epoch": 2.11,
-      "learning_rate": 0.00015783783783783785,
-      "loss": 0.6721,
+      "epoch": 4.42,
+      "learning_rate": 2.7924528301886794e-05,
+      "loss": 0.9653,
       "step": 234
     },
     {
-      "epoch": 2.12,
-      "learning_rate": 0.00015765765765765767,
-      "loss": 0.5985,
+      "epoch": 4.43,
+      "learning_rate": 2.7830188679245282e-05,
+      "loss": 0.9458,
       "step": 235
     },
     {
-      "epoch": 2.13,
-      "learning_rate": 0.00015747747747747747,
-      "loss": 0.6828,
+      "epoch": 4.45,
+      "learning_rate": 2.7735849056603773e-05,
+      "loss": 0.91,
       "step": 236
     },
     {
-      "epoch": 2.14,
-      "learning_rate": 0.0001572972972972973,
-      "loss": 0.5669,
+      "epoch": 4.47,
+      "learning_rate": 2.7641509433962264e-05,
+      "loss": 0.871,
       "step": 237
     },
     {
-      "epoch": 2.14,
-      "learning_rate": 0.00015711711711711713,
-      "loss": 0.7135,
+      "epoch": 4.49,
+      "learning_rate": 2.7547169811320755e-05,
+      "loss": 0.8209,
       "step": 238
     },
     {
-      "epoch": 2.15,
-      "learning_rate": 0.00015693693693693693,
-      "loss": 0.5734,
+      "epoch": 4.51,
+      "learning_rate": 2.7452830188679247e-05,
+      "loss": 0.8097,
       "step": 239
     },
     {
-      "epoch": 2.16,
-      "learning_rate": 0.00015675675675675676,
-      "loss": 0.7206,
+      "epoch": 4.53,
+      "learning_rate": 2.7358490566037738e-05,
+      "loss": 0.7661,
       "step": 240
     },
     {
-      "epoch": 2.17,
-      "learning_rate": 0.0001565765765765766,
-      "loss": 0.7708,
+      "epoch": 4.55,
+      "learning_rate": 2.726415094339623e-05,
+      "loss": 0.895,
       "step": 241
     },
     {
-      "epoch": 2.18,
-      "learning_rate": 0.00015639639639639642,
-      "loss": 0.744,
+      "epoch": 4.57,
+      "learning_rate": 2.7169811320754716e-05,
+      "loss": 0.88,
       "step": 242
     },
     {
-      "epoch": 2.19,
-      "learning_rate": 0.00015621621621621622,
-      "loss": 0.5623,
+      "epoch": 4.58,
+      "learning_rate": 2.7075471698113207e-05,
+      "loss": 0.8906,
       "step": 243
     },
     {
-      "epoch": 2.2,
-      "learning_rate": 0.00015603603603603605,
-      "loss": 0.712,
+      "epoch": 4.6,
+      "learning_rate": 2.69811320754717e-05,
+      "loss": 0.9012,
       "step": 244
     },
     {
-      "epoch": 2.21,
-      "learning_rate": 0.00015585585585585588,
-      "loss": 0.5958,
+      "epoch": 4.62,
+      "learning_rate": 2.688679245283019e-05,
+      "loss": 0.7683,
       "step": 245
     },
     {
-      "epoch": 2.22,
-      "learning_rate": 0.0001556756756756757,
-      "loss": 0.7317,
+      "epoch": 4.64,
+      "learning_rate": 2.679245283018868e-05,
+      "loss": 0.8865,
       "step": 246
     },
     {
-      "epoch": 2.23,
-      "learning_rate": 0.0001554954954954955,
-      "loss": 0.7904,
+      "epoch": 4.66,
+      "learning_rate": 2.6698113207547172e-05,
+      "loss": 0.8114,
       "step": 247
     },
     {
-      "epoch": 2.23,
-      "learning_rate": 0.0001553153153153153,
-      "loss": 0.6661,
+      "epoch": 4.68,
+      "learning_rate": 2.6603773584905663e-05,
+      "loss": 0.852,
       "step": 248
     },
     {
-      "epoch": 2.24,
-      "learning_rate": 0.00015513513513513514,
-      "loss": 0.6191,
+      "epoch": 4.7,
+      "learning_rate": 2.650943396226415e-05,
+      "loss": 0.8537,
       "step": 249
     },
     {
-      "epoch": 2.25,
-      "learning_rate": 0.00015495495495495496,
-      "loss": 0.6944,
+      "epoch": 4.72,
+      "learning_rate": 2.641509433962264e-05,
+      "loss": 0.8999,
       "step": 250
     },
     {
-      "epoch": 2.26,
-      "learning_rate": 0.00015477477477477477,
-      "loss": 0.7057,
+      "epoch": 4.74,
+      "learning_rate": 2.6320754716981133e-05,
+      "loss": 0.8895,
       "step": 251
     },
     {
-      "epoch": 2.27,
-      "learning_rate": 0.0001545945945945946,
-      "loss": 0.6131,
+      "epoch": 4.75,
+      "learning_rate": 2.6226415094339624e-05,
+      "loss": 0.8271,
       "step": 252
     },
     {
-      "epoch": 2.28,
-      "learning_rate": 0.00015441441441441442,
-      "loss": 0.5626,
+      "epoch": 4.77,
+      "learning_rate": 2.6132075471698115e-05,
+      "loss": 0.7964,
       "step": 253
     },
     {
-      "epoch": 2.29,
-      "learning_rate": 0.00015423423423423425,
-      "loss": 0.6566,
+      "epoch": 4.79,
+      "learning_rate": 2.6037735849056606e-05,
+      "loss": 0.9496,
       "step": 254
     },
     {
-      "epoch": 2.3,
-      "learning_rate": 0.00015405405405405405,
-      "loss": 0.7677,
+      "epoch": 4.81,
+      "learning_rate": 2.5943396226415094e-05,
+      "loss": 0.8386,
       "step": 255
     },
     {
-      "epoch": 2.31,
-      "learning_rate": 0.00015387387387387388,
-      "loss": 0.7221,
+      "epoch": 4.83,
+      "learning_rate": 2.5849056603773585e-05,
+      "loss": 0.7688,
       "step": 256
     },
     {
-      "epoch": 2.32,
-      "learning_rate": 0.0001536936936936937,
-      "loss": 0.7011,
+      "epoch": 4.85,
+      "learning_rate": 2.5754716981132076e-05,
+      "loss": 0.8199,
       "step": 257
     },
     {
-      "epoch": 2.32,
-      "learning_rate": 0.00015351351351351354,
-      "loss": 0.651,
+      "epoch": 4.87,
+      "learning_rate": 2.5660377358490567e-05,
+      "loss": 0.8596,
       "step": 258
     },
     {
-      "epoch": 2.33,
-      "learning_rate": 0.00015333333333333334,
-      "loss": 0.7339,
+      "epoch": 4.89,
+      "learning_rate": 2.5566037735849058e-05,
+      "loss": 0.8176,
       "step": 259
     },
     {
-      "epoch": 2.34,
-      "learning_rate": 0.00015315315315315314,
-      "loss": 0.7311,
+      "epoch": 4.91,
+      "learning_rate": 2.547169811320755e-05,
+      "loss": 0.8093,
       "step": 260
     },
     {
-      "epoch": 2.35,
-      "learning_rate": 0.00015297297297297297,
-      "loss": 0.799,
+      "epoch": 4.92,
+      "learning_rate": 2.537735849056604e-05,
+      "loss": 0.7932,
       "step": 261
     },
     {
-      "epoch": 2.36,
-      "learning_rate": 0.0001527927927927928,
-      "loss": 0.5861,
+      "epoch": 4.94,
+      "learning_rate": 2.5283018867924528e-05,
+      "loss": 0.7955,
       "step": 262
     },
     {
-      "epoch": 2.37,
-      "learning_rate": 0.00015261261261261262,
-      "loss": 0.7177,
+      "epoch": 4.96,
+      "learning_rate": 2.518867924528302e-05,
+      "loss": 0.8786,
       "step": 263
     },
     {
-      "epoch": 2.38,
-      "learning_rate": 0.00015243243243243243,
-      "loss": 0.6037,
+      "epoch": 4.98,
+      "learning_rate": 2.509433962264151e-05,
+      "loss": 0.7967,
       "step": 264
     },
     {
-      "epoch": 2.39,
-      "learning_rate": 0.00015225225225225225,
-      "loss": 0.595,
+      "epoch": 5.0,
+      "learning_rate": 2.5e-05,
+      "loss": 0.7464,
       "step": 265
     },
     {
-      "epoch": 2.4,
-      "learning_rate": 0.00015207207207207208,
-      "loss": 0.4945,
+      "epoch": 5.02,
+      "learning_rate": 2.4905660377358492e-05,
+      "loss": 0.7578,
       "step": 266
     },
     {
-      "epoch": 2.41,
-      "learning_rate": 0.0001518918918918919,
-      "loss": 0.7155,
+      "epoch": 5.04,
+      "learning_rate": 2.4811320754716983e-05,
+      "loss": 0.7802,
       "step": 267
     },
     {
-      "epoch": 2.41,
-      "learning_rate": 0.0001517117117117117,
-      "loss": 0.633,
+      "epoch": 5.06,
+      "learning_rate": 2.4716981132075474e-05,
+      "loss": 0.7665,
       "step": 268
     },
     {
-      "epoch": 2.42,
-      "learning_rate": 0.00015153153153153154,
-      "loss": 0.698,
+      "epoch": 5.08,
+      "learning_rate": 2.4622641509433962e-05,
+      "loss": 0.797,
       "step": 269
     },
     {
-      "epoch": 2.43,
-      "learning_rate": 0.00015135135135135137,
-      "loss": 0.7286,
+      "epoch": 5.09,
+      "learning_rate": 2.4528301886792453e-05,
+      "loss": 0.7681,
       "step": 270
     },
     {
-      "epoch": 2.44,
-      "learning_rate": 0.0001511711711711712,
-      "loss": 0.8749,
+      "epoch": 5.11,
+      "learning_rate": 2.4433962264150944e-05,
+      "loss": 0.8316,
       "step": 271
     },
     {
-      "epoch": 2.45,
-      "learning_rate": 0.000150990990990991,
-      "loss": 0.638,
+      "epoch": 5.13,
+      "learning_rate": 2.4339622641509435e-05,
+      "loss": 0.8532,
       "step": 272
     },
     {
-      "epoch": 2.46,
-      "learning_rate": 0.0001508108108108108,
-      "loss": 0.7524,
+      "epoch": 5.15,
+      "learning_rate": 2.4245283018867926e-05,
+      "loss": 0.7485,
       "step": 273
     },
     {
-      "epoch": 2.47,
-      "learning_rate": 0.00015063063063063063,
-      "loss": 1.5031,
+      "epoch": 5.17,
+      "learning_rate": 2.4150943396226418e-05,
+      "loss": 0.7837,
       "step": 274
     },
     {
-      "epoch": 2.48,
-      "learning_rate": 0.00015045045045045046,
-      "loss": 1.8588,
+      "epoch": 5.19,
+      "learning_rate": 2.405660377358491e-05,
+      "loss": 0.7994,
       "step": 275
     },
     {
-      "epoch": 2.49,
-      "learning_rate": 0.00015027027027027028,
-      "loss": 0.8453,
+      "epoch": 5.21,
+      "learning_rate": 2.3962264150943396e-05,
+      "loss": 0.7448,
       "step": 276
     },
     {
-      "epoch": 2.5,
-      "learning_rate": 0.00015009009009009009,
-      "loss": 1.0263,
+      "epoch": 5.23,
+      "learning_rate": 2.3867924528301887e-05,
+      "loss": 0.842,
       "step": 277
     },
     {
-      "epoch": 2.5,
-      "learning_rate": 0.00014990990990990991,
-      "loss": 0.7366,
+      "epoch": 5.25,
+      "learning_rate": 2.377358490566038e-05,
+      "loss": 0.8137,
       "step": 278
     },
     {
-      "epoch": 2.51,
-      "learning_rate": 0.00014972972972972974,
-      "loss": 0.7533,
+      "epoch": 5.26,
+      "learning_rate": 2.367924528301887e-05,
+      "loss": 0.7739,
       "step": 279
     },
     {
-      "epoch": 2.52,
-      "learning_rate": 0.00014954954954954957,
-      "loss": 0.7375,
+      "epoch": 5.28,
+      "learning_rate": 2.358490566037736e-05,
+      "loss": 0.7935,
       "step": 280
     },
     {
-      "epoch": 2.53,
-      "learning_rate": 0.00014936936936936937,
-      "loss": 0.8677,
+      "epoch": 5.3,
+      "learning_rate": 2.3490566037735852e-05,
+      "loss": 0.8022,
       "step": 281
     },
     {
-      "epoch": 2.54,
-      "learning_rate": 0.0001491891891891892,
-      "loss": 0.689,
+      "epoch": 5.32,
+      "learning_rate": 2.339622641509434e-05,
+      "loss": 0.7823,
       "step": 282
     },
     {
-      "epoch": 2.55,
-      "learning_rate": 0.00014900900900900903,
-      "loss": 0.6659,
+      "epoch": 5.34,
+      "learning_rate": 2.330188679245283e-05,
+      "loss": 0.7645,
       "step": 283
     },
     {
-      "epoch": 2.56,
-      "learning_rate": 0.00014882882882882883,
-      "loss": 1.3045,
+      "epoch": 5.36,
+      "learning_rate": 2.320754716981132e-05,
+      "loss": 0.7344,
       "step": 284
     },
     {
-      "epoch": 2.57,
-      "learning_rate": 0.00014864864864864866,
-      "loss": 0.8731,
+      "epoch": 5.38,
+      "learning_rate": 2.3113207547169813e-05,
+      "loss": 0.8098,
       "step": 285
     },
     {
-      "epoch": 2.58,
-      "learning_rate": 0.00014846846846846846,
-      "loss": 0.9237,
+      "epoch": 5.4,
+      "learning_rate": 2.3018867924528304e-05,
+      "loss": 0.7387,
       "step": 286
     },
     {
-      "epoch": 2.59,
-      "learning_rate": 0.0001482882882882883,
-      "loss": 0.7298,
+      "epoch": 5.42,
+      "learning_rate": 2.2924528301886795e-05,
+      "loss": 0.7694,
       "step": 287
     },
     {
-      "epoch": 2.59,
-      "learning_rate": 0.00014810810810810812,
-      "loss": 0.5712,
+      "epoch": 5.43,
+      "learning_rate": 2.2830188679245286e-05,
+      "loss": 0.7766,
       "step": 288
     },
     {
-      "epoch": 2.6,
-      "learning_rate": 0.00014792792792792795,
-      "loss": 0.8097,
+      "epoch": 5.45,
+      "learning_rate": 2.2735849056603774e-05,
+      "loss": 0.7933,
       "step": 289
     },
     {
-      "epoch": 2.61,
-      "learning_rate": 0.00014774774774774775,
-      "loss": 0.5713,
+      "epoch": 5.47,
+      "learning_rate": 2.2641509433962265e-05,
+      "loss": 0.737,
       "step": 290
     },
     {
-      "epoch": 2.62,
-      "learning_rate": 0.00014756756756756758,
-      "loss": 0.7031,
+      "epoch": 5.49,
+      "learning_rate": 2.2547169811320756e-05,
+      "loss": 0.8105,
       "step": 291
     },
     {
-      "epoch": 2.63,
-      "learning_rate": 0.0001473873873873874,
-      "loss": 0.5795,
+      "epoch": 5.51,
+      "learning_rate": 2.2452830188679247e-05,
+      "loss": 0.7348,
       "step": 292
     },
     {
-      "epoch": 2.64,
-      "learning_rate": 0.00014720720720720723,
-      "loss": 0.6107,
+      "epoch": 5.53,
+      "learning_rate": 2.2358490566037738e-05,
+      "loss": 0.7407,
       "step": 293
     },
     {
-      "epoch": 2.65,
-      "learning_rate": 0.00014702702702702703,
-      "loss": 0.7417,
+      "epoch": 5.55,
+      "learning_rate": 2.226415094339623e-05,
+      "loss": 0.7406,
       "step": 294
     },
     {
-      "epoch": 2.66,
-      "learning_rate": 0.00014684684684684686,
-      "loss": 0.7342,
+      "epoch": 5.57,
+      "learning_rate": 2.216981132075472e-05,
+      "loss": 0.7485,
       "step": 295
     },
     {
-      "epoch": 2.67,
-      "learning_rate": 0.00014666666666666666,
-      "loss": 0.6121,
+      "epoch": 5.58,
+      "learning_rate": 2.2075471698113208e-05,
+      "loss": 0.7226,
       "step": 296
     },
     {
-      "epoch": 2.68,
-      "learning_rate": 0.0001464864864864865,
-      "loss": 0.801,
+      "epoch": 5.6,
+      "learning_rate": 2.19811320754717e-05,
+      "loss": 0.7865,
       "step": 297
     },
     {
-      "epoch": 2.68,
-      "learning_rate": 0.00014630630630630632,
-      "loss": 0.5626,
+      "epoch": 5.62,
+      "learning_rate": 2.188679245283019e-05,
+      "loss": 0.7983,
       "step": 298
     },
     {
-      "epoch": 2.69,
-      "learning_rate": 0.00014612612612612612,
-      "loss": 0.6167,
+      "epoch": 5.64,
+      "learning_rate": 2.179245283018868e-05,
+      "loss": 0.7524,
       "step": 299
     },
     {
-      "epoch": 2.7,
-      "learning_rate": 0.00014594594594594595,
-      "loss": 0.606,
+      "epoch": 5.66,
+      "learning_rate": 2.1698113207547172e-05,
+      "loss": 0.6711,
       "step": 300
     },
     {
-      "epoch": 2.71,
-      "learning_rate": 0.00014576576576576578,
-      "loss": 0.7919,
+      "epoch": 5.68,
+      "learning_rate": 2.1603773584905663e-05,
+      "loss": 0.8228,
       "step": 301
     },
     {
-      "epoch": 2.72,
-      "learning_rate": 0.0001455855855855856,
-      "loss": 0.9143,
+      "epoch": 5.7,
+      "learning_rate": 2.1509433962264154e-05,
+      "loss": 0.72,
       "step": 302
     },
     {
-      "epoch": 2.73,
-      "learning_rate": 0.0001454054054054054,
-      "loss": 0.5416,
+      "epoch": 5.72,
+      "learning_rate": 2.1415094339622642e-05,
+      "loss": 0.8165,
       "step": 303
     },
     {
-      "epoch": 2.74,
-      "learning_rate": 0.00014522522522522524,
-      "loss": 0.9303,
+      "epoch": 5.74,
+      "learning_rate": 2.1320754716981133e-05,
+      "loss": 0.7387,
       "step": 304
     },
     {
-      "epoch": 2.75,
-      "learning_rate": 0.00014504504504504506,
-      "loss": 0.7065,
+      "epoch": 5.75,
+      "learning_rate": 2.1226415094339624e-05,
+      "loss": 0.7303,
       "step": 305
     },
     {
-      "epoch": 2.76,
-      "learning_rate": 0.0001448648648648649,
-      "loss": 0.7287,
+      "epoch": 5.77,
+      "learning_rate": 2.1132075471698115e-05,
+      "loss": 0.7795,
       "step": 306
     },
     {
-      "epoch": 2.77,
-      "learning_rate": 0.0001446846846846847,
-      "loss": 0.6792,
+      "epoch": 5.79,
+      "learning_rate": 2.1037735849056606e-05,
+      "loss": 0.7651,
       "step": 307
     },
     {
-      "epoch": 2.77,
-      "learning_rate": 0.0001445045045045045,
-      "loss": 0.762,
+      "epoch": 5.81,
+      "learning_rate": 2.0943396226415098e-05,
+      "loss": 0.8202,
       "step": 308
     },
     {
-      "epoch": 2.78,
-      "learning_rate": 0.00014432432432432432,
-      "loss": 0.5965,
+      "epoch": 5.83,
+      "learning_rate": 2.0849056603773585e-05,
+      "loss": 0.807,
       "step": 309
     },
     {
-      "epoch": 2.79,
-      "learning_rate": 0.00014414414414414415,
-      "loss": 0.6565,
+      "epoch": 5.85,
+      "learning_rate": 2.0754716981132076e-05,
+      "loss": 0.7871,
       "step": 310
     },
     {
-      "epoch": 2.8,
-      "learning_rate": 0.00014396396396396395,
-      "loss": 0.734,
+      "epoch": 5.87,
+      "learning_rate": 2.0660377358490567e-05,
+      "loss": 0.6727,
       "step": 311
     },
     {
-      "epoch": 2.81,
-      "learning_rate": 0.00014378378378378378,
-      "loss": 0.6371,
+      "epoch": 5.89,
+      "learning_rate": 2.056603773584906e-05,
+      "loss": 0.7548,
       "step": 312
     },
     {
-      "epoch": 2.82,
-      "learning_rate": 0.0001436036036036036,
-      "loss": 0.67,
+      "epoch": 5.91,
+      "learning_rate": 2.047169811320755e-05,
+      "loss": 0.7253,
       "step": 313
     },
     {
-      "epoch": 2.83,
-      "learning_rate": 0.00014342342342342344,
-      "loss": 0.6313,
+      "epoch": 5.92,
+      "learning_rate": 2.037735849056604e-05,
+      "loss": 0.7422,
       "step": 314
     },
     {
-      "epoch": 2.84,
-      "learning_rate": 0.00014324324324324324,
-      "loss": 0.6081,
+      "epoch": 5.94,
+      "learning_rate": 2.0283018867924532e-05,
+      "loss": 0.8262,
       "step": 315
     },
     {
-      "epoch": 2.85,
-      "learning_rate": 0.00014306306306306307,
-      "loss": 0.7904,
+      "epoch": 5.96,
+      "learning_rate": 2.018867924528302e-05,
+      "loss": 0.7564,
       "step": 316
     },
     {
-      "epoch": 2.86,
-      "learning_rate": 0.0001428828828828829,
-      "loss": 0.6455,
+      "epoch": 5.98,
+      "learning_rate": 2.009433962264151e-05,
+      "loss": 0.8395,
       "step": 317
     },
     {
-      "epoch": 2.86,
-      "learning_rate": 0.00014270270270270272,
-      "loss": 0.6323,
+      "epoch": 6.0,
+      "learning_rate": 2e-05,
+      "loss": 0.7952,
       "step": 318
     },
     {
-      "epoch": 2.87,
-      "learning_rate": 0.00014252252252252253,
-      "loss": 0.779,
+      "epoch": 6.02,
+      "learning_rate": 1.9905660377358493e-05,
+      "loss": 0.719,
       "step": 319
     },
     {
-      "epoch": 2.88,
-      "learning_rate": 0.00014234234234234233,
-      "loss": 0.5011,
+      "epoch": 6.04,
+      "learning_rate": 1.9811320754716984e-05,
+      "loss": 0.7088,
       "step": 320
     },
     {
-      "epoch": 2.89,
-      "learning_rate": 0.00014216216216216216,
-      "loss": 0.6007,
+      "epoch": 6.06,
+      "learning_rate": 1.9716981132075475e-05,
+      "loss": 0.7672,
       "step": 321
     },
     {
-      "epoch": 2.9,
-      "learning_rate": 0.00014198198198198198,
-      "loss": 0.7067,
+      "epoch": 6.08,
+      "learning_rate": 1.9622641509433966e-05,
+      "loss": 0.6742,
       "step": 322
     },
     {
-      "epoch": 2.91,
-      "learning_rate": 0.0001418018018018018,
-      "loss": 0.6962,
+      "epoch": 6.09,
+      "learning_rate": 1.9528301886792454e-05,
+      "loss": 0.7195,
       "step": 323
     },
     {
-      "epoch": 2.92,
-      "learning_rate": 0.00014162162162162161,
-      "loss": 0.7823,
+      "epoch": 6.11,
+      "learning_rate": 1.9433962264150945e-05,
+      "loss": 0.773,
       "step": 324
     },
     {
-      "epoch": 2.93,
-      "learning_rate": 0.00014144144144144144,
-      "loss": 0.7142,
+      "epoch": 6.13,
+      "learning_rate": 1.9339622641509436e-05,
+      "loss": 0.6531,
       "step": 325
     },
     {
-      "epoch": 2.94,
-      "learning_rate": 0.00014126126126126127,
-      "loss": 0.6356,
+      "epoch": 6.15,
+      "learning_rate": 1.9245283018867927e-05,
+      "loss": 0.7301,
       "step": 326
     },
     {
-      "epoch": 2.95,
-      "learning_rate": 0.0001410810810810811,
-      "loss": 0.7062,
+      "epoch": 6.17,
+      "learning_rate": 1.9150943396226418e-05,
+      "loss": 0.6485,
       "step": 327
     },
     {
-      "epoch": 2.95,
-      "learning_rate": 0.0001409009009009009,
-      "loss": 0.7607,
+      "epoch": 6.19,
+      "learning_rate": 1.905660377358491e-05,
+      "loss": 0.6735,
       "step": 328
     },
     {
-      "epoch": 2.96,
-      "learning_rate": 0.00014072072072072073,
-      "loss": 0.6222,
+      "epoch": 6.21,
+      "learning_rate": 1.8962264150943397e-05,
+      "loss": 0.7351,
       "step": 329
     },
     {
-      "epoch": 2.97,
-      "learning_rate": 0.00014054054054054056,
-      "loss": 0.7505,
+      "epoch": 6.23,
+      "learning_rate": 1.8867924528301888e-05,
+      "loss": 0.7142,
       "step": 330
     },
     {
-      "epoch": 2.98,
-      "learning_rate": 0.00014036036036036039,
-      "loss": 0.6673,
+      "epoch": 6.25,
+      "learning_rate": 1.877358490566038e-05,
+      "loss": 0.6725,
       "step": 331
     },
     {
-      "epoch": 2.99,
-      "learning_rate": 0.0001401801801801802,
-      "loss": 0.6934,
+      "epoch": 6.26,
+      "learning_rate": 1.8679245283018867e-05,
+      "loss": 0.7494,
       "step": 332
     },
     {
-      "epoch": 3.0,
-      "learning_rate": 0.00014,
-      "loss": 0.5489,
+      "epoch": 6.28,
+      "learning_rate": 1.8584905660377358e-05,
+      "loss": 0.6543,
       "step": 333
     },
     {
-      "epoch": 3.01,
-      "learning_rate": 0.00013981981981981982,
-      "loss": 0.4571,
+      "epoch": 6.3,
+      "learning_rate": 1.849056603773585e-05,
+      "loss": 0.8346,
       "step": 334
     },
     {
-      "epoch": 3.02,
-      "learning_rate": 0.00013963963963963964,
-      "loss": 0.4375,
+      "epoch": 6.32,
+      "learning_rate": 1.839622641509434e-05,
+      "loss": 0.8046,
       "step": 335
     },
     {
-      "epoch": 3.03,
-      "learning_rate": 0.00013945945945945947,
-      "loss": 0.4438,
+      "epoch": 6.34,
+      "learning_rate": 1.830188679245283e-05,
+      "loss": 0.8155,
       "step": 336
     },
     {
-      "epoch": 3.04,
-      "learning_rate": 0.00013927927927927927,
-      "loss": 0.4085,
+      "epoch": 6.36,
+      "learning_rate": 1.820754716981132e-05,
+      "loss": 0.647,
       "step": 337
     },
     {
-      "epoch": 3.05,
-      "learning_rate": 0.0001390990990990991,
-      "loss": 0.5438,
+      "epoch": 6.38,
+      "learning_rate": 1.811320754716981e-05,
+      "loss": 0.7001,
       "step": 338
     },
     {
-      "epoch": 3.05,
-      "learning_rate": 0.00013891891891891893,
-      "loss": 0.5038,
+      "epoch": 6.4,
+      "learning_rate": 1.80188679245283e-05,
+      "loss": 0.7545,
       "step": 339
     },
     {
-      "epoch": 3.06,
-      "learning_rate": 0.00013873873873873876,
-      "loss": 0.479,
+      "epoch": 6.42,
+      "learning_rate": 1.7924528301886792e-05,
+      "loss": 0.6668,
       "step": 340
     },
     {
-      "epoch": 3.07,
-      "learning_rate": 0.00013855855855855856,
-      "loss": 0.405,
+      "epoch": 6.43,
+      "learning_rate": 1.7830188679245283e-05,
+      "loss": 0.8008,
       "step": 341
     },
     {
-      "epoch": 3.08,
-      "learning_rate": 0.0001383783783783784,
-      "loss": 0.4181,
+      "epoch": 6.45,
+      "learning_rate": 1.7735849056603774e-05,
+      "loss": 0.6139,
       "step": 342
     },
     {
-      "epoch": 3.09,
-      "learning_rate": 0.00013819819819819822,
-      "loss": 0.4513,
+      "epoch": 6.47,
+      "learning_rate": 1.7641509433962265e-05,
+      "loss": 0.7745,
       "step": 343
     },
     {
-      "epoch": 3.1,
-      "learning_rate": 0.00013801801801801802,
-      "loss": 0.4685,
+      "epoch": 6.49,
+      "learning_rate": 1.7547169811320753e-05,
+      "loss": 0.6493,
       "step": 344
     },
     {
-      "epoch": 3.11,
-      "learning_rate": 0.00013783783783783785,
-      "loss": 0.5291,
+      "epoch": 6.51,
+      "learning_rate": 1.7452830188679244e-05,
+      "loss": 0.6302,
       "step": 345
     },
     {
-      "epoch": 3.12,
-      "learning_rate": 0.00013765765765765765,
-      "loss": 0.5497,
+      "epoch": 6.53,
+      "learning_rate": 1.7358490566037735e-05,
+      "loss": 0.7609,
       "step": 346
     },
     {
-      "epoch": 3.13,
-      "learning_rate": 0.00013747747747747748,
-      "loss": 0.4366,
+      "epoch": 6.55,
+      "learning_rate": 1.7264150943396226e-05,
+      "loss": 0.765,
       "step": 347
     },
     {
-      "epoch": 3.14,
-      "learning_rate": 0.0001372972972972973,
-      "loss": 0.5815,
+      "epoch": 6.57,
+      "learning_rate": 1.7169811320754717e-05,
+      "loss": 0.627,
       "step": 348
     },
     {
-      "epoch": 3.14,
-      "learning_rate": 0.00013711711711711713,
-      "loss": 0.5056,
+      "epoch": 6.58,
+      "learning_rate": 1.707547169811321e-05,
+      "loss": 0.7065,
       "step": 349
     },
     {
-      "epoch": 3.15,
-      "learning_rate": 0.00013693693693693693,
-      "loss": 0.4975,
+      "epoch": 6.6,
+      "learning_rate": 1.69811320754717e-05,
+      "loss": 0.7555,
       "step": 350
     },
     {
-      "epoch": 3.16,
-      "learning_rate": 0.00013675675675675676,
-      "loss": 0.4601,
+      "epoch": 6.62,
+      "learning_rate": 1.6886792452830187e-05,
+      "loss": 0.7766,
       "step": 351
     },
     {
-      "epoch": 3.17,
-      "learning_rate": 0.0001365765765765766,
-      "loss": 0.4328,
+      "epoch": 6.64,
+      "learning_rate": 1.6792452830188678e-05,
+      "loss": 0.6657,
       "step": 352
     },
     {
-      "epoch": 3.18,
-      "learning_rate": 0.00013639639639639642,
-      "loss": 0.5189,
+      "epoch": 6.66,
+      "learning_rate": 1.669811320754717e-05,
+      "loss": 0.5936,
       "step": 353
     },
     {
-      "epoch": 3.19,
-      "learning_rate": 0.00013621621621621622,
-      "loss": 0.4054,
+      "epoch": 6.68,
+      "learning_rate": 1.660377358490566e-05,
+      "loss": 0.6822,
       "step": 354
     },
     {
-      "epoch": 3.2,
-      "learning_rate": 0.00013603603603603605,
-      "loss": 0.4604,
+      "epoch": 6.7,
+      "learning_rate": 1.650943396226415e-05,
+      "loss": 0.7292,
       "step": 355
     },
     {
-      "epoch": 3.21,
-      "learning_rate": 0.00013585585585585585,
-      "loss": 0.5539,
+      "epoch": 6.72,
+      "learning_rate": 1.6415094339622643e-05,
+      "loss": 0.7883,
       "step": 356
     },
     {
-      "epoch": 3.22,
-      "learning_rate": 0.00013567567567567568,
-      "loss": 0.3773,
+      "epoch": 6.74,
+      "learning_rate": 1.6320754716981134e-05,
+      "loss": 0.7028,
       "step": 357
     },
     {
-      "epoch": 3.23,
-      "learning_rate": 0.0001354954954954955,
-      "loss": 0.4631,
+      "epoch": 6.75,
+      "learning_rate": 1.622641509433962e-05,
+      "loss": 0.6749,
       "step": 358
     },
     {
-      "epoch": 3.23,
-      "learning_rate": 0.0001353153153153153,
-      "loss": 0.463,
+      "epoch": 6.77,
+      "learning_rate": 1.6132075471698112e-05,
+      "loss": 0.7383,
       "step": 359
     },
     {
-      "epoch": 3.24,
-      "learning_rate": 0.00013513513513513514,
-      "loss": 0.4423,
+      "epoch": 6.79,
+      "learning_rate": 1.6037735849056604e-05,
+      "loss": 0.7755,
       "step": 360
     },
     {
-      "epoch": 3.25,
-      "learning_rate": 0.00013495495495495497,
-      "loss": 0.5197,
+      "epoch": 6.81,
+      "learning_rate": 1.5943396226415095e-05,
+      "loss": 0.6989,
       "step": 361
     },
     {
-      "epoch": 3.26,
-      "learning_rate": 0.0001347747747747748,
-      "loss": 0.6306,
+      "epoch": 6.83,
+      "learning_rate": 1.5849056603773586e-05,
+      "loss": 0.7402,
       "step": 362
     },
     {
-      "epoch": 3.27,
-      "learning_rate": 0.0001345945945945946,
-      "loss": 0.431,
+      "epoch": 6.85,
+      "learning_rate": 1.5754716981132077e-05,
+      "loss": 0.675,
       "step": 363
     },
     {
-      "epoch": 3.28,
-      "learning_rate": 0.00013441441441441442,
-      "loss": 0.4192,
+      "epoch": 6.87,
+      "learning_rate": 1.5660377358490564e-05,
+      "loss": 0.598,
       "step": 364
     },
     {
-      "epoch": 3.29,
-      "learning_rate": 0.00013423423423423425,
-      "loss": 0.5175,
+      "epoch": 6.89,
+      "learning_rate": 1.5566037735849056e-05,
+      "loss": 0.6434,
       "step": 365
     },
     {
-      "epoch": 3.3,
-      "learning_rate": 0.00013405405405405408,
-      "loss": 0.4623,
+      "epoch": 6.91,
+      "learning_rate": 1.5471698113207547e-05,
+      "loss": 0.744,
       "step": 366
     },
     {
-      "epoch": 3.31,
-      "learning_rate": 0.00013387387387387388,
-      "loss": 0.4332,
+      "epoch": 6.92,
+      "learning_rate": 1.5377358490566038e-05,
+      "loss": 0.7623,
       "step": 367
     },
     {
-      "epoch": 3.32,
-      "learning_rate": 0.00013369369369369368,
-      "loss": 0.4969,
+      "epoch": 6.94,
+      "learning_rate": 1.528301886792453e-05,
+      "loss": 0.728,
       "step": 368
     },
     {
-      "epoch": 3.32,
-      "learning_rate": 0.0001335135135135135,
-      "loss": 0.5908,
+      "epoch": 6.96,
+      "learning_rate": 1.5188679245283018e-05,
+      "loss": 0.6612,
       "step": 369
     },
     {
-      "epoch": 3.33,
-      "learning_rate": 0.00013333333333333334,
-      "loss": 0.4597,
+      "epoch": 6.98,
+      "learning_rate": 1.509433962264151e-05,
+      "loss": 0.6723,
       "step": 370
     },
     {
-      "epoch": 3.34,
-      "learning_rate": 0.00013315315315315314,
-      "loss": 0.4516,
+      "epoch": 7.0,
+      "learning_rate": 1.5e-05,
+      "loss": 0.5828,
       "step": 371
     },
     {
-      "epoch": 3.35,
-      "learning_rate": 0.00013297297297297297,
-      "loss": 0.498,
+      "epoch": 7.02,
+      "learning_rate": 1.4905660377358491e-05,
+      "loss": 0.6875,
       "step": 372
     },
     {
-      "epoch": 3.36,
-      "learning_rate": 0.0001327927927927928,
-      "loss": 0.4829,
+      "epoch": 7.04,
+      "learning_rate": 1.4811320754716981e-05,
+      "loss": 0.6812,
       "step": 373
     },
     {
-      "epoch": 3.37,
-      "learning_rate": 0.00013261261261261263,
-      "loss": 0.4679,
+      "epoch": 7.06,
+      "learning_rate": 1.4716981132075472e-05,
+      "loss": 0.6529,
       "step": 374
     },
     {
-      "epoch": 3.38,
-      "learning_rate": 0.00013243243243243243,
-      "loss": 0.5148,
+      "epoch": 7.08,
+      "learning_rate": 1.4622641509433963e-05,
+      "loss": 0.682,
       "step": 375
     },
     {
-      "epoch": 3.39,
-      "learning_rate": 0.00013225225225225226,
-      "loss": 0.4495,
+      "epoch": 7.09,
+      "learning_rate": 1.4528301886792452e-05,
+      "loss": 0.6836,
       "step": 376
     },
     {
-      "epoch": 3.4,
-      "learning_rate": 0.00013207207207207208,
-      "loss": 0.4408,
+      "epoch": 7.11,
+      "learning_rate": 1.4433962264150944e-05,
+      "loss": 0.6603,
       "step": 377
     },
     {
-      "epoch": 3.41,
-      "learning_rate": 0.0001318918918918919,
-      "loss": 0.4732,
+      "epoch": 7.13,
+      "learning_rate": 1.4339622641509435e-05,
+      "loss": 0.6198,
       "step": 378
     },
     {
-      "epoch": 3.41,
-      "learning_rate": 0.00013171171171171171,
-      "loss": 0.4048,
+      "epoch": 7.15,
+      "learning_rate": 1.4245283018867924e-05,
+      "loss": 0.6459,
       "step": 379
     },
     {
-      "epoch": 3.42,
-      "learning_rate": 0.00013153153153153154,
-      "loss": 0.4607,
+      "epoch": 7.17,
+      "learning_rate": 1.4150943396226415e-05,
+      "loss": 0.682,
       "step": 380
     },
     {
-      "epoch": 3.43,
-      "learning_rate": 0.00013135135135135134,
-      "loss": 0.457,
+      "epoch": 7.19,
+      "learning_rate": 1.4056603773584906e-05,
+      "loss": 0.6039,
       "step": 381
     },
     {
-      "epoch": 3.44,
-      "learning_rate": 0.00013117117117117117,
-      "loss": 0.4446,
+      "epoch": 7.21,
+      "learning_rate": 1.3962264150943397e-05,
+      "loss": 0.7004,
       "step": 382
     },
     {
-      "epoch": 3.45,
-      "learning_rate": 0.000130990990990991,
-      "loss": 0.5565,
+      "epoch": 7.23,
+      "learning_rate": 1.3867924528301887e-05,
+      "loss": 0.622,
       "step": 383
     },
     {
-      "epoch": 3.46,
-      "learning_rate": 0.0001308108108108108,
-      "loss": 0.452,
+      "epoch": 7.25,
+      "learning_rate": 1.3773584905660378e-05,
+      "loss": 0.6887,
       "step": 384
     },
     {
-      "epoch": 3.47,
-      "learning_rate": 0.00013063063063063063,
-      "loss": 0.6298,
+      "epoch": 7.26,
+      "learning_rate": 1.3679245283018869e-05,
+      "loss": 0.6733,
       "step": 385
     },
     {
-      "epoch": 3.48,
-      "learning_rate": 0.00013045045045045046,
-      "loss": 0.5399,
+      "epoch": 7.28,
+      "learning_rate": 1.3584905660377358e-05,
+      "loss": 0.6231,
       "step": 386
     },
     {
-      "epoch": 3.49,
-      "learning_rate": 0.0001302702702702703,
-      "loss": 0.4797,
+      "epoch": 7.3,
+      "learning_rate": 1.349056603773585e-05,
+      "loss": 0.6759,
       "step": 387
     },
     {
-      "epoch": 3.5,
-      "learning_rate": 0.0001300900900900901,
-      "loss": 0.5867,
+      "epoch": 7.32,
+      "learning_rate": 1.339622641509434e-05,
+      "loss": 0.6818,
       "step": 388
     },
     {
-      "epoch": 3.5,
-      "learning_rate": 0.00012990990990990992,
-      "loss": 0.4448,
+      "epoch": 7.34,
+      "learning_rate": 1.3301886792452831e-05,
+      "loss": 0.6215,
       "step": 389
     },
     {
-      "epoch": 3.51,
-      "learning_rate": 0.00012972972972972974,
-      "loss": 0.5132,
+      "epoch": 7.36,
+      "learning_rate": 1.320754716981132e-05,
+      "loss": 0.6892,
       "step": 390
     },
     {
-      "epoch": 3.52,
-      "learning_rate": 0.00012954954954954957,
-      "loss": 0.4374,
+      "epoch": 7.38,
+      "learning_rate": 1.3113207547169812e-05,
+      "loss": 0.6298,
       "step": 391
     },
     {
-      "epoch": 3.53,
-      "learning_rate": 0.00012936936936936937,
-      "loss": 0.5119,
+      "epoch": 7.4,
+      "learning_rate": 1.3018867924528303e-05,
+      "loss": 0.6314,
       "step": 392
     },
     {
-      "epoch": 3.54,
-      "learning_rate": 0.00012918918918918918,
-      "loss": 0.5887,
+      "epoch": 7.42,
+      "learning_rate": 1.2924528301886792e-05,
+      "loss": 0.6547,
       "step": 393
     },
     {
-      "epoch": 3.55,
-      "learning_rate": 0.000129009009009009,
-      "loss": 0.477,
+      "epoch": 7.43,
+      "learning_rate": 1.2830188679245283e-05,
+      "loss": 0.6266,
       "step": 394
     },
     {
-      "epoch": 3.56,
-      "learning_rate": 0.00012882882882882883,
-      "loss": 0.5135,
+      "epoch": 7.45,
+      "learning_rate": 1.2735849056603775e-05,
+      "loss": 0.756,
       "step": 395
     },
     {
-      "epoch": 3.57,
-      "learning_rate": 0.00012864864864864866,
-      "loss": 0.4852,
+      "epoch": 7.47,
+      "learning_rate": 1.2641509433962264e-05,
+      "loss": 0.6805,
       "step": 396
     },
     {
-      "epoch": 3.58,
-      "learning_rate": 0.00012846846846846846,
-      "loss": 0.5327,
+      "epoch": 7.49,
+      "learning_rate": 1.2547169811320755e-05,
+      "loss": 0.6736,
       "step": 397
     },
     {
-      "epoch": 3.59,
-      "learning_rate": 0.0001282882882882883,
-      "loss": 0.4726,
+      "epoch": 7.51,
+      "learning_rate": 1.2452830188679246e-05,
+      "loss": 0.6708,
       "step": 398
     },
     {
-      "epoch": 3.59,
-      "learning_rate": 0.00012810810810810812,
-      "loss": 0.4805,
+      "epoch": 7.53,
+      "learning_rate": 1.2358490566037737e-05,
+      "loss": 0.6287,
       "step": 399
     },
     {
-      "epoch": 3.6,
-      "learning_rate": 0.00012792792792792795,
-      "loss": 0.4577,
+      "epoch": 7.55,
+      "learning_rate": 1.2264150943396227e-05,
+      "loss": 0.7074,
       "step": 400
     },
     {
-      "epoch": 3.61,
-      "learning_rate": 0.00012774774774774775,
-      "loss": 0.4823,
+      "epoch": 7.57,
+      "learning_rate": 1.2169811320754718e-05,
+      "loss": 0.6036,
       "step": 401
     },
     {
-      "epoch": 3.62,
-      "learning_rate": 0.00012756756756756758,
-      "loss": 0.4779,
+      "epoch": 7.58,
+      "learning_rate": 1.2075471698113209e-05,
+      "loss": 0.6801,
       "step": 402
     },
     {
-      "epoch": 3.63,
-      "learning_rate": 0.0001273873873873874,
-      "loss": 0.586,
+      "epoch": 7.6,
+      "learning_rate": 1.1981132075471698e-05,
+      "loss": 0.6523,
       "step": 403
     },
     {
-      "epoch": 3.64,
-      "learning_rate": 0.00012720720720720723,
-      "loss": 0.4533,
+      "epoch": 7.62,
+      "learning_rate": 1.188679245283019e-05,
+      "loss": 0.6658,
       "step": 404
     },
     {
-      "epoch": 3.65,
-      "learning_rate": 0.00012702702702702703,
-      "loss": 0.5355,
+      "epoch": 7.64,
+      "learning_rate": 1.179245283018868e-05,
+      "loss": 0.6517,
       "step": 405
     },
     {
-      "epoch": 3.66,
-      "learning_rate": 0.00012684684684684684,
-      "loss": 0.581,
+      "epoch": 7.66,
+      "learning_rate": 1.169811320754717e-05,
+      "loss": 0.7265,
       "step": 406
     },
     {
-      "epoch": 3.67,
-      "learning_rate": 0.00012666666666666666,
-      "loss": 0.4933,
+      "epoch": 7.68,
+      "learning_rate": 1.160377358490566e-05,
+      "loss": 0.6105,
       "step": 407
     },
     {
-      "epoch": 3.68,
-      "learning_rate": 0.0001264864864864865,
-      "loss": 0.5883,
+      "epoch": 7.7,
+      "learning_rate": 1.1509433962264152e-05,
+      "loss": 0.6036,
       "step": 408
     },
     {
-      "epoch": 3.68,
-      "learning_rate": 0.00012630630630630632,
-      "loss": 0.5552,
+      "epoch": 7.72,
+      "learning_rate": 1.1415094339622643e-05,
+      "loss": 0.6095,
       "step": 409
     },
     {
-      "epoch": 3.69,
-      "learning_rate": 0.00012612612612612612,
-      "loss": 0.5098,
+      "epoch": 7.74,
+      "learning_rate": 1.1320754716981132e-05,
+      "loss": 0.7576,
       "step": 410
     },
     {
-      "epoch": 3.7,
-      "learning_rate": 0.00012594594594594595,
-      "loss": 0.4585,
+      "epoch": 7.75,
+      "learning_rate": 1.1226415094339623e-05,
+      "loss": 0.6719,
       "step": 411
     },
     {
-      "epoch": 3.71,
-      "learning_rate": 0.00012576576576576578,
-      "loss": 0.5726,
+      "epoch": 7.77,
+      "learning_rate": 1.1132075471698115e-05,
+      "loss": 0.7032,
       "step": 412
     },
     {
-      "epoch": 3.72,
-      "learning_rate": 0.0001255855855855856,
-      "loss": 0.5016,
+      "epoch": 7.79,
+      "learning_rate": 1.1037735849056604e-05,
+      "loss": 0.6761,
       "step": 413
     },
     {
-      "epoch": 3.73,
-      "learning_rate": 0.0001254054054054054,
-      "loss": 0.6272,
+      "epoch": 7.81,
+      "learning_rate": 1.0943396226415095e-05,
+      "loss": 0.6498,
       "step": 414
     },
     {
-      "epoch": 3.74,
-      "learning_rate": 0.00012522522522522524,
-      "loss": 0.5357,
+      "epoch": 7.83,
+      "learning_rate": 1.0849056603773586e-05,
+      "loss": 0.6711,
       "step": 415
     },
     {
-      "epoch": 3.75,
-      "learning_rate": 0.00012504504504504507,
-      "loss": 0.4562,
+      "epoch": 7.85,
+      "learning_rate": 1.0754716981132077e-05,
+      "loss": 0.5983,
       "step": 416
     },
     {
-      "epoch": 3.76,
-      "learning_rate": 0.00012486486486486487,
-      "loss": 0.5279,
+      "epoch": 7.87,
+      "learning_rate": 1.0660377358490567e-05,
+      "loss": 0.6335,
       "step": 417
     },
     {
-      "epoch": 3.77,
-      "learning_rate": 0.0001246846846846847,
-      "loss": 0.5381,
+      "epoch": 7.89,
+      "learning_rate": 1.0566037735849058e-05,
+      "loss": 0.6277,
       "step": 418
     },
     {
-      "epoch": 3.77,
-      "learning_rate": 0.0001245045045045045,
-      "loss": 0.5124,
+      "epoch": 7.91,
+      "learning_rate": 1.0471698113207549e-05,
+      "loss": 0.6953,
       "step": 419
     },
     {
-      "epoch": 3.78,
-      "learning_rate": 0.00012432432432432433,
-      "loss": 0.5242,
+      "epoch": 7.92,
+      "learning_rate": 1.0377358490566038e-05,
+      "loss": 0.6101,
       "step": 420
     },
     {
-      "epoch": 3.79,
-      "learning_rate": 0.00012414414414414415,
-      "loss": 0.4864,
+      "epoch": 7.94,
+      "learning_rate": 1.028301886792453e-05,
+      "loss": 0.5842,
       "step": 421
     },
     {
-      "epoch": 3.8,
-      "learning_rate": 0.00012396396396396398,
-      "loss": 0.4408,
+      "epoch": 7.96,
+      "learning_rate": 1.018867924528302e-05,
+      "loss": 0.6547,
       "step": 422
     },
     {
-      "epoch": 3.81,
-      "learning_rate": 0.00012378378378378378,
-      "loss": 0.4686,
+      "epoch": 7.98,
+      "learning_rate": 1.009433962264151e-05,
+      "loss": 0.581,
       "step": 423
     },
     {
-      "epoch": 3.82,
-      "learning_rate": 0.0001236036036036036,
-      "loss": 0.4583,
+      "epoch": 8.0,
+      "learning_rate": 1e-05,
+      "loss": 0.6649,
       "step": 424
     },
     {
-      "epoch": 3.83,
-      "learning_rate": 0.00012342342342342344,
-      "loss": 0.5698,
+      "epoch": 8.02,
+      "learning_rate": 9.905660377358492e-06,
+      "loss": 0.6135,
       "step": 425
     },
     {
-      "epoch": 3.84,
-      "learning_rate": 0.00012324324324324327,
-      "loss": 0.5193,
+      "epoch": 8.04,
+      "learning_rate": 9.811320754716983e-06,
+      "loss": 0.6005,
       "step": 426
     },
     {
-      "epoch": 3.85,
-      "learning_rate": 0.00012306306306306307,
-      "loss": 0.6061,
+      "epoch": 8.06,
+      "learning_rate": 9.716981132075472e-06,
+      "loss": 0.5888,
       "step": 427
     },
     {
-      "epoch": 3.86,
-      "learning_rate": 0.0001228828828828829,
-      "loss": 0.4556,
+      "epoch": 8.08,
+      "learning_rate": 9.622641509433963e-06,
+      "loss": 0.5913,
       "step": 428
     },
     {
-      "epoch": 3.86,
-      "learning_rate": 0.0001227027027027027,
-      "loss": 0.4494,
+      "epoch": 8.09,
+      "learning_rate": 9.528301886792455e-06,
+      "loss": 0.6436,
       "step": 429
     },
     {
-      "epoch": 3.87,
-      "learning_rate": 0.00012252252252252253,
-      "loss": 0.4351,
+      "epoch": 8.11,
+      "learning_rate": 9.433962264150944e-06,
+      "loss": 0.6632,
       "step": 430
     },
     {
-      "epoch": 3.88,
-      "learning_rate": 0.00012234234234234233,
-      "loss": 0.4312,
+      "epoch": 8.13,
+      "learning_rate": 9.339622641509433e-06,
+      "loss": 0.5931,
       "step": 431
     },
     {
-      "epoch": 3.89,
-      "learning_rate": 0.00012216216216216216,
-      "loss": 0.5729,
+      "epoch": 8.15,
+      "learning_rate": 9.245283018867924e-06,
+      "loss": 0.563,
       "step": 432
     },
     {
-      "epoch": 3.9,
-      "learning_rate": 0.00012198198198198199,
-      "loss": 0.4375,
+      "epoch": 8.17,
+      "learning_rate": 9.150943396226416e-06,
+      "loss": 0.6361,
       "step": 433
     },
     {
-      "epoch": 3.91,
-      "learning_rate": 0.0001218018018018018,
-      "loss": 0.5225,
+      "epoch": 8.19,
+      "learning_rate": 9.056603773584905e-06,
+      "loss": 0.605,
       "step": 434
     },
     {
-      "epoch": 3.92,
-      "learning_rate": 0.00012162162162162163,
-      "loss": 0.4699,
+      "epoch": 8.21,
+      "learning_rate": 8.962264150943396e-06,
+      "loss": 0.6416,
       "step": 435
     },
     {
-      "epoch": 3.93,
-      "learning_rate": 0.00012144144144144144,
-      "loss": 0.4862,
+      "epoch": 8.23,
+      "learning_rate": 8.867924528301887e-06,
+      "loss": 0.6744,
       "step": 436
     },
     {
-      "epoch": 3.94,
-      "learning_rate": 0.00012126126126126127,
-      "loss": 0.5477,
+      "epoch": 8.25,
+      "learning_rate": 8.773584905660376e-06,
+      "loss": 0.5623,
       "step": 437
     },
     {
-      "epoch": 3.95,
-      "learning_rate": 0.00012108108108108109,
-      "loss": 0.5063,
+      "epoch": 8.26,
+      "learning_rate": 8.679245283018868e-06,
+      "loss": 0.6513,
       "step": 438
     },
     {
-      "epoch": 3.95,
-      "learning_rate": 0.00012090090090090092,
-      "loss": 0.4708,
+      "epoch": 8.28,
+      "learning_rate": 8.584905660377359e-06,
+      "loss": 0.6099,
       "step": 439
     },
     {
-      "epoch": 3.96,
-      "learning_rate": 0.00012072072072072073,
-      "loss": 0.3925,
+      "epoch": 8.3,
+      "learning_rate": 8.49056603773585e-06,
+      "loss": 0.5376,
       "step": 440
     },
     {
-      "epoch": 3.97,
-      "learning_rate": 0.00012054054054054053,
-      "loss": 0.5277,
+      "epoch": 8.32,
+      "learning_rate": 8.396226415094339e-06,
+      "loss": 0.657,
       "step": 441
     },
     {
-      "epoch": 3.98,
-      "learning_rate": 0.00012036036036036036,
-      "loss": 0.5806,
+      "epoch": 8.34,
+      "learning_rate": 8.30188679245283e-06,
+      "loss": 0.62,
       "step": 442
     },
     {
-      "epoch": 3.99,
-      "learning_rate": 0.00012018018018018017,
-      "loss": 0.4542,
+      "epoch": 8.36,
+      "learning_rate": 8.207547169811321e-06,
+      "loss": 0.6364,
       "step": 443
     },
     {
-      "epoch": 4.0,
-      "learning_rate": 0.00012,
-      "loss": 0.6045,
+      "epoch": 8.38,
+      "learning_rate": 8.11320754716981e-06,
+      "loss": 0.7565,
       "step": 444
     },
     {
-      "epoch": 4.01,
-      "learning_rate": 0.00011981981981981982,
-      "loss": 0.3265,
+      "epoch": 8.4,
+      "learning_rate": 8.018867924528302e-06,
+      "loss": 0.6923,
       "step": 445
     },
     {
-      "epoch": 4.02,
-      "learning_rate": 0.00011963963963963965,
-      "loss": 0.3141,
+      "epoch": 8.42,
+      "learning_rate": 7.924528301886793e-06,
+      "loss": 0.6294,
       "step": 446
     },
     {
-      "epoch": 4.03,
-      "learning_rate": 0.00011945945945945946,
-      "loss": 0.312,
+      "epoch": 8.43,
+      "learning_rate": 7.830188679245282e-06,
+      "loss": 0.6152,
       "step": 447
     },
     {
-      "epoch": 4.04,
-      "learning_rate": 0.00011927927927927929,
-      "loss": 0.3917,
+      "epoch": 8.45,
+      "learning_rate": 7.735849056603773e-06,
+      "loss": 0.6197,
       "step": 448
     },
     {
-      "epoch": 4.05,
-      "learning_rate": 0.0001190990990990991,
-      "loss": 0.4704,
+      "epoch": 8.47,
+      "learning_rate": 7.641509433962264e-06,
+      "loss": 0.6689,
       "step": 449
     },
     {
-      "epoch": 4.05,
-      "learning_rate": 0.00011891891891891893,
-      "loss": 0.3541,
+      "epoch": 8.49,
+      "learning_rate": 7.547169811320755e-06,
+      "loss": 0.5863,
       "step": 450
     },
     {
-      "epoch": 4.06,
-      "learning_rate": 0.00011873873873873875,
-      "loss": 0.401,
+      "epoch": 8.51,
+      "learning_rate": 7.452830188679246e-06,
+      "loss": 0.6549,
       "step": 451
     },
     {
-      "epoch": 4.07,
-      "learning_rate": 0.00011855855855855858,
-      "loss": 0.3266,
+      "epoch": 8.53,
+      "learning_rate": 7.358490566037736e-06,
+      "loss": 0.6476,
       "step": 452
     },
     {
-      "epoch": 4.08,
-      "learning_rate": 0.00011837837837837838,
-      "loss": 0.3155,
+      "epoch": 8.55,
+      "learning_rate": 7.264150943396226e-06,
+      "loss": 0.5644,
       "step": 453
     },
     {
-      "epoch": 4.09,
-      "learning_rate": 0.00011819819819819819,
-      "loss": 0.3531,
+      "epoch": 8.57,
+      "learning_rate": 7.169811320754717e-06,
+      "loss": 0.5327,
       "step": 454
     },
     {
-      "epoch": 4.1,
-      "learning_rate": 0.00011801801801801802,
-      "loss": 0.3775,
+      "epoch": 8.58,
+      "learning_rate": 7.0754716981132075e-06,
+      "loss": 0.6086,
       "step": 455
     },
     {
-      "epoch": 4.11,
-      "learning_rate": 0.00011783783783783784,
-      "loss": 0.3703,
+      "epoch": 8.6,
+      "learning_rate": 6.981132075471699e-06,
+      "loss": 0.7013,
       "step": 456
     },
     {
-      "epoch": 4.12,
-      "learning_rate": 0.00011765765765765766,
-      "loss": 0.3814,
+      "epoch": 8.62,
+      "learning_rate": 6.886792452830189e-06,
+      "loss": 0.6546,
       "step": 457
     },
     {
-      "epoch": 4.13,
-      "learning_rate": 0.00011747747747747748,
-      "loss": 0.3786,
+      "epoch": 8.64,
+      "learning_rate": 6.792452830188679e-06,
+      "loss": 0.636,
       "step": 458
     },
     {
-      "epoch": 4.14,
-      "learning_rate": 0.00011729729729729731,
-      "loss": 0.3343,
+      "epoch": 8.66,
+      "learning_rate": 6.69811320754717e-06,
+      "loss": 0.6997,
       "step": 459
     },
     {
-      "epoch": 4.14,
-      "learning_rate": 0.00011711711711711712,
-      "loss": 0.3516,
+      "epoch": 8.68,
+      "learning_rate": 6.60377358490566e-06,
+      "loss": 0.6163,
       "step": 460
     },
     {
-      "epoch": 4.15,
-      "learning_rate": 0.00011693693693693695,
-      "loss": 0.3729,
+      "epoch": 8.7,
+      "learning_rate": 6.5094339622641515e-06,
+      "loss": 0.6237,
       "step": 461
     },
     {
-      "epoch": 4.16,
-      "learning_rate": 0.00011675675675675676,
-      "loss": 0.3886,
+      "epoch": 8.72,
+      "learning_rate": 6.415094339622642e-06,
+      "loss": 0.631,
       "step": 462
     },
     {
-      "epoch": 4.17,
-      "learning_rate": 0.0001165765765765766,
-      "loss": 0.364,
+      "epoch": 8.74,
+      "learning_rate": 6.320754716981132e-06,
+      "loss": 0.5505,
       "step": 463
     },
     {
-      "epoch": 4.18,
-      "learning_rate": 0.00011639639639639641,
-      "loss": 0.3486,
+      "epoch": 8.75,
+      "learning_rate": 6.226415094339623e-06,
+      "loss": 0.5707,
       "step": 464
     },
     {
-      "epoch": 4.19,
-      "learning_rate": 0.00011621621621621621,
-      "loss": 0.3411,
+      "epoch": 8.77,
+      "learning_rate": 6.132075471698113e-06,
+      "loss": 0.6712,
       "step": 465
     },
     {
-      "epoch": 4.2,
-      "learning_rate": 0.00011603603603603604,
-      "loss": 0.4359,
+      "epoch": 8.79,
+      "learning_rate": 6.037735849056604e-06,
+      "loss": 0.6895,
       "step": 466
     },
     {
-      "epoch": 4.21,
-      "learning_rate": 0.00011585585585585585,
-      "loss": 0.336,
+      "epoch": 8.81,
+      "learning_rate": 5.943396226415095e-06,
+      "loss": 0.6574,
       "step": 467
     },
     {
-      "epoch": 4.22,
-      "learning_rate": 0.00011567567567567568,
-      "loss": 0.4453,
+      "epoch": 8.83,
+      "learning_rate": 5.849056603773585e-06,
+      "loss": 0.6445,
       "step": 468
     },
     {
-      "epoch": 4.23,
-      "learning_rate": 0.0001154954954954955,
-      "loss": 0.3626,
+      "epoch": 8.85,
+      "learning_rate": 5.754716981132076e-06,
+      "loss": 0.5688,
       "step": 469
     },
     {
-      "epoch": 4.23,
-      "learning_rate": 0.00011531531531531532,
-      "loss": 0.3608,
+      "epoch": 8.87,
+      "learning_rate": 5.660377358490566e-06,
+      "loss": 0.6162,
       "step": 470
     },
     {
-      "epoch": 4.24,
-      "learning_rate": 0.00011513513513513514,
-      "loss": 0.3803,
+      "epoch": 8.89,
+      "learning_rate": 5.566037735849057e-06,
+      "loss": 0.5972,
       "step": 471
     },
     {
-      "epoch": 4.25,
-      "learning_rate": 0.00011495495495495497,
-      "loss": 0.3862,
+      "epoch": 8.91,
+      "learning_rate": 5.4716981132075475e-06,
+      "loss": 0.5262,
       "step": 472
     },
     {
-      "epoch": 4.26,
-      "learning_rate": 0.00011477477477477478,
-      "loss": 0.3639,
+      "epoch": 8.92,
+      "learning_rate": 5.377358490566039e-06,
+      "loss": 0.7048,
       "step": 473
     },
     {
-      "epoch": 4.27,
-      "learning_rate": 0.00011459459459459461,
-      "loss": 0.3345,
+      "epoch": 8.94,
+      "learning_rate": 5.283018867924529e-06,
+      "loss": 0.6142,
       "step": 474
     },
     {
-      "epoch": 4.28,
-      "learning_rate": 0.00011441441441441443,
-      "loss": 0.3343,
+      "epoch": 8.96,
+      "learning_rate": 5.188679245283019e-06,
+      "loss": 0.6144,
       "step": 475
     },
     {
-      "epoch": 4.29,
-      "learning_rate": 0.00011423423423423425,
-      "loss": 0.3943,
+      "epoch": 8.98,
+      "learning_rate": 5.09433962264151e-06,
+      "loss": 0.6002,
       "step": 476
     },
     {
-      "epoch": 4.3,
-      "learning_rate": 0.00011405405405405406,
-      "loss": 0.3802,
+      "epoch": 9.0,
+      "learning_rate": 5e-06,
+      "loss": 0.5671,
       "step": 477
     },
     {
-      "epoch": 4.31,
-      "learning_rate": 0.00011387387387387387,
-      "loss": 0.3523,
+      "epoch": 9.02,
+      "learning_rate": 4.9056603773584915e-06,
+      "loss": 0.5556,
       "step": 478
     },
     {
-      "epoch": 4.32,
-      "learning_rate": 0.0001136936936936937,
-      "loss": 0.4398,
+      "epoch": 9.04,
+      "learning_rate": 4.811320754716982e-06,
+      "loss": 0.6501,
       "step": 479
     },
     {
-      "epoch": 4.32,
-      "learning_rate": 0.00011351351351351351,
-      "loss": 0.3851,
+      "epoch": 9.06,
+      "learning_rate": 4.716981132075472e-06,
+      "loss": 0.5688,
       "step": 480
     },
     {
-      "epoch": 4.33,
-      "learning_rate": 0.00011333333333333334,
-      "loss": 0.3755,
+      "epoch": 9.08,
+      "learning_rate": 4.622641509433962e-06,
+      "loss": 0.5437,
       "step": 481
     },
-    {
-      "epoch": 4.34,
-      "learning_rate": 0.00011315315315315316,
-      "loss": 0.3402,
-      "step": 482
-    },
-    {
-      "epoch": 4.35,
-      "learning_rate": 0.00011297297297297298,
-      "loss": 0.3227,
-      "step": 483
-    },
-    {
-      "epoch": 4.36,
-      "learning_rate": 0.0001127927927927928,
-      "loss": 0.3282,
-      "step": 484
-    },
-    {
-      "epoch": 4.37,
-      "learning_rate": 0.00011261261261261263,
-      "loss": 0.3621,
-      "step": 485
-    },
-    {
-      "epoch": 4.38,
-      "learning_rate": 0.00011243243243243244,
-      "loss": 0.3507,
-      "step": 486
-    },
-    {
-      "epoch": 4.39,
-      "learning_rate": 0.00011225225225225227,
-      "loss": 0.412,
-      "step": 487
-    },
-    {
-      "epoch": 4.4,
-      "learning_rate": 0.00011207207207207209,
-      "loss": 0.5337,
-      "step": 488
-    },
-    {
-      "epoch": 4.41,
-      "learning_rate": 0.00011189189189189189,
-      "loss": 0.4157,
-      "step": 489
-    },
-    {
-      "epoch": 4.41,
-      "learning_rate": 0.0001117117117117117,
-      "loss": 0.411,
-      "step": 490
-    },
-    {
-      "epoch": 4.42,
-      "learning_rate": 0.00011153153153153153,
-      "loss": 0.4512,
-      "step": 491
-    },
-    {
-      "epoch": 4.43,
-      "learning_rate": 0.00011135135135135135,
-      "loss": 0.3734,
-      "step": 492
-    },
-    {
-      "epoch": 4.44,
-      "learning_rate": 0.00011117117117117117,
-      "loss": 0.3837,
-      "step": 493
-    },
-    {
-      "epoch": 4.45,
-      "learning_rate": 0.00011099099099099099,
-      "loss": 0.4167,
-      "step": 494
-    },
-    {
-      "epoch": 4.46,
-      "learning_rate": 0.00011081081081081082,
-      "loss": 0.3796,
-      "step": 495
-    },
-    {
-      "epoch": 4.47,
-      "learning_rate": 0.00011063063063063063,
-      "loss": 0.3891,
-      "step": 496
-    },
-    {
-      "epoch": 4.48,
-      "learning_rate": 0.00011045045045045046,
-      "loss": 0.3086,
-      "step": 497
-    },
-    {
-      "epoch": 4.49,
-      "learning_rate": 0.00011027027027027029,
-      "loss": 0.4173,
-      "step": 498
-    },
-    {
-      "epoch": 4.5,
-      "learning_rate": 0.0001100900900900901,
-      "loss": 0.3275,
-      "step": 499
-    },
-    {
-      "epoch": 4.5,
-      "learning_rate": 0.00010990990990990993,
-      "loss": 0.357,
-      "step": 500
-    },
-    {
-      "epoch": 4.51,
-      "learning_rate": 0.00010972972972972972,
-      "loss": 0.4611,
-      "step": 501
-    },
-    {
-      "epoch": 4.52,
-      "learning_rate": 0.00010954954954954955,
-      "loss": 0.3544,
-      "step": 502
-    },
-    {
-      "epoch": 4.53,
-      "learning_rate": 0.00010936936936936936,
-      "loss": 0.3564,
-      "step": 503
-    },
-    {
-      "epoch": 4.54,
-      "learning_rate": 0.00010918918918918919,
-      "loss": 0.3306,
-      "step": 504
-    },
-    {
-      "epoch": 4.55,
-      "learning_rate": 0.000109009009009009,
-      "loss": 0.4079,
-      "step": 505
-    },
-    {
-      "epoch": 4.56,
-      "learning_rate": 0.00010882882882882883,
-      "loss": 0.3906,
-      "step": 506
-    },
-    {
-      "epoch": 4.57,
-      "learning_rate": 0.00010864864864864865,
-      "loss": 0.3863,
-      "step": 507
-    },
-    {
-      "epoch": 4.58,
-      "learning_rate": 0.00010846846846846848,
-      "loss": 0.3378,
-      "step": 508
-    },
-    {
-      "epoch": 4.59,
-      "learning_rate": 0.00010828828828828829,
-      "loss": 0.3556,
-      "step": 509
-    },
-    {
-      "epoch": 4.59,
-      "learning_rate": 0.00010810810810810812,
-      "loss": 0.2989,
-      "step": 510
-    },
-    {
-      "epoch": 4.6,
-      "learning_rate": 0.00010792792792792794,
-      "loss": 0.4635,
-      "step": 511
-    },
-    {
-      "epoch": 4.61,
-      "learning_rate": 0.00010774774774774776,
-      "loss": 0.3909,
-      "step": 512
-    },
-    {
-      "epoch": 4.62,
-      "learning_rate": 0.00010756756756756757,
-      "loss": 0.3124,
-      "step": 513
-    },
-    {
-      "epoch": 4.63,
-      "learning_rate": 0.00010738738738738738,
-      "loss": 0.4026,
-      "step": 514
-    },
-    {
-      "epoch": 4.64,
-      "learning_rate": 0.00010720720720720721,
-      "loss": 0.3955,
-      "step": 515
-    },
-    {
-      "epoch": 4.65,
-      "learning_rate": 0.00010702702702702702,
-      "loss": 0.3776,
-      "step": 516
-    },
-    {
-      "epoch": 4.66,
-      "learning_rate": 0.00010684684684684685,
-      "loss": 0.3324,
-      "step": 517
-    },
-    {
-      "epoch": 4.67,
-      "learning_rate": 0.00010666666666666667,
-      "loss": 0.3567,
-      "step": 518
-    },
-    {
-      "epoch": 4.68,
-      "learning_rate": 0.0001064864864864865,
-      "loss": 0.3758,
-      "step": 519
-    },
-    {
-      "epoch": 4.68,
-      "learning_rate": 0.00010630630630630631,
-      "loss": 0.3658,
-      "step": 520
-    },
-    {
-      "epoch": 4.69,
-      "learning_rate": 0.00010612612612612614,
-      "loss": 0.4016,
-      "step": 521
-    },
-    {
-      "epoch": 4.7,
-      "learning_rate": 0.00010594594594594595,
-      "loss": 0.3351,
-      "step": 522
-    },
-    {
-      "epoch": 4.71,
-      "learning_rate": 0.00010576576576576578,
-      "loss": 0.4547,
-      "step": 523
-    },
-    {
-      "epoch": 4.72,
-      "learning_rate": 0.0001055855855855856,
-      "loss": 0.4222,
-      "step": 524
-    },
-    {
-      "epoch": 4.73,
-      "learning_rate": 0.0001054054054054054,
-      "loss": 0.3836,
-      "step": 525
-    },
-    {
-      "epoch": 4.74,
-      "learning_rate": 0.00010522522522522523,
-      "loss": 0.3061,
-      "step": 526
-    },
-    {
-      "epoch": 4.75,
-      "learning_rate": 0.00010504504504504504,
-      "loss": 0.3326,
-      "step": 527
-    },
-    {
-      "epoch": 4.76,
-      "learning_rate": 0.00010486486486486487,
-      "loss": 0.3756,
-      "step": 528
-    },
-    {
-      "epoch": 4.77,
-      "learning_rate": 0.00010468468468468468,
-      "loss": 0.3527,
-      "step": 529
-    },
-    {
-      "epoch": 4.77,
-      "learning_rate": 0.00010450450450450451,
-      "loss": 0.3688,
-      "step": 530
-    },
-    {
-      "epoch": 4.78,
-      "learning_rate": 0.00010432432432432433,
-      "loss": 0.3322,
-      "step": 531
-    },
-    {
-      "epoch": 4.79,
-      "learning_rate": 0.00010414414414414416,
-      "loss": 0.4191,
-      "step": 532
-    },
-    {
-      "epoch": 4.8,
-      "learning_rate": 0.00010396396396396397,
-      "loss": 0.3586,
-      "step": 533
-    },
-    {
-      "epoch": 4.81,
-      "learning_rate": 0.0001037837837837838,
-      "loss": 0.4528,
-      "step": 534
-    },
-    {
-      "epoch": 4.82,
-      "learning_rate": 0.00010360360360360361,
-      "loss": 0.2971,
-      "step": 535
-    },
-    {
-      "epoch": 4.83,
-      "learning_rate": 0.00010342342342342344,
-      "loss": 0.3338,
-      "step": 536
-    },
-    {
-      "epoch": 4.84,
-      "learning_rate": 0.00010324324324324324,
-      "loss": 0.3805,
-      "step": 537
-    },
-    {
-      "epoch": 4.85,
-      "learning_rate": 0.00010306306306306306,
-      "loss": 0.3958,
-      "step": 538
-    },
-    {
-      "epoch": 4.86,
-      "learning_rate": 0.00010288288288288289,
-      "loss": 0.3995,
-      "step": 539
-    },
-    {
-      "epoch": 4.86,
-      "learning_rate": 0.0001027027027027027,
-      "loss": 0.3661,
-      "step": 540
-    },
-    {
-      "epoch": 4.87,
-      "learning_rate": 0.00010252252252252253,
-      "loss": 0.2927,
-      "step": 541
-    },
-    {
-      "epoch": 4.88,
-      "learning_rate": 0.00010234234234234234,
-      "loss": 0.4004,
-      "step": 542
-    },
-    {
-      "epoch": 4.89,
-      "learning_rate": 0.00010216216216216217,
-      "loss": 0.3619,
-      "step": 543
-    },
-    {
-      "epoch": 4.9,
-      "learning_rate": 0.00010198198198198199,
-      "loss": 0.3478,
-      "step": 544
-    },
-    {
-      "epoch": 4.91,
-      "learning_rate": 0.00010180180180180182,
-      "loss": 0.417,
-      "step": 545
-    },
-    {
-      "epoch": 4.92,
-      "learning_rate": 0.00010162162162162163,
-      "loss": 0.4245,
-      "step": 546
-    },
-    {
-      "epoch": 4.93,
-      "learning_rate": 0.00010144144144144146,
-      "loss": 0.319,
-      "step": 547
-    },
-    {
-      "epoch": 4.94,
-      "learning_rate": 0.00010126126126126127,
-      "loss": 0.281,
-      "step": 548
-    },
-    {
-      "epoch": 4.95,
-      "learning_rate": 0.00010108108108108108,
-      "loss": 0.3438,
-      "step": 549
-    },
-    {
-      "epoch": 4.95,
-      "learning_rate": 0.00010090090090090089,
-      "loss": 0.3441,
-      "step": 550
-    },
-    {
-      "epoch": 4.96,
-      "learning_rate": 0.00010072072072072072,
-      "loss": 0.3684,
-      "step": 551
-    },
-    {
-      "epoch": 4.97,
-      "learning_rate": 0.00010054054054054053,
-      "loss": 0.3679,
-      "step": 552
-    },
-    {
-      "epoch": 4.98,
-      "learning_rate": 0.00010036036036036036,
-      "loss": 0.3563,
-      "step": 553
-    },
-    {
-      "epoch": 4.99,
-      "learning_rate": 0.00010018018018018018,
-      "loss": 0.4197,
-      "step": 554
-    },
-    {
-      "epoch": 5.0,
-      "learning_rate": 0.0001,
-      "loss": 0.4142,
-      "step": 555
-    },
-    {
-      "epoch": 5.01,
-      "learning_rate": 9.981981981981983e-05,
-      "loss": 0.2757,
-      "step": 556
-    },
-    {
-      "epoch": 5.02,
-      "learning_rate": 9.963963963963965e-05,
-      "loss": 0.3093,
-      "step": 557
-    },
-    {
-      "epoch": 5.03,
-      "learning_rate": 9.945945945945948e-05,
-      "loss": 0.2858,
-      "step": 558
-    },
-    {
-      "epoch": 5.04,
-      "learning_rate": 9.927927927927928e-05,
-      "loss": 0.2819,
-      "step": 559
-    },
-    {
-      "epoch": 5.05,
-      "learning_rate": 9.90990990990991e-05,
-      "loss": 0.2873,
-      "step": 560
-    },
-    {
-      "epoch": 5.05,
-      "learning_rate": 9.891891891891892e-05,
-      "loss": 0.2807,
-      "step": 561
-    },
-    {
-      "epoch": 5.06,
-      "learning_rate": 9.873873873873875e-05,
-      "loss": 0.3117,
-      "step": 562
-    },
-    {
-      "epoch": 5.07,
-      "learning_rate": 9.855855855855856e-05,
-      "loss": 0.2886,
-      "step": 563
-    },
-    {
-      "epoch": 5.08,
-      "learning_rate": 9.837837837837839e-05,
-      "loss": 0.2797,
-      "step": 564
-    },
-    {
-      "epoch": 5.09,
-      "learning_rate": 9.81981981981982e-05,
-      "loss": 0.3165,
-      "step": 565
-    },
-    {
-      "epoch": 5.1,
-      "learning_rate": 9.801801801801802e-05,
-      "loss": 0.3353,
-      "step": 566
-    },
-    {
-      "epoch": 5.11,
-      "learning_rate": 9.783783783783784e-05,
-      "loss": 0.2902,
-      "step": 567
-    },
-    {
-      "epoch": 5.12,
-      "learning_rate": 9.765765765765767e-05,
-      "loss": 0.2807,
-      "step": 568
-    },
-    {
-      "epoch": 5.13,
-      "learning_rate": 9.747747747747748e-05,
-      "loss": 0.3003,
-      "step": 569
-    },
-    {
-      "epoch": 5.14,
-      "learning_rate": 9.729729729729731e-05,
-      "loss": 0.2983,
-      "step": 570
-    },
-    {
-      "epoch": 5.14,
-      "learning_rate": 9.711711711711712e-05,
-      "loss": 0.2716,
-      "step": 571
-    },
-    {
-      "epoch": 5.15,
-      "learning_rate": 9.693693693693694e-05,
-      "loss": 0.3266,
-      "step": 572
-    },
-    {
-      "epoch": 5.16,
-      "learning_rate": 9.675675675675677e-05,
-      "loss": 0.2585,
-      "step": 573
-    },
-    {
-      "epoch": 5.17,
-      "learning_rate": 9.657657657657658e-05,
-      "loss": 0.2978,
-      "step": 574
-    },
-    {
-      "epoch": 5.18,
-      "learning_rate": 9.639639639639641e-05,
-      "loss": 0.3214,
-      "step": 575
-    },
-    {
-      "epoch": 5.19,
-      "learning_rate": 9.621621621621622e-05,
-      "loss": 0.2546,
-      "step": 576
-    },
-    {
-      "epoch": 5.2,
-      "learning_rate": 9.603603603603604e-05,
-      "loss": 0.2801,
-      "step": 577
-    },
-    {
-      "epoch": 5.21,
-      "learning_rate": 9.585585585585585e-05,
-      "loss": 0.2491,
-      "step": 578
-    },
-    {
-      "epoch": 5.22,
-      "learning_rate": 9.567567567567568e-05,
-      "loss": 0.2667,
-      "step": 579
-    },
-    {
-      "epoch": 5.23,
-      "learning_rate": 9.54954954954955e-05,
-      "loss": 0.2588,
-      "step": 580
-    },
-    {
-      "epoch": 5.23,
-      "learning_rate": 9.531531531531533e-05,
-      "loss": 0.2887,
-      "step": 581
-    },
-    {
-      "epoch": 5.24,
-      "learning_rate": 9.513513513513514e-05,
-      "loss": 0.2701,
-      "step": 582
-    },
-    {
-      "epoch": 5.25,
-      "learning_rate": 9.495495495495496e-05,
-      "loss": 0.2838,
-      "step": 583
-    },
-    {
-      "epoch": 5.26,
-      "learning_rate": 9.477477477477478e-05,
-      "loss": 0.3002,
-      "step": 584
-    },
-    {
-      "epoch": 5.27,
-      "learning_rate": 9.45945945945946e-05,
-      "loss": 0.2738,
-      "step": 585
-    },
-    {
-      "epoch": 5.28,
-      "learning_rate": 9.441441441441443e-05,
-      "loss": 0.2904,
-      "step": 586
-    },
-    {
-      "epoch": 5.29,
-      "learning_rate": 9.423423423423424e-05,
-      "loss": 0.2691,
-      "step": 587
-    },
-    {
-      "epoch": 5.3,
-      "learning_rate": 9.405405405405407e-05,
-      "loss": 0.3163,
-      "step": 588
-    },
-    {
-      "epoch": 5.31,
-      "learning_rate": 9.387387387387387e-05,
-      "loss": 0.2756,
-      "step": 589
-    },
-    {
-      "epoch": 5.32,
-      "learning_rate": 9.36936936936937e-05,
-      "loss": 0.2878,
-      "step": 590
-    },
-    {
-      "epoch": 5.32,
-      "learning_rate": 9.351351351351351e-05,
-      "loss": 0.2815,
-      "step": 591
-    },
-    {
-      "epoch": 5.33,
-      "learning_rate": 9.333333333333334e-05,
-      "loss": 0.3392,
-      "step": 592
-    },
-    {
-      "epoch": 5.34,
-      "learning_rate": 9.315315315315316e-05,
-      "loss": 0.3289,
-      "step": 593
-    },
-    {
-      "epoch": 5.35,
-      "learning_rate": 9.297297297297299e-05,
-      "loss": 0.3027,
-      "step": 594
-    },
-    {
-      "epoch": 5.36,
-      "learning_rate": 9.279279279279279e-05,
-      "loss": 0.3511,
-      "step": 595
-    },
-    {
-      "epoch": 5.37,
-      "learning_rate": 9.261261261261262e-05,
-      "loss": 0.2502,
-      "step": 596
-    },
-    {
-      "epoch": 5.38,
-      "learning_rate": 9.243243243243243e-05,
-      "loss": 0.281,
-      "step": 597
-    },
-    {
-      "epoch": 5.39,
-      "learning_rate": 9.225225225225226e-05,
-      "loss": 0.3211,
-      "step": 598
-    },
-    {
-      "epoch": 5.4,
-      "learning_rate": 9.207207207207207e-05,
-      "loss": 0.2933,
-      "step": 599
-    },
-    {
-      "epoch": 5.41,
-      "learning_rate": 9.18918918918919e-05,
-      "loss": 0.3106,
-      "step": 600
-    },
-    {
-      "epoch": 5.41,
-      "learning_rate": 9.171171171171172e-05,
-      "loss": 0.2699,
-      "step": 601
-    },
-    {
-      "epoch": 5.42,
-      "learning_rate": 9.153153153153153e-05,
-      "loss": 0.2849,
-      "step": 602
-    },
-    {
-      "epoch": 5.43,
-      "learning_rate": 9.135135135135136e-05,
-      "loss": 0.2974,
-      "step": 603
-    },
-    {
-      "epoch": 5.44,
-      "learning_rate": 9.117117117117118e-05,
-      "loss": 0.2942,
-      "step": 604
-    },
-    {
-      "epoch": 5.45,
-      "learning_rate": 9.0990990990991e-05,
-      "loss": 0.2788,
-      "step": 605
-    },
-    {
-      "epoch": 5.46,
-      "learning_rate": 9.081081081081082e-05,
-      "loss": 0.2658,
-      "step": 606
-    },
-    {
-      "epoch": 5.47,
-      "learning_rate": 9.063063063063063e-05,
-      "loss": 0.3066,
-      "step": 607
-    },
-    {
-      "epoch": 5.48,
-      "learning_rate": 9.045045045045045e-05,
-      "loss": 0.3215,
-      "step": 608
-    },
-    {
-      "epoch": 5.49,
-      "learning_rate": 9.027027027027028e-05,
-      "loss": 0.2737,
-      "step": 609
-    },
-    {
-      "epoch": 5.5,
-      "learning_rate": 9.009009009009009e-05,
-      "loss": 0.279,
-      "step": 610
-    },
-    {
-      "epoch": 5.5,
-      "learning_rate": 8.990990990990992e-05,
-      "loss": 0.2721,
-      "step": 611
-    },
-    {
-      "epoch": 5.51,
-      "learning_rate": 8.972972972972973e-05,
-      "loss": 0.2429,
-      "step": 612
-    },
-    {
-      "epoch": 5.52,
-      "learning_rate": 8.954954954954955e-05,
-      "loss": 0.2799,
-      "step": 613
-    },
-    {
-      "epoch": 5.53,
-      "learning_rate": 8.936936936936938e-05,
-      "loss": 0.2756,
-      "step": 614
-    },
-    {
-      "epoch": 5.54,
-      "learning_rate": 8.918918918918919e-05,
-      "loss": 0.2578,
-      "step": 615
-    },
-    {
-      "epoch": 5.55,
-      "learning_rate": 8.900900900900902e-05,
-      "loss": 0.3057,
-      "step": 616
-    },
-    {
-      "epoch": 5.56,
-      "learning_rate": 8.882882882882884e-05,
-      "loss": 0.262,
-      "step": 617
-    },
-    {
-      "epoch": 5.57,
-      "learning_rate": 8.864864864864866e-05,
-      "loss": 0.3209,
-      "step": 618
-    },
-    {
-      "epoch": 5.58,
-      "learning_rate": 8.846846846846847e-05,
-      "loss": 0.2809,
-      "step": 619
-    },
-    {
-      "epoch": 5.59,
-      "learning_rate": 8.82882882882883e-05,
-      "loss": 0.299,
-      "step": 620
-    },
-    {
-      "epoch": 5.59,
-      "learning_rate": 8.810810810810811e-05,
-      "loss": 0.3359,
-      "step": 621
-    },
-    {
-      "epoch": 5.6,
-      "learning_rate": 8.792792792792794e-05,
-      "loss": 0.3272,
-      "step": 622
-    },
-    {
-      "epoch": 5.61,
-      "learning_rate": 8.774774774774775e-05,
-      "loss": 0.3184,
-      "step": 623
-    },
-    {
-      "epoch": 5.62,
-      "learning_rate": 8.756756756756758e-05,
-      "loss": 0.3008,
-      "step": 624
-    },
-    {
-      "epoch": 5.63,
-      "learning_rate": 8.738738738738738e-05,
-      "loss": 0.2611,
-      "step": 625
-    },
-    {
-      "epoch": 5.64,
-      "learning_rate": 8.720720720720721e-05,
-      "loss": 0.3845,
-      "step": 626
-    },
-    {
-      "epoch": 5.65,
-      "learning_rate": 8.702702702702702e-05,
-      "loss": 0.2434,
-      "step": 627
-    },
-    {
-      "epoch": 5.66,
-      "learning_rate": 8.684684684684685e-05,
-      "loss": 0.3178,
-      "step": 628
-    },
-    {
-      "epoch": 5.67,
-      "learning_rate": 8.666666666666667e-05,
-      "loss": 0.2643,
-      "step": 629
-    },
-    {
-      "epoch": 5.68,
-      "learning_rate": 8.64864864864865e-05,
-      "loss": 0.3346,
-      "step": 630
-    },
-    {
-      "epoch": 5.68,
-      "learning_rate": 8.630630630630631e-05,
-      "loss": 0.3385,
-      "step": 631
-    },
-    {
-      "epoch": 5.69,
-      "learning_rate": 8.612612612612613e-05,
-      "loss": 0.2635,
-      "step": 632
-    },
-    {
-      "epoch": 5.7,
-      "learning_rate": 8.594594594594595e-05,
-      "loss": 0.2559,
-      "step": 633
-    },
-    {
-      "epoch": 5.71,
-      "learning_rate": 8.576576576576577e-05,
-      "loss": 0.2956,
-      "step": 634
-    },
-    {
-      "epoch": 5.72,
-      "learning_rate": 8.55855855855856e-05,
-      "loss": 0.3279,
-      "step": 635
-    },
-    {
-      "epoch": 5.73,
-      "learning_rate": 8.540540540540541e-05,
-      "loss": 0.2796,
-      "step": 636
-    },
-    {
-      "epoch": 5.74,
-      "learning_rate": 8.522522522522523e-05,
-      "loss": 0.32,
-      "step": 637
-    },
-    {
-      "epoch": 5.75,
-      "learning_rate": 8.504504504504504e-05,
-      "loss": 0.2425,
-      "step": 638
-    },
-    {
-      "epoch": 5.76,
-      "learning_rate": 8.486486486486487e-05,
-      "loss": 0.3221,
-      "step": 639
-    },
-    {
-      "epoch": 5.77,
-      "learning_rate": 8.468468468468469e-05,
-      "loss": 0.3018,
-      "step": 640
-    },
-    {
-      "epoch": 5.77,
-      "learning_rate": 8.450450450450451e-05,
-      "loss": 0.2249,
-      "step": 641
-    },
-    {
-      "epoch": 5.78,
-      "learning_rate": 8.432432432432433e-05,
-      "loss": 0.2763,
-      "step": 642
-    },
-    {
-      "epoch": 5.79,
-      "learning_rate": 8.414414414414414e-05,
-      "loss": 0.2784,
-      "step": 643
-    },
-    {
-      "epoch": 5.8,
-      "learning_rate": 8.396396396396397e-05,
-      "loss": 0.3659,
-      "step": 644
-    },
-    {
-      "epoch": 5.81,
-      "learning_rate": 8.378378378378379e-05,
-      "loss": 0.2453,
-      "step": 645
-    },
-    {
-      "epoch": 5.82,
-      "learning_rate": 8.360360360360362e-05,
-      "loss": 0.3197,
-      "step": 646
-    },
-    {
-      "epoch": 5.83,
-      "learning_rate": 8.342342342342343e-05,
-      "loss": 0.3055,
-      "step": 647
-    },
-    {
-      "epoch": 5.84,
-      "learning_rate": 8.324324324324326e-05,
-      "loss": 0.2642,
-      "step": 648
-    },
-    {
-      "epoch": 5.85,
-      "learning_rate": 8.306306306306306e-05,
-      "loss": 0.2442,
-      "step": 649
-    },
-    {
-      "epoch": 5.86,
-      "learning_rate": 8.288288288288289e-05,
-      "loss": 0.274,
-      "step": 650
-    },
-    {
-      "epoch": 5.86,
-      "learning_rate": 8.27027027027027e-05,
-      "loss": 0.2701,
-      "step": 651
-    },
-    {
-      "epoch": 5.87,
-      "learning_rate": 8.252252252252253e-05,
-      "loss": 0.2802,
-      "step": 652
-    },
-    {
-      "epoch": 5.88,
-      "learning_rate": 8.234234234234235e-05,
-      "loss": 0.2709,
-      "step": 653
-    },
-    {
-      "epoch": 5.89,
-      "learning_rate": 8.216216216216217e-05,
-      "loss": 0.3404,
-      "step": 654
-    },
-    {
-      "epoch": 5.9,
-      "learning_rate": 8.198198198198198e-05,
-      "loss": 0.3093,
-      "step": 655
-    },
-    {
-      "epoch": 5.91,
-      "learning_rate": 8.18018018018018e-05,
-      "loss": 0.3159,
-      "step": 656
-    },
-    {
-      "epoch": 5.92,
-      "learning_rate": 8.162162162162162e-05,
-      "loss": 0.3564,
-      "step": 657
-    },
-    {
-      "epoch": 5.93,
-      "learning_rate": 8.144144144144145e-05,
-      "loss": 0.2453,
-      "step": 658
-    },
-    {
-      "epoch": 5.94,
-      "learning_rate": 8.126126126126126e-05,
-      "loss": 0.2488,
-      "step": 659
-    },
-    {
-      "epoch": 5.95,
-      "learning_rate": 8.108108108108109e-05,
-      "loss": 0.3104,
-      "step": 660
-    },
-    {
-      "epoch": 5.95,
-      "learning_rate": 8.09009009009009e-05,
-      "loss": 0.3627,
-      "step": 661
-    },
-    {
-      "epoch": 5.96,
-      "learning_rate": 8.072072072072072e-05,
-      "loss": 0.3423,
-      "step": 662
-    },
-    {
-      "epoch": 5.97,
-      "learning_rate": 8.054054054054055e-05,
-      "loss": 0.2948,
-      "step": 663
-    },
-    {
-      "epoch": 5.98,
-      "learning_rate": 8.036036036036036e-05,
-      "loss": 0.2747,
-      "step": 664
-    },
-    {
-      "epoch": 5.99,
-      "learning_rate": 8.018018018018019e-05,
-      "loss": 0.2977,
-      "step": 665
-    },
-    {
-      "epoch": 6.0,
-      "learning_rate": 8e-05,
-      "loss": 0.2801,
-      "step": 666
-    },
-    {
-      "epoch": 6.01,
-      "learning_rate": 7.981981981981982e-05,
-      "loss": 0.2422,
-      "step": 667
-    },
-    {
-      "epoch": 6.02,
-      "learning_rate": 7.963963963963964e-05,
-      "loss": 0.2353,
-      "step": 668
-    },
-    {
-      "epoch": 6.03,
-      "learning_rate": 7.945945945945946e-05,
-      "loss": 0.2646,
-      "step": 669
-    },
-    {
-      "epoch": 6.04,
-      "learning_rate": 7.927927927927928e-05,
-      "loss": 0.2443,
-      "step": 670
-    },
-    {
-      "epoch": 6.05,
-      "learning_rate": 7.909909909909911e-05,
-      "loss": 0.2291,
-      "step": 671
-    },
-    {
-      "epoch": 6.05,
-      "learning_rate": 7.891891891891892e-05,
-      "loss": 0.2014,
-      "step": 672
-    },
-    {
-      "epoch": 6.06,
-      "learning_rate": 7.873873873873874e-05,
-      "loss": 0.2264,
-      "step": 673
-    },
-    {
-      "epoch": 6.07,
-      "learning_rate": 7.855855855855857e-05,
-      "loss": 0.2373,
-      "step": 674
-    },
-    {
-      "epoch": 6.08,
-      "learning_rate": 7.837837837837838e-05,
-      "loss": 0.2223,
-      "step": 675
-    },
-    {
-      "epoch": 6.09,
-      "learning_rate": 7.819819819819821e-05,
-      "loss": 0.2465,
-      "step": 676
-    },
-    {
-      "epoch": 6.1,
-      "learning_rate": 7.801801801801802e-05,
-      "loss": 0.2812,
-      "step": 677
-    },
-    {
-      "epoch": 6.11,
-      "learning_rate": 7.783783783783785e-05,
-      "loss": 0.2418,
-      "step": 678
-    },
-    {
-      "epoch": 6.12,
-      "learning_rate": 7.765765765765765e-05,
-      "loss": 0.2626,
-      "step": 679
-    },
-    {
-      "epoch": 6.13,
-      "learning_rate": 7.747747747747748e-05,
-      "loss": 0.2225,
-      "step": 680
-    },
-    {
-      "epoch": 6.14,
-      "learning_rate": 7.72972972972973e-05,
-      "loss": 0.2215,
-      "step": 681
-    },
-    {
-      "epoch": 6.14,
-      "learning_rate": 7.711711711711713e-05,
-      "loss": 0.2634,
-      "step": 682
-    },
-    {
-      "epoch": 6.15,
-      "learning_rate": 7.693693693693694e-05,
-      "loss": 0.2405,
-      "step": 683
-    },
-    {
-      "epoch": 6.16,
-      "learning_rate": 7.675675675675677e-05,
-      "loss": 0.2635,
-      "step": 684
-    },
-    {
-      "epoch": 6.17,
-      "learning_rate": 7.657657657657657e-05,
-      "loss": 0.2163,
-      "step": 685
-    },
-    {
-      "epoch": 6.18,
-      "learning_rate": 7.63963963963964e-05,
-      "loss": 0.2326,
-      "step": 686
-    },
-    {
-      "epoch": 6.19,
-      "learning_rate": 7.621621621621621e-05,
-      "loss": 0.2794,
-      "step": 687
-    },
-    {
-      "epoch": 6.2,
-      "learning_rate": 7.603603603603604e-05,
-      "loss": 0.2338,
-      "step": 688
-    },
-    {
-      "epoch": 6.21,
-      "learning_rate": 7.585585585585586e-05,
-      "loss": 0.2364,
-      "step": 689
-    },
-    {
-      "epoch": 6.22,
-      "learning_rate": 7.567567567567568e-05,
-      "loss": 0.2184,
-      "step": 690
-    },
-    {
-      "epoch": 6.23,
-      "learning_rate": 7.54954954954955e-05,
-      "loss": 0.2297,
-      "step": 691
-    },
-    {
-      "epoch": 6.23,
-      "learning_rate": 7.531531531531531e-05,
-      "loss": 0.2275,
-      "step": 692
-    },
-    {
-      "epoch": 6.24,
-      "learning_rate": 7.513513513513514e-05,
-      "loss": 0.2431,
-      "step": 693
-    },
-    {
-      "epoch": 6.25,
-      "learning_rate": 7.495495495495496e-05,
-      "loss": 0.2014,
-      "step": 694
-    },
-    {
-      "epoch": 6.26,
-      "learning_rate": 7.477477477477479e-05,
-      "loss": 0.232,
-      "step": 695
-    },
-    {
-      "epoch": 6.27,
-      "learning_rate": 7.45945945945946e-05,
-      "loss": 0.2394,
-      "step": 696
-    },
-    {
-      "epoch": 6.28,
-      "learning_rate": 7.441441441441442e-05,
-      "loss": 0.252,
-      "step": 697
-    },
-    {
-      "epoch": 6.29,
-      "learning_rate": 7.423423423423423e-05,
-      "loss": 0.2705,
-      "step": 698
-    },
-    {
-      "epoch": 6.3,
-      "learning_rate": 7.405405405405406e-05,
-      "loss": 0.243,
-      "step": 699
-    },
-    {
-      "epoch": 6.31,
-      "learning_rate": 7.387387387387387e-05,
-      "loss": 0.2024,
-      "step": 700
-    },
-    {
-      "epoch": 6.32,
-      "learning_rate": 7.36936936936937e-05,
-      "loss": 0.2171,
-      "step": 701
-    },
-    {
-      "epoch": 6.32,
-      "learning_rate": 7.351351351351352e-05,
-      "loss": 0.2162,
-      "step": 702
-    },
-    {
-      "epoch": 6.33,
-      "learning_rate": 7.333333333333333e-05,
-      "loss": 0.2465,
-      "step": 703
-    },
-    {
-      "epoch": 6.34,
-      "learning_rate": 7.315315315315316e-05,
-      "loss": 0.2256,
-      "step": 704
-    },
-    {
-      "epoch": 6.35,
-      "learning_rate": 7.297297297297297e-05,
-      "loss": 0.2539,
-      "step": 705
-    },
-    {
-      "epoch": 6.36,
-      "learning_rate": 7.27927927927928e-05,
-      "loss": 0.2265,
-      "step": 706
-    },
-    {
-      "epoch": 6.37,
-      "learning_rate": 7.261261261261262e-05,
-      "loss": 0.266,
-      "step": 707
-    },
-    {
-      "epoch": 6.38,
-      "learning_rate": 7.243243243243245e-05,
-      "loss": 0.227,
-      "step": 708
-    },
-    {
-      "epoch": 6.39,
-      "learning_rate": 7.225225225225225e-05,
-      "loss": 0.2219,
-      "step": 709
-    },
-    {
-      "epoch": 6.4,
-      "learning_rate": 7.207207207207208e-05,
-      "loss": 0.2071,
-      "step": 710
-    },
-    {
-      "epoch": 6.41,
-      "learning_rate": 7.189189189189189e-05,
-      "loss": 0.247,
-      "step": 711
-    },
-    {
-      "epoch": 6.41,
-      "learning_rate": 7.171171171171172e-05,
-      "loss": 0.2306,
-      "step": 712
-    },
-    {
-      "epoch": 6.42,
-      "learning_rate": 7.153153153153153e-05,
-      "loss": 0.2531,
-      "step": 713
-    },
-    {
-      "epoch": 6.43,
-      "learning_rate": 7.135135135135136e-05,
-      "loss": 0.2534,
-      "step": 714
-    },
-    {
-      "epoch": 6.44,
-      "learning_rate": 7.117117117117116e-05,
-      "loss": 0.2102,
-      "step": 715
-    },
-    {
-      "epoch": 6.45,
-      "learning_rate": 7.099099099099099e-05,
-      "loss": 0.2476,
-      "step": 716
-    },
-    {
-      "epoch": 6.46,
-      "learning_rate": 7.081081081081081e-05,
-      "loss": 0.2229,
-      "step": 717
-    },
-    {
-      "epoch": 6.47,
-      "learning_rate": 7.063063063063064e-05,
-      "loss": 0.217,
-      "step": 718
-    },
-    {
-      "epoch": 6.48,
-      "learning_rate": 7.045045045045045e-05,
-      "loss": 0.2887,
-      "step": 719
-    },
-    {
-      "epoch": 6.49,
-      "learning_rate": 7.027027027027028e-05,
-      "loss": 0.2373,
-      "step": 720
-    },
-    {
-      "epoch": 6.5,
-      "learning_rate": 7.00900900900901e-05,
-      "loss": 0.2376,
-      "step": 721
-    },
-    {
-      "epoch": 6.5,
-      "learning_rate": 6.990990990990991e-05,
-      "loss": 0.2175,
-      "step": 722
-    },
-    {
-      "epoch": 6.51,
-      "learning_rate": 6.972972972972974e-05,
-      "loss": 0.2444,
-      "step": 723
-    },
-    {
-      "epoch": 6.52,
-      "learning_rate": 6.954954954954955e-05,
-      "loss": 0.2278,
-      "step": 724
-    },
-    {
-      "epoch": 6.53,
-      "learning_rate": 6.936936936936938e-05,
-      "loss": 0.2493,
-      "step": 725
-    },
-    {
-      "epoch": 6.54,
-      "learning_rate": 6.91891891891892e-05,
-      "loss": 0.2598,
-      "step": 726
-    },
-    {
-      "epoch": 6.55,
-      "learning_rate": 6.900900900900901e-05,
-      "loss": 0.2349,
-      "step": 727
-    },
-    {
-      "epoch": 6.56,
-      "learning_rate": 6.882882882882882e-05,
-      "loss": 0.1987,
-      "step": 728
-    },
-    {
-      "epoch": 6.57,
-      "learning_rate": 6.864864864864865e-05,
-      "loss": 0.2043,
-      "step": 729
-    },
-    {
-      "epoch": 6.58,
-      "learning_rate": 6.846846846846847e-05,
-      "loss": 0.2178,
-      "step": 730
-    },
-    {
-      "epoch": 6.59,
-      "learning_rate": 6.82882882882883e-05,
-      "loss": 0.2323,
-      "step": 731
-    },
-    {
-      "epoch": 6.59,
-      "learning_rate": 6.810810810810811e-05,
-      "loss": 0.1981,
-      "step": 732
-    },
-    {
-      "epoch": 6.6,
-      "learning_rate": 6.792792792792793e-05,
-      "loss": 0.2229,
-      "step": 733
-    },
-    {
-      "epoch": 6.61,
-      "learning_rate": 6.774774774774775e-05,
-      "loss": 0.2083,
-      "step": 734
-    },
-    {
-      "epoch": 6.62,
-      "learning_rate": 6.756756756756757e-05,
-      "loss": 0.2394,
-      "step": 735
-    },
-    {
-      "epoch": 6.63,
-      "learning_rate": 6.73873873873874e-05,
-      "loss": 0.2321,
-      "step": 736
-    },
-    {
-      "epoch": 6.64,
-      "learning_rate": 6.720720720720721e-05,
-      "loss": 0.2422,
-      "step": 737
-    },
-    {
-      "epoch": 6.65,
-      "learning_rate": 6.702702702702704e-05,
-      "loss": 0.2542,
-      "step": 738
-    },
-    {
-      "epoch": 6.66,
-      "learning_rate": 6.684684684684684e-05,
-      "loss": 0.2108,
-      "step": 739
-    },
-    {
-      "epoch": 6.67,
-      "learning_rate": 6.666666666666667e-05,
-      "loss": 0.2947,
-      "step": 740
-    },
-    {
-      "epoch": 6.68,
-      "learning_rate": 6.648648648648648e-05,
-      "loss": 0.2397,
-      "step": 741
-    },
-    {
-      "epoch": 6.68,
-      "learning_rate": 6.630630630630631e-05,
-      "loss": 0.2606,
-      "step": 742
-    },
-    {
-      "epoch": 6.69,
-      "learning_rate": 6.612612612612613e-05,
-      "loss": 0.2134,
-      "step": 743
-    },
-    {
-      "epoch": 6.7,
-      "learning_rate": 6.594594594594596e-05,
-      "loss": 0.2437,
-      "step": 744
-    },
-    {
-      "epoch": 6.71,
-      "learning_rate": 6.576576576576577e-05,
-      "loss": 0.2596,
-      "step": 745
-    },
-    {
-      "epoch": 6.72,
-      "learning_rate": 6.558558558558559e-05,
-      "loss": 0.3366,
-      "step": 746
-    },
-    {
-      "epoch": 6.73,
-      "learning_rate": 6.54054054054054e-05,
-      "loss": 0.2668,
-      "step": 747
-    },
-    {
-      "epoch": 6.74,
-      "learning_rate": 6.522522522522523e-05,
-      "loss": 0.2131,
-      "step": 748
-    },
-    {
-      "epoch": 6.75,
-      "learning_rate": 6.504504504504504e-05,
-      "loss": 0.2526,
-      "step": 749
-    },
-    {
-      "epoch": 6.76,
-      "learning_rate": 6.486486486486487e-05,
-      "loss": 0.2533,
-      "step": 750
-    },
-    {
-      "epoch": 6.77,
-      "learning_rate": 6.468468468468469e-05,
-      "loss": 0.2435,
-      "step": 751
-    },
-    {
-      "epoch": 6.77,
-      "learning_rate": 6.45045045045045e-05,
-      "loss": 0.2438,
-      "step": 752
-    },
-    {
-      "epoch": 6.78,
-      "learning_rate": 6.432432432432433e-05,
-      "loss": 0.2876,
-      "step": 753
-    },
-    {
-      "epoch": 6.79,
-      "learning_rate": 6.414414414414415e-05,
-      "loss": 0.2223,
-      "step": 754
-    },
-    {
-      "epoch": 6.8,
-      "learning_rate": 6.396396396396397e-05,
-      "loss": 0.2464,
-      "step": 755
-    },
-    {
-      "epoch": 6.81,
-      "learning_rate": 6.378378378378379e-05,
-      "loss": 0.2879,
-      "step": 756
-    },
-    {
-      "epoch": 6.82,
-      "learning_rate": 6.360360360360362e-05,
-      "loss": 0.2551,
-      "step": 757
-    },
-    {
-      "epoch": 6.83,
-      "learning_rate": 6.342342342342342e-05,
-      "loss": 0.2715,
-      "step": 758
-    },
-    {
-      "epoch": 6.84,
-      "learning_rate": 6.324324324324325e-05,
-      "loss": 0.2768,
-      "step": 759
-    },
-    {
-      "epoch": 6.85,
-      "learning_rate": 6.306306306306306e-05,
-      "loss": 0.2601,
-      "step": 760
-    },
-    {
-      "epoch": 6.86,
-      "learning_rate": 6.288288288288289e-05,
-      "loss": 0.2163,
-      "step": 761
-    },
-    {
-      "epoch": 6.86,
-      "learning_rate": 6.27027027027027e-05,
-      "loss": 0.1886,
-      "step": 762
-    },
-    {
-      "epoch": 6.87,
-      "learning_rate": 6.252252252252253e-05,
-      "loss": 0.2579,
-      "step": 763
-    },
-    {
-      "epoch": 6.88,
-      "learning_rate": 6.234234234234235e-05,
-      "loss": 0.2352,
-      "step": 764
-    },
-    {
-      "epoch": 6.89,
-      "learning_rate": 6.216216216216216e-05,
-      "loss": 0.2469,
-      "step": 765
-    },
-    {
-      "epoch": 6.9,
-      "learning_rate": 6.198198198198199e-05,
-      "loss": 0.2886,
-      "step": 766
-    },
-    {
-      "epoch": 6.91,
-      "learning_rate": 6.18018018018018e-05,
-      "loss": 0.2168,
-      "step": 767
-    },
-    {
-      "epoch": 6.92,
-      "learning_rate": 6.162162162162163e-05,
-      "loss": 0.2607,
-      "step": 768
-    },
-    {
-      "epoch": 6.93,
-      "learning_rate": 6.144144144144145e-05,
-      "loss": 0.253,
-      "step": 769
-    },
-    {
-      "epoch": 6.94,
-      "learning_rate": 6.126126126126126e-05,
-      "loss": 0.2079,
-      "step": 770
-    },
-    {
-      "epoch": 6.95,
-      "learning_rate": 6.108108108108108e-05,
-      "loss": 0.2232,
-      "step": 771
-    },
-    {
-      "epoch": 6.95,
-      "learning_rate": 6.09009009009009e-05,
-      "loss": 0.2432,
-      "step": 772
-    },
-    {
-      "epoch": 6.96,
-      "learning_rate": 6.072072072072072e-05,
-      "loss": 0.2366,
-      "step": 773
-    },
-    {
-      "epoch": 6.97,
-      "learning_rate": 6.0540540540540543e-05,
-      "loss": 0.2189,
-      "step": 774
-    },
-    {
-      "epoch": 6.98,
-      "learning_rate": 6.0360360360360365e-05,
-      "loss": 0.253,
-      "step": 775
-    },
-    {
-      "epoch": 6.99,
-      "learning_rate": 6.018018018018018e-05,
-      "loss": 0.2137,
-      "step": 776
-    },
-    {
-      "epoch": 7.0,
-      "learning_rate": 6e-05,
-      "loss": 0.2828,
-      "step": 777
-    },
-    {
-      "epoch": 7.01,
-      "learning_rate": 5.981981981981982e-05,
-      "loss": 0.1999,
-      "step": 778
-    },
-    {
-      "epoch": 7.02,
-      "learning_rate": 5.9639639639639645e-05,
-      "loss": 0.1985,
-      "step": 779
-    },
-    {
-      "epoch": 7.03,
-      "learning_rate": 5.9459459459459466e-05,
-      "loss": 0.183,
-      "step": 780
-    },
-    {
-      "epoch": 7.04,
-      "learning_rate": 5.927927927927929e-05,
-      "loss": 0.1728,
-      "step": 781
-    },
-    {
-      "epoch": 7.05,
-      "learning_rate": 5.9099099099099096e-05,
-      "loss": 0.184,
-      "step": 782
-    },
-    {
-      "epoch": 7.05,
-      "learning_rate": 5.891891891891892e-05,
-      "loss": 0.2041,
-      "step": 783
-    },
-    {
-      "epoch": 7.06,
-      "learning_rate": 5.873873873873874e-05,
-      "loss": 0.1711,
-      "step": 784
-    },
-    {
-      "epoch": 7.07,
-      "learning_rate": 5.855855855855856e-05,
-      "loss": 0.1968,
-      "step": 785
-    },
-    {
-      "epoch": 7.08,
-      "learning_rate": 5.837837837837838e-05,
-      "loss": 0.2036,
-      "step": 786
-    },
-    {
-      "epoch": 7.09,
-      "learning_rate": 5.8198198198198204e-05,
-      "loss": 0.2364,
-      "step": 787
-    },
-    {
-      "epoch": 7.1,
-      "learning_rate": 5.801801801801802e-05,
-      "loss": 0.2074,
-      "step": 788
-    },
-    {
-      "epoch": 7.11,
-      "learning_rate": 5.783783783783784e-05,
-      "loss": 0.2597,
-      "step": 789
-    },
-    {
-      "epoch": 7.12,
-      "learning_rate": 5.765765765765766e-05,
-      "loss": 0.1965,
-      "step": 790
-    },
-    {
-      "epoch": 7.13,
-      "learning_rate": 5.7477477477477484e-05,
-      "loss": 0.1815,
-      "step": 791
-    },
-    {
-      "epoch": 7.14,
-      "learning_rate": 5.7297297297297305e-05,
-      "loss": 0.1948,
-      "step": 792
-    },
-    {
-      "epoch": 7.14,
-      "learning_rate": 5.711711711711713e-05,
-      "loss": 0.2361,
-      "step": 793
-    },
-    {
-      "epoch": 7.15,
-      "learning_rate": 5.6936936936936935e-05,
-      "loss": 0.2092,
-      "step": 794
-    },
-    {
-      "epoch": 7.16,
-      "learning_rate": 5.6756756756756757e-05,
-      "loss": 0.2038,
-      "step": 795
-    },
-    {
-      "epoch": 7.17,
-      "learning_rate": 5.657657657657658e-05,
-      "loss": 0.1696,
-      "step": 796
-    },
-    {
-      "epoch": 7.18,
-      "learning_rate": 5.63963963963964e-05,
-      "loss": 0.1944,
-      "step": 797
-    },
-    {
-      "epoch": 7.19,
-      "learning_rate": 5.621621621621622e-05,
-      "loss": 0.2115,
-      "step": 798
-    },
-    {
-      "epoch": 7.2,
-      "learning_rate": 5.603603603603604e-05,
-      "loss": 0.1758,
-      "step": 799
-    },
-    {
-      "epoch": 7.21,
-      "learning_rate": 5.585585585585585e-05,
-      "loss": 0.1908,
-      "step": 800
-    },
-    {
-      "epoch": 7.22,
-      "learning_rate": 5.567567567567567e-05,
-      "loss": 0.2106,
-      "step": 801
-    },
-    {
-      "epoch": 7.23,
-      "learning_rate": 5.5495495495495494e-05,
-      "loss": 0.2048,
-      "step": 802
-    },
-    {
-      "epoch": 7.23,
-      "learning_rate": 5.5315315315315316e-05,
-      "loss": 0.192,
-      "step": 803
-    },
-    {
-      "epoch": 7.24,
-      "learning_rate": 5.5135135135135144e-05,
-      "loss": 0.1656,
-      "step": 804
-    },
-    {
-      "epoch": 7.25,
-      "learning_rate": 5.4954954954954966e-05,
-      "loss": 0.2072,
-      "step": 805
-    },
-    {
-      "epoch": 7.26,
-      "learning_rate": 5.4774774774774774e-05,
-      "loss": 0.1902,
-      "step": 806
-    },
-    {
-      "epoch": 7.27,
-      "learning_rate": 5.4594594594594595e-05,
-      "loss": 0.2155,
-      "step": 807
-    },
-    {
-      "epoch": 7.28,
-      "learning_rate": 5.441441441441442e-05,
-      "loss": 0.1983,
-      "step": 808
-    },
-    {
-      "epoch": 7.29,
-      "learning_rate": 5.423423423423424e-05,
-      "loss": 0.19,
-      "step": 809
-    },
-    {
-      "epoch": 7.3,
-      "learning_rate": 5.405405405405406e-05,
-      "loss": 0.2128,
-      "step": 810
-    },
-    {
-      "epoch": 7.31,
-      "learning_rate": 5.387387387387388e-05,
-      "loss": 0.1976,
-      "step": 811
-    },
-    {
-      "epoch": 7.32,
-      "learning_rate": 5.369369369369369e-05,
-      "loss": 0.228,
-      "step": 812
-    },
-    {
-      "epoch": 7.32,
-      "learning_rate": 5.351351351351351e-05,
-      "loss": 0.2673,
-      "step": 813
-    },
-    {
-      "epoch": 7.33,
-      "learning_rate": 5.333333333333333e-05,
-      "loss": 0.218,
-      "step": 814
-    },
-    {
-      "epoch": 7.34,
-      "learning_rate": 5.3153153153153155e-05,
-      "loss": 0.1898,
-      "step": 815
-    },
-    {
-      "epoch": 7.35,
-      "learning_rate": 5.2972972972972976e-05,
-      "loss": 0.1921,
-      "step": 816
-    },
-    {
-      "epoch": 7.36,
-      "learning_rate": 5.27927927927928e-05,
-      "loss": 0.1932,
-      "step": 817
-    },
-    {
-      "epoch": 7.37,
-      "learning_rate": 5.261261261261261e-05,
-      "loss": 0.2193,
-      "step": 818
-    },
-    {
-      "epoch": 7.38,
-      "learning_rate": 5.2432432432432434e-05,
-      "loss": 0.1821,
-      "step": 819
-    },
-    {
-      "epoch": 7.39,
-      "learning_rate": 5.2252252252252256e-05,
-      "loss": 0.2112,
-      "step": 820
-    },
-    {
-      "epoch": 7.4,
-      "learning_rate": 5.207207207207208e-05,
-      "loss": 0.2047,
-      "step": 821
-    },
-    {
-      "epoch": 7.41,
-      "learning_rate": 5.18918918918919e-05,
-      "loss": 0.1936,
-      "step": 822
-    },
-    {
-      "epoch": 7.41,
-      "learning_rate": 5.171171171171172e-05,
-      "loss": 0.2419,
-      "step": 823
-    },
-    {
-      "epoch": 7.42,
-      "learning_rate": 5.153153153153153e-05,
-      "loss": 0.1825,
-      "step": 824
-    },
-    {
-      "epoch": 7.43,
-      "learning_rate": 5.135135135135135e-05,
-      "loss": 0.2011,
-      "step": 825
-    },
-    {
-      "epoch": 7.44,
-      "learning_rate": 5.117117117117117e-05,
-      "loss": 0.2126,
-      "step": 826
-    },
-    {
-      "epoch": 7.45,
-      "learning_rate": 5.0990990990990994e-05,
-      "loss": 0.2394,
-      "step": 827
-    },
-    {
-      "epoch": 7.46,
-      "learning_rate": 5.0810810810810815e-05,
-      "loss": 0.2179,
-      "step": 828
-    },
-    {
-      "epoch": 7.47,
-      "learning_rate": 5.063063063063064e-05,
-      "loss": 0.232,
-      "step": 829
-    },
-    {
-      "epoch": 7.48,
-      "learning_rate": 5.0450450450450445e-05,
-      "loss": 0.2246,
-      "step": 830
-    },
-    {
-      "epoch": 7.49,
-      "learning_rate": 5.0270270270270267e-05,
-      "loss": 0.1931,
-      "step": 831
-    },
-    {
-      "epoch": 7.5,
-      "learning_rate": 5.009009009009009e-05,
-      "loss": 0.1878,
-      "step": 832
-    },
-    {
-      "epoch": 7.5,
-      "learning_rate": 4.9909909909909917e-05,
-      "loss": 0.2133,
-      "step": 833
-    },
-    {
-      "epoch": 7.51,
-      "learning_rate": 4.972972972972974e-05,
-      "loss": 0.1934,
-      "step": 834
-    },
-    {
-      "epoch": 7.52,
-      "learning_rate": 4.954954954954955e-05,
-      "loss": 0.2257,
-      "step": 835
-    },
-    {
-      "epoch": 7.53,
-      "learning_rate": 4.9369369369369375e-05,
-      "loss": 0.1962,
-      "step": 836
-    },
-    {
-      "epoch": 7.54,
-      "learning_rate": 4.9189189189189196e-05,
-      "loss": 0.1722,
-      "step": 837
-    },
-    {
-      "epoch": 7.55,
-      "learning_rate": 4.900900900900901e-05,
-      "loss": 0.1857,
-      "step": 838
-    },
-    {
-      "epoch": 7.56,
-      "learning_rate": 4.882882882882883e-05,
-      "loss": 0.2043,
-      "step": 839
-    },
-    {
-      "epoch": 7.57,
-      "learning_rate": 4.8648648648648654e-05,
-      "loss": 0.1977,
-      "step": 840
-    },
-    {
-      "epoch": 7.58,
-      "learning_rate": 4.846846846846847e-05,
-      "loss": 0.2503,
-      "step": 841
-    },
-    {
-      "epoch": 7.59,
-      "learning_rate": 4.828828828828829e-05,
-      "loss": 0.2246,
-      "step": 842
-    },
-    {
-      "epoch": 7.59,
-      "learning_rate": 4.810810810810811e-05,
-      "loss": 0.2141,
-      "step": 843
-    },
-    {
-      "epoch": 7.6,
-      "learning_rate": 4.792792792792793e-05,
-      "loss": 0.2114,
-      "step": 844
-    },
-    {
-      "epoch": 7.61,
-      "learning_rate": 4.774774774774775e-05,
-      "loss": 0.207,
-      "step": 845
-    },
-    {
-      "epoch": 7.62,
-      "learning_rate": 4.756756756756757e-05,
-      "loss": 0.2045,
-      "step": 846
-    },
-    {
-      "epoch": 7.63,
-      "learning_rate": 4.738738738738739e-05,
-      "loss": 0.1935,
-      "step": 847
-    },
-    {
-      "epoch": 7.64,
-      "learning_rate": 4.7207207207207214e-05,
-      "loss": 0.2299,
-      "step": 848
-    },
-    {
-      "epoch": 7.65,
-      "learning_rate": 4.7027027027027035e-05,
-      "loss": 0.2157,
-      "step": 849
-    },
-    {
-      "epoch": 7.66,
-      "learning_rate": 4.684684684684685e-05,
-      "loss": 0.2246,
-      "step": 850
-    },
-    {
-      "epoch": 7.67,
-      "learning_rate": 4.666666666666667e-05,
-      "loss": 0.1902,
-      "step": 851
-    },
-    {
-      "epoch": 7.68,
-      "learning_rate": 4.648648648648649e-05,
-      "loss": 0.2111,
-      "step": 852
-    },
-    {
-      "epoch": 7.68,
-      "learning_rate": 4.630630630630631e-05,
-      "loss": 0.198,
-      "step": 853
-    },
-    {
-      "epoch": 7.69,
-      "learning_rate": 4.612612612612613e-05,
-      "loss": 0.2126,
-      "step": 854
-    },
-    {
-      "epoch": 7.7,
-      "learning_rate": 4.594594594594595e-05,
-      "loss": 0.2093,
-      "step": 855
-    },
-    {
-      "epoch": 7.71,
-      "learning_rate": 4.5765765765765766e-05,
-      "loss": 0.2755,
-      "step": 856
-    },
-    {
-      "epoch": 7.72,
-      "learning_rate": 4.558558558558559e-05,
-      "loss": 0.2028,
-      "step": 857
-    },
-    {
-      "epoch": 7.73,
-      "learning_rate": 4.540540540540541e-05,
-      "loss": 0.2026,
-      "step": 858
-    },
-    {
-      "epoch": 7.74,
-      "learning_rate": 4.5225225225225224e-05,
-      "loss": 0.1991,
-      "step": 859
-    },
-    {
-      "epoch": 7.75,
-      "learning_rate": 4.5045045045045046e-05,
-      "loss": 0.2005,
-      "step": 860
-    },
-    {
-      "epoch": 7.76,
-      "learning_rate": 4.486486486486487e-05,
-      "loss": 0.1966,
-      "step": 861
-    },
-    {
-      "epoch": 7.77,
-      "learning_rate": 4.468468468468469e-05,
-      "loss": 0.2325,
-      "step": 862
-    },
-    {
-      "epoch": 7.77,
-      "learning_rate": 4.450450450450451e-05,
-      "loss": 0.1986,
-      "step": 863
-    },
-    {
-      "epoch": 7.78,
-      "learning_rate": 4.432432432432433e-05,
-      "loss": 0.2288,
-      "step": 864
-    },
-    {
-      "epoch": 7.79,
-      "learning_rate": 4.414414414414415e-05,
-      "loss": 0.1883,
-      "step": 865
-    },
-    {
-      "epoch": 7.8,
-      "learning_rate": 4.396396396396397e-05,
-      "loss": 0.2734,
-      "step": 866
-    },
-    {
-      "epoch": 7.81,
-      "learning_rate": 4.378378378378379e-05,
-      "loss": 0.2244,
-      "step": 867
-    },
-    {
-      "epoch": 7.82,
-      "learning_rate": 4.3603603603603605e-05,
-      "loss": 0.2134,
-      "step": 868
-    },
-    {
-      "epoch": 7.83,
-      "learning_rate": 4.342342342342343e-05,
-      "loss": 0.2495,
-      "step": 869
-    },
-    {
-      "epoch": 7.84,
-      "learning_rate": 4.324324324324325e-05,
-      "loss": 0.2268,
-      "step": 870
-    },
-    {
-      "epoch": 7.85,
-      "learning_rate": 4.306306306306306e-05,
-      "loss": 0.2232,
-      "step": 871
-    },
-    {
-      "epoch": 7.86,
-      "learning_rate": 4.2882882882882885e-05,
-      "loss": 0.2142,
-      "step": 872
-    },
-    {
-      "epoch": 7.86,
-      "learning_rate": 4.2702702702702706e-05,
-      "loss": 0.1927,
-      "step": 873
-    },
-    {
-      "epoch": 7.87,
-      "learning_rate": 4.252252252252252e-05,
-      "loss": 0.2356,
-      "step": 874
-    },
-    {
-      "epoch": 7.88,
-      "learning_rate": 4.234234234234234e-05,
-      "loss": 0.1987,
-      "step": 875
-    },
-    {
-      "epoch": 7.89,
-      "learning_rate": 4.2162162162162164e-05,
-      "loss": 0.203,
-      "step": 876
-    },
-    {
-      "epoch": 7.9,
-      "learning_rate": 4.1981981981981986e-05,
-      "loss": 0.2034,
-      "step": 877
-    },
-    {
-      "epoch": 7.91,
-      "learning_rate": 4.180180180180181e-05,
-      "loss": 0.221,
-      "step": 878
-    },
-    {
-      "epoch": 7.92,
-      "learning_rate": 4.162162162162163e-05,
-      "loss": 0.1952,
-      "step": 879
-    },
-    {
-      "epoch": 7.93,
-      "learning_rate": 4.1441441441441444e-05,
-      "loss": 0.1798,
-      "step": 880
-    },
-    {
-      "epoch": 7.94,
-      "learning_rate": 4.1261261261261266e-05,
-      "loss": 0.2057,
-      "step": 881
-    },
-    {
-      "epoch": 7.95,
-      "learning_rate": 4.108108108108109e-05,
-      "loss": 0.2084,
-      "step": 882
-    },
-    {
-      "epoch": 7.95,
-      "learning_rate": 4.09009009009009e-05,
-      "loss": 0.2067,
-      "step": 883
-    },
-    {
-      "epoch": 7.96,
-      "learning_rate": 4.0720720720720724e-05,
-      "loss": 0.2017,
-      "step": 884
-    },
-    {
-      "epoch": 7.97,
-      "learning_rate": 4.0540540540540545e-05,
-      "loss": 0.1786,
-      "step": 885
-    },
-    {
-      "epoch": 7.98,
-      "learning_rate": 4.036036036036036e-05,
-      "loss": 0.2265,
-      "step": 886
-    },
-    {
-      "epoch": 7.99,
-      "learning_rate": 4.018018018018018e-05,
-      "loss": 0.2221,
-      "step": 887
-    },
-    {
-      "epoch": 8.0,
-      "learning_rate": 4e-05,
-      "loss": 0.1592,
-      "step": 888
-    },
-    {
-      "epoch": 8.01,
-      "learning_rate": 3.981981981981982e-05,
-      "loss": 0.1533,
-      "step": 889
-    },
-    {
-      "epoch": 8.02,
-      "learning_rate": 3.963963963963964e-05,
-      "loss": 0.201,
-      "step": 890
-    },
-    {
-      "epoch": 8.03,
-      "learning_rate": 3.945945945945946e-05,
-      "loss": 0.2034,
-      "step": 891
-    },
-    {
-      "epoch": 8.04,
-      "learning_rate": 3.927927927927928e-05,
-      "loss": 0.1815,
-      "step": 892
-    },
-    {
-      "epoch": 8.05,
-      "learning_rate": 3.9099099099099105e-05,
-      "loss": 0.1729,
-      "step": 893
-    },
-    {
-      "epoch": 8.05,
-      "learning_rate": 3.8918918918918926e-05,
-      "loss": 0.203,
-      "step": 894
-    },
-    {
-      "epoch": 8.06,
-      "learning_rate": 3.873873873873874e-05,
-      "loss": 0.1611,
-      "step": 895
-    },
-    {
-      "epoch": 8.07,
-      "learning_rate": 3.855855855855856e-05,
-      "loss": 0.2099,
-      "step": 896
-    },
-    {
-      "epoch": 8.08,
-      "learning_rate": 3.8378378378378384e-05,
-      "loss": 0.1435,
-      "step": 897
-    },
-    {
-      "epoch": 8.09,
-      "learning_rate": 3.81981981981982e-05,
-      "loss": 0.1857,
-      "step": 898
-    },
-    {
-      "epoch": 8.1,
-      "learning_rate": 3.801801801801802e-05,
-      "loss": 0.2327,
-      "step": 899
-    },
-    {
-      "epoch": 8.11,
-      "learning_rate": 3.783783783783784e-05,
-      "loss": 0.1875,
-      "step": 900
-    },
-    {
-      "epoch": 8.12,
-      "learning_rate": 3.765765765765766e-05,
-      "loss": 0.1824,
-      "step": 901
-    },
-    {
-      "epoch": 8.13,
-      "learning_rate": 3.747747747747748e-05,
-      "loss": 0.1671,
-      "step": 902
-    },
-    {
-      "epoch": 8.14,
-      "learning_rate": 3.72972972972973e-05,
-      "loss": 0.157,
-      "step": 903
-    },
-    {
-      "epoch": 8.14,
-      "learning_rate": 3.7117117117117115e-05,
-      "loss": 0.2169,
-      "step": 904
-    },
-    {
-      "epoch": 8.15,
-      "learning_rate": 3.693693693693694e-05,
-      "loss": 0.1918,
-      "step": 905
-    },
-    {
-      "epoch": 8.16,
-      "learning_rate": 3.675675675675676e-05,
-      "loss": 0.2088,
-      "step": 906
-    },
-    {
-      "epoch": 8.17,
-      "learning_rate": 3.657657657657658e-05,
-      "loss": 0.1751,
-      "step": 907
-    },
-    {
-      "epoch": 8.18,
-      "learning_rate": 3.63963963963964e-05,
-      "loss": 0.2023,
-      "step": 908
-    },
-    {
-      "epoch": 8.19,
-      "learning_rate": 3.621621621621622e-05,
-      "loss": 0.1897,
-      "step": 909
-    },
-    {
-      "epoch": 8.2,
-      "learning_rate": 3.603603603603604e-05,
-      "loss": 0.1792,
-      "step": 910
-    },
-    {
-      "epoch": 8.21,
-      "learning_rate": 3.585585585585586e-05,
-      "loss": 0.177,
-      "step": 911
-    },
-    {
-      "epoch": 8.22,
-      "learning_rate": 3.567567567567568e-05,
-      "loss": 0.1938,
-      "step": 912
-    },
-    {
-      "epoch": 8.23,
-      "learning_rate": 3.5495495495495496e-05,
-      "loss": 0.1987,
-      "step": 913
-    },
-    {
-      "epoch": 8.23,
-      "learning_rate": 3.531531531531532e-05,
-      "loss": 0.1757,
-      "step": 914
-    },
-    {
-      "epoch": 8.24,
-      "learning_rate": 3.513513513513514e-05,
-      "loss": 0.2028,
-      "step": 915
-    },
-    {
-      "epoch": 8.25,
-      "learning_rate": 3.4954954954954954e-05,
-      "loss": 0.1704,
-      "step": 916
-    },
-    {
-      "epoch": 8.26,
-      "learning_rate": 3.4774774774774776e-05,
-      "loss": 0.1984,
-      "step": 917
-    },
-    {
-      "epoch": 8.27,
-      "learning_rate": 3.45945945945946e-05,
-      "loss": 0.2117,
-      "step": 918
-    },
-    {
-      "epoch": 8.28,
-      "learning_rate": 3.441441441441441e-05,
-      "loss": 0.1817,
-      "step": 919
-    },
-    {
-      "epoch": 8.29,
-      "learning_rate": 3.4234234234234234e-05,
-      "loss": 0.2177,
-      "step": 920
-    },
-    {
-      "epoch": 8.3,
-      "learning_rate": 3.4054054054054055e-05,
-      "loss": 0.1737,
-      "step": 921
-    },
-    {
-      "epoch": 8.31,
-      "learning_rate": 3.387387387387388e-05,
-      "loss": 0.1605,
-      "step": 922
-    },
-    {
-      "epoch": 8.32,
-      "learning_rate": 3.36936936936937e-05,
-      "loss": 0.1568,
-      "step": 923
-    },
-    {
-      "epoch": 8.32,
-      "learning_rate": 3.351351351351352e-05,
-      "loss": 0.1898,
-      "step": 924
-    },
-    {
-      "epoch": 8.33,
-      "learning_rate": 3.3333333333333335e-05,
-      "loss": 0.1615,
-      "step": 925
-    },
-    {
-      "epoch": 8.34,
-      "learning_rate": 3.3153153153153157e-05,
-      "loss": 0.1867,
-      "step": 926
-    },
-    {
-      "epoch": 8.35,
-      "learning_rate": 3.297297297297298e-05,
-      "loss": 0.1879,
-      "step": 927
-    },
-    {
-      "epoch": 8.36,
-      "learning_rate": 3.279279279279279e-05,
-      "loss": 0.1822,
-      "step": 928
-    },
-    {
-      "epoch": 8.37,
-      "learning_rate": 3.2612612612612615e-05,
-      "loss": 0.1782,
-      "step": 929
-    },
-    {
-      "epoch": 8.38,
-      "learning_rate": 3.2432432432432436e-05,
-      "loss": 0.1718,
-      "step": 930
-    },
-    {
-      "epoch": 8.39,
-      "learning_rate": 3.225225225225225e-05,
-      "loss": 0.1944,
-      "step": 931
-    },
-    {
-      "epoch": 8.4,
-      "learning_rate": 3.207207207207207e-05,
-      "loss": 0.2228,
-      "step": 932
-    },
-    {
-      "epoch": 8.41,
-      "learning_rate": 3.1891891891891894e-05,
-      "loss": 0.1929,
-      "step": 933
-    },
-    {
-      "epoch": 8.41,
-      "learning_rate": 3.171171171171171e-05,
-      "loss": 0.1967,
-      "step": 934
-    },
-    {
-      "epoch": 8.42,
-      "learning_rate": 3.153153153153153e-05,
-      "loss": 0.1918,
-      "step": 935
-    },
-    {
-      "epoch": 8.43,
-      "learning_rate": 3.135135135135135e-05,
-      "loss": 0.1658,
-      "step": 936
-    },
-    {
-      "epoch": 8.44,
-      "learning_rate": 3.1171171171171174e-05,
-      "loss": 0.1878,
-      "step": 937
-    },
-    {
-      "epoch": 8.45,
-      "learning_rate": 3.0990990990990995e-05,
-      "loss": 0.1712,
-      "step": 938
-    },
-    {
-      "epoch": 8.46,
-      "learning_rate": 3.081081081081082e-05,
-      "loss": 0.1775,
-      "step": 939
-    },
-    {
-      "epoch": 8.47,
-      "learning_rate": 3.063063063063063e-05,
-      "loss": 0.1869,
-      "step": 940
-    },
-    {
-      "epoch": 8.48,
-      "learning_rate": 3.045045045045045e-05,
-      "loss": 0.1943,
-      "step": 941
-    },
-    {
-      "epoch": 8.49,
-      "learning_rate": 3.0270270270270272e-05,
-      "loss": 0.1863,
-      "step": 942
-    },
-    {
-      "epoch": 8.5,
-      "learning_rate": 3.009009009009009e-05,
-      "loss": 0.1746,
-      "step": 943
-    },
-    {
-      "epoch": 8.5,
-      "learning_rate": 2.990990990990991e-05,
-      "loss": 0.175,
-      "step": 944
-    },
-    {
-      "epoch": 8.51,
-      "learning_rate": 2.9729729729729733e-05,
-      "loss": 0.2085,
-      "step": 945
-    },
-    {
-      "epoch": 8.52,
-      "learning_rate": 2.9549549549549548e-05,
-      "loss": 0.1638,
-      "step": 946
-    },
-    {
-      "epoch": 8.53,
-      "learning_rate": 2.936936936936937e-05,
-      "loss": 0.2147,
-      "step": 947
-    },
-    {
-      "epoch": 8.54,
-      "learning_rate": 2.918918918918919e-05,
-      "loss": 0.1774,
-      "step": 948
-    },
-    {
-      "epoch": 8.55,
-      "learning_rate": 2.900900900900901e-05,
-      "loss": 0.179,
-      "step": 949
-    },
-    {
-      "epoch": 8.56,
-      "learning_rate": 2.882882882882883e-05,
-      "loss": 0.1927,
-      "step": 950
-    },
-    {
-      "epoch": 8.57,
-      "learning_rate": 2.8648648648648653e-05,
-      "loss": 0.1809,
-      "step": 951
-    },
-    {
-      "epoch": 8.58,
-      "learning_rate": 2.8468468468468467e-05,
-      "loss": 0.168,
-      "step": 952
-    },
-    {
-      "epoch": 8.59,
-      "learning_rate": 2.828828828828829e-05,
-      "loss": 0.1853,
-      "step": 953
-    },
-    {
-      "epoch": 8.59,
-      "learning_rate": 2.810810810810811e-05,
-      "loss": 0.2057,
-      "step": 954
-    },
-    {
-      "epoch": 8.6,
-      "learning_rate": 2.7927927927927926e-05,
-      "loss": 0.1757,
-      "step": 955
-    },
-    {
-      "epoch": 8.61,
-      "learning_rate": 2.7747747747747747e-05,
-      "loss": 0.1706,
-      "step": 956
-    },
-    {
-      "epoch": 8.62,
-      "learning_rate": 2.7567567567567572e-05,
-      "loss": 0.1854,
-      "step": 957
-    },
-    {
-      "epoch": 8.63,
-      "learning_rate": 2.7387387387387387e-05,
-      "loss": 0.1873,
-      "step": 958
-    },
-    {
-      "epoch": 8.64,
-      "learning_rate": 2.720720720720721e-05,
-      "loss": 0.1783,
-      "step": 959
-    },
-    {
-      "epoch": 8.65,
-      "learning_rate": 2.702702702702703e-05,
-      "loss": 0.1709,
-      "step": 960
-    },
-    {
-      "epoch": 8.66,
-      "learning_rate": 2.6846846846846845e-05,
-      "loss": 0.1766,
-      "step": 961
-    },
-    {
-      "epoch": 8.67,
-      "learning_rate": 2.6666666666666667e-05,
-      "loss": 0.21,
-      "step": 962
-    },
-    {
-      "epoch": 8.68,
-      "learning_rate": 2.6486486486486488e-05,
-      "loss": 0.196,
-      "step": 963
-    },
-    {
-      "epoch": 8.68,
-      "learning_rate": 2.6306306306306306e-05,
-      "loss": 0.2047,
-      "step": 964
-    },
-    {
-      "epoch": 8.69,
-      "learning_rate": 2.6126126126126128e-05,
-      "loss": 0.1878,
-      "step": 965
-    },
-    {
-      "epoch": 8.7,
-      "learning_rate": 2.594594594594595e-05,
-      "loss": 0.1713,
-      "step": 966
-    },
-    {
-      "epoch": 8.71,
-      "learning_rate": 2.5765765765765764e-05,
-      "loss": 0.169,
-      "step": 967
-    },
-    {
-      "epoch": 8.72,
-      "learning_rate": 2.5585585585585586e-05,
-      "loss": 0.1749,
-      "step": 968
-    },
-    {
-      "epoch": 8.73,
-      "learning_rate": 2.5405405405405408e-05,
-      "loss": 0.1968,
-      "step": 969
-    },
-    {
-      "epoch": 8.74,
-      "learning_rate": 2.5225225225225222e-05,
-      "loss": 0.173,
-      "step": 970
-    },
-    {
-      "epoch": 8.75,
-      "learning_rate": 2.5045045045045044e-05,
-      "loss": 0.208,
-      "step": 971
-    },
-    {
-      "epoch": 8.76,
-      "learning_rate": 2.486486486486487e-05,
-      "loss": 0.1588,
-      "step": 972
-    },
-    {
-      "epoch": 8.77,
-      "learning_rate": 2.4684684684684687e-05,
-      "loss": 0.1704,
-      "step": 973
-    },
-    {
-      "epoch": 8.77,
-      "learning_rate": 2.4504504504504506e-05,
-      "loss": 0.1907,
-      "step": 974
-    },
-    {
-      "epoch": 8.78,
-      "learning_rate": 2.4324324324324327e-05,
-      "loss": 0.1939,
-      "step": 975
-    },
-    {
-      "epoch": 8.79,
-      "learning_rate": 2.4144144144144145e-05,
-      "loss": 0.1676,
-      "step": 976
-    },
-    {
-      "epoch": 8.8,
-      "learning_rate": 2.3963963963963964e-05,
-      "loss": 0.2301,
-      "step": 977
-    },
-    {
-      "epoch": 8.81,
-      "learning_rate": 2.3783783783783785e-05,
-      "loss": 0.1836,
-      "step": 978
-    },
-    {
-      "epoch": 8.82,
-      "learning_rate": 2.3603603603603607e-05,
-      "loss": 0.1972,
-      "step": 979
-    },
-    {
-      "epoch": 8.83,
-      "learning_rate": 2.3423423423423425e-05,
-      "loss": 0.1865,
-      "step": 980
-    },
-    {
-      "epoch": 8.84,
-      "learning_rate": 2.3243243243243247e-05,
-      "loss": 0.168,
-      "step": 981
-    },
-    {
-      "epoch": 8.85,
-      "learning_rate": 2.3063063063063065e-05,
-      "loss": 0.1778,
-      "step": 982
-    },
-    {
-      "epoch": 8.86,
-      "learning_rate": 2.2882882882882883e-05,
-      "loss": 0.1863,
-      "step": 983
-    },
-    {
-      "epoch": 8.86,
-      "learning_rate": 2.2702702702702705e-05,
-      "loss": 0.1549,
-      "step": 984
-    },
-    {
-      "epoch": 8.87,
-      "learning_rate": 2.2522522522522523e-05,
-      "loss": 0.1869,
-      "step": 985
-    },
-    {
-      "epoch": 8.88,
-      "learning_rate": 2.2342342342342344e-05,
-      "loss": 0.2235,
-      "step": 986
-    },
-    {
-      "epoch": 8.89,
-      "learning_rate": 2.2162162162162166e-05,
-      "loss": 0.172,
-      "step": 987
-    },
-    {
-      "epoch": 8.9,
-      "learning_rate": 2.1981981981981984e-05,
-      "loss": 0.2134,
-      "step": 988
-    },
-    {
-      "epoch": 8.91,
-      "learning_rate": 2.1801801801801803e-05,
-      "loss": 0.1754,
-      "step": 989
-    },
-    {
-      "epoch": 8.92,
-      "learning_rate": 2.1621621621621624e-05,
-      "loss": 0.1436,
-      "step": 990
-    },
-    {
-      "epoch": 8.93,
-      "learning_rate": 2.1441441441441442e-05,
-      "loss": 0.1933,
-      "step": 991
-    },
-    {
-      "epoch": 8.94,
-      "learning_rate": 2.126126126126126e-05,
-      "loss": 0.2173,
-      "step": 992
-    },
-    {
-      "epoch": 8.95,
-      "learning_rate": 2.1081081081081082e-05,
-      "loss": 0.1751,
-      "step": 993
-    },
-    {
-      "epoch": 8.95,
-      "learning_rate": 2.0900900900900904e-05,
-      "loss": 0.177,
-      "step": 994
-    },
-    {
-      "epoch": 8.96,
-      "learning_rate": 2.0720720720720722e-05,
-      "loss": 0.1967,
-      "step": 995
-    },
-    {
-      "epoch": 8.97,
-      "learning_rate": 2.0540540540540544e-05,
-      "loss": 0.1582,
-      "step": 996
-    },
-    {
-      "epoch": 8.98,
-      "learning_rate": 2.0360360360360362e-05,
-      "loss": 0.1755,
-      "step": 997
-    },
-    {
-      "epoch": 8.99,
-      "learning_rate": 2.018018018018018e-05,
-      "loss": 0.1799,
-      "step": 998
-    },
-    {
-      "epoch": 9.0,
-      "learning_rate": 2e-05,
-      "loss": 0.2062,
-      "step": 999
-    },
-    {
-      "epoch": 9.01,
-      "learning_rate": 1.981981981981982e-05,
-      "loss": 0.1766,
-      "step": 1000
-    },
-    {
-      "epoch": 9.02,
-      "learning_rate": 1.963963963963964e-05,
-      "loss": 0.1705,
-      "step": 1001
-    },
-    {
-      "epoch": 9.03,
-      "learning_rate": 1.9459459459459463e-05,
-      "loss": 0.1559,
-      "step": 1002
-    },
-    {
-      "epoch": 9.04,
-      "learning_rate": 1.927927927927928e-05,
-      "loss": 0.1595,
-      "step": 1003
-    },
-    {
-      "epoch": 9.05,
-      "learning_rate": 1.90990990990991e-05,
-      "loss": 0.1711,
-      "step": 1004
-    },
-    {
-      "epoch": 9.05,
-      "learning_rate": 1.891891891891892e-05,
-      "loss": 0.1782,
-      "step": 1005
-    },
-    {
-      "epoch": 9.06,
-      "learning_rate": 1.873873873873874e-05,
-      "loss": 0.1841,
-      "step": 1006
-    },
-    {
-      "epoch": 9.07,
-      "learning_rate": 1.8558558558558558e-05,
-      "loss": 0.1725,
-      "step": 1007
-    },
-    {
-      "epoch": 9.08,
-      "learning_rate": 1.837837837837838e-05,
-      "loss": 0.1974,
-      "step": 1008
-    },
     {
       "epoch": 9.09,
-      "learning_rate": 1.81981981981982e-05,
-      "loss": 0.1765,
-      "step": 1009
-    },
-    {
-      "epoch": 9.1,
-      "learning_rate": 1.801801801801802e-05,
-      "loss": 0.1505,
-      "step": 1010
+      "learning_rate": 4.5283018867924524e-06,
+      "loss": 0.6673,
+      "step": 482
     },
     {
       "epoch": 9.11,
-      "learning_rate": 1.783783783783784e-05,
-      "loss": 0.1642,
-      "step": 1011
-    },
-    {
-      "epoch": 9.12,
-      "learning_rate": 1.765765765765766e-05,
-      "loss": 0.1547,
-      "step": 1012
+      "learning_rate": 4.4339622641509435e-06,
+      "loss": 0.5573,
+      "step": 483
     },
     {
       "epoch": 9.13,
-      "learning_rate": 1.7477477477477477e-05,
-      "loss": 0.163,
-      "step": 1013
-    },
-    {
-      "epoch": 9.14,
-      "learning_rate": 1.72972972972973e-05,
-      "loss": 0.1567,
-      "step": 1014
-    },
-    {
-      "epoch": 9.14,
-      "learning_rate": 1.7117117117117117e-05,
-      "loss": 0.1563,
-      "step": 1015
+      "learning_rate": 4.339622641509434e-06,
+      "loss": 0.5971,
+      "step": 484
     },
     {
       "epoch": 9.15,
-      "learning_rate": 1.693693693693694e-05,
-      "loss": 0.1349,
-      "step": 1016
-    },
-    {
-      "epoch": 9.16,
-      "learning_rate": 1.675675675675676e-05,
-      "loss": 0.1765,
-      "step": 1017
+      "learning_rate": 4.245283018867925e-06,
+      "loss": 0.5829,
+      "step": 485
     },
     {
       "epoch": 9.17,
-      "learning_rate": 1.6576576576576578e-05,
-      "loss": 0.1638,
-      "step": 1018
-    },
-    {
-      "epoch": 9.18,
-      "learning_rate": 1.6396396396396396e-05,
-      "loss": 0.16,
-      "step": 1019
+      "learning_rate": 4.150943396226415e-06,
+      "loss": 0.5953,
+      "step": 486
     },
     {
       "epoch": 9.19,
-      "learning_rate": 1.6216216216216218e-05,
-      "loss": 0.2045,
-      "step": 1020
-    },
-    {
-      "epoch": 9.2,
-      "learning_rate": 1.6036036036036036e-05,
-      "loss": 0.1729,
-      "step": 1021
+      "learning_rate": 4.056603773584905e-06,
+      "loss": 0.6096,
+      "step": 487
     },
     {
       "epoch": 9.21,
-      "learning_rate": 1.5855855855855855e-05,
-      "loss": 0.1687,
-      "step": 1022
-    },
-    {
-      "epoch": 9.22,
-      "learning_rate": 1.5675675675675676e-05,
-      "loss": 0.1639,
-      "step": 1023
-    },
-    {
-      "epoch": 9.23,
-      "learning_rate": 1.5495495495495498e-05,
-      "loss": 0.1709,
-      "step": 1024
+      "learning_rate": 3.962264150943396e-06,
+      "loss": 0.6193,
+      "step": 488
     },
     {
       "epoch": 9.23,
-      "learning_rate": 1.5315315315315316e-05,
-      "loss": 0.1806,
-      "step": 1025
-    },
-    {
-      "epoch": 9.24,
-      "learning_rate": 1.5135135135135136e-05,
-      "loss": 0.192,
-      "step": 1026
+      "learning_rate": 3.867924528301887e-06,
+      "loss": 0.6021,
+      "step": 489
     },
     {
       "epoch": 9.25,
-      "learning_rate": 1.4954954954954956e-05,
-      "loss": 0.1636,
-      "step": 1027
+      "learning_rate": 3.7735849056603773e-06,
+      "loss": 0.6095,
+      "step": 490
     },
     {
       "epoch": 9.26,
-      "learning_rate": 1.4774774774774774e-05,
-      "loss": 0.1797,
-      "step": 1028
-    },
-    {
-      "epoch": 9.27,
-      "learning_rate": 1.4594594594594596e-05,
-      "loss": 0.1976,
-      "step": 1029
+      "learning_rate": 3.679245283018868e-06,
+      "loss": 0.5709,
+      "step": 491
     },
     {
       "epoch": 9.28,
-      "learning_rate": 1.4414414414414416e-05,
-      "loss": 0.1657,
-      "step": 1030
-    },
-    {
-      "epoch": 9.29,
-      "learning_rate": 1.4234234234234234e-05,
-      "loss": 0.178,
-      "step": 1031
+      "learning_rate": 3.5849056603773586e-06,
+      "loss": 0.6103,
+      "step": 492
     },
     {
       "epoch": 9.3,
-      "learning_rate": 1.4054054054054055e-05,
-      "loss": 0.1709,
-      "step": 1032
-    },
-    {
-      "epoch": 9.31,
-      "learning_rate": 1.3873873873873874e-05,
-      "loss": 0.1604,
-      "step": 1033
-    },
-    {
-      "epoch": 9.32,
-      "learning_rate": 1.3693693693693693e-05,
-      "loss": 0.1804,
-      "step": 1034
+      "learning_rate": 3.4905660377358493e-06,
+      "loss": 0.6074,
+      "step": 493
     },
     {
       "epoch": 9.32,
-      "learning_rate": 1.3513513513513515e-05,
-      "loss": 0.1661,
-      "step": 1035
-    },
-    {
-      "epoch": 9.33,
-      "learning_rate": 1.3333333333333333e-05,
-      "loss": 0.1542,
-      "step": 1036
+      "learning_rate": 3.3962264150943395e-06,
+      "loss": 0.5336,
+      "step": 494
     },
     {
       "epoch": 9.34,
-      "learning_rate": 1.3153153153153153e-05,
-      "loss": 0.1807,
-      "step": 1037
-    },
-    {
-      "epoch": 9.35,
-      "learning_rate": 1.2972972972972975e-05,
-      "loss": 0.1976,
-      "step": 1038
+      "learning_rate": 3.30188679245283e-06,
+      "loss": 0.5793,
+      "step": 495
     },
     {
       "epoch": 9.36,
-      "learning_rate": 1.2792792792792793e-05,
-      "loss": 0.162,
-      "step": 1039
-    },
-    {
-      "epoch": 9.37,
-      "learning_rate": 1.2612612612612611e-05,
-      "loss": 0.173,
-      "step": 1040
+      "learning_rate": 3.207547169811321e-06,
+      "loss": 0.5899,
+      "step": 496
     },
     {
       "epoch": 9.38,
-      "learning_rate": 1.2432432432432435e-05,
-      "loss": 0.1679,
-      "step": 1041
-    },
-    {
-      "epoch": 9.39,
-      "learning_rate": 1.2252252252252253e-05,
-      "loss": 0.1732,
-      "step": 1042
+      "learning_rate": 3.1132075471698115e-06,
+      "loss": 0.6602,
+      "step": 497
     },
     {
       "epoch": 9.4,
-      "learning_rate": 1.2072072072072073e-05,
-      "loss": 0.1774,
-      "step": 1043
-    },
-    {
-      "epoch": 9.41,
-      "learning_rate": 1.1891891891891893e-05,
-      "loss": 0.1579,
-      "step": 1044
-    },
-    {
-      "epoch": 9.41,
-      "learning_rate": 1.1711711711711713e-05,
-      "loss": 0.1631,
-      "step": 1045
+      "learning_rate": 3.018867924528302e-06,
+      "loss": 0.5698,
+      "step": 498
     },
     {
       "epoch": 9.42,
-      "learning_rate": 1.1531531531531532e-05,
-      "loss": 0.1698,
-      "step": 1046
+      "learning_rate": 2.9245283018867924e-06,
+      "loss": 0.5968,
+      "step": 499
     },
     {
       "epoch": 9.43,
-      "learning_rate": 1.1351351351351352e-05,
-      "loss": 0.1702,
-      "step": 1047
-    },
-    {
-      "epoch": 9.44,
-      "learning_rate": 1.1171171171171172e-05,
-      "loss": 0.1565,
-      "step": 1048
+      "learning_rate": 2.830188679245283e-06,
+      "loss": 0.6084,
+      "step": 500
     },
     {
       "epoch": 9.45,
-      "learning_rate": 1.0990990990990992e-05,
-      "loss": 0.1443,
-      "step": 1049
-    },
-    {
-      "epoch": 9.46,
-      "learning_rate": 1.0810810810810812e-05,
-      "loss": 0.1805,
-      "step": 1050
+      "learning_rate": 2.7358490566037738e-06,
+      "loss": 0.5757,
+      "step": 501
     },
     {
       "epoch": 9.47,
-      "learning_rate": 1.063063063063063e-05,
-      "loss": 0.1462,
-      "step": 1051
-    },
-    {
-      "epoch": 9.48,
-      "learning_rate": 1.0450450450450452e-05,
-      "loss": 0.1723,
-      "step": 1052
+      "learning_rate": 2.6415094339622644e-06,
+      "loss": 0.5803,
+      "step": 502
     },
     {
       "epoch": 9.49,
-      "learning_rate": 1.0270270270270272e-05,
-      "loss": 0.1684,
-      "step": 1053
-    },
-    {
-      "epoch": 9.5,
-      "learning_rate": 1.009009009009009e-05,
-      "loss": 0.1866,
-      "step": 1054
-    },
-    {
-      "epoch": 9.5,
-      "learning_rate": 9.90990990990991e-06,
-      "loss": 0.1562,
-      "step": 1055
+      "learning_rate": 2.547169811320755e-06,
+      "loss": 0.6314,
+      "step": 503
     },
     {
       "epoch": 9.51,
-      "learning_rate": 9.729729729729732e-06,
-      "loss": 0.1614,
-      "step": 1056
-    },
-    {
-      "epoch": 9.52,
-      "learning_rate": 9.54954954954955e-06,
-      "loss": 0.1592,
-      "step": 1057
+      "learning_rate": 2.4528301886792457e-06,
+      "loss": 0.6091,
+      "step": 504
     },
     {
       "epoch": 9.53,
-      "learning_rate": 9.36936936936937e-06,
-      "loss": 0.1659,
-      "step": 1058
-    },
-    {
-      "epoch": 9.54,
-      "learning_rate": 9.18918918918919e-06,
-      "loss": 0.1632,
-      "step": 1059
+      "learning_rate": 2.358490566037736e-06,
+      "loss": 0.6383,
+      "step": 505
     },
     {
       "epoch": 9.55,
-      "learning_rate": 9.00900900900901e-06,
-      "loss": 0.1697,
-      "step": 1060
-    },
-    {
-      "epoch": 9.56,
-      "learning_rate": 8.82882882882883e-06,
-      "loss": 0.1893,
-      "step": 1061
+      "learning_rate": 2.2641509433962262e-06,
+      "loss": 0.697,
+      "step": 506
     },
     {
       "epoch": 9.57,
-      "learning_rate": 8.64864864864865e-06,
-      "loss": 0.1768,
-      "step": 1062
+      "learning_rate": 2.169811320754717e-06,
+      "loss": 0.6193,
+      "step": 507
     },
     {
       "epoch": 9.58,
-      "learning_rate": 8.46846846846847e-06,
-      "loss": 0.1828,
-      "step": 1063
-    },
-    {
-      "epoch": 9.59,
-      "learning_rate": 8.288288288288289e-06,
-      "loss": 0.1687,
-      "step": 1064
-    },
-    {
-      "epoch": 9.59,
-      "learning_rate": 8.108108108108109e-06,
-      "loss": 0.1662,
-      "step": 1065
+      "learning_rate": 2.0754716981132075e-06,
+      "loss": 0.5657,
+      "step": 508
     },
     {
       "epoch": 9.6,
-      "learning_rate": 7.927927927927927e-06,
-      "loss": 0.1681,
-      "step": 1066
-    },
-    {
-      "epoch": 9.61,
-      "learning_rate": 7.747747747747749e-06,
-      "loss": 0.1521,
-      "step": 1067
+      "learning_rate": 1.981132075471698e-06,
+      "loss": 0.5766,
+      "step": 509
     },
     {
       "epoch": 9.62,
-      "learning_rate": 7.567567567567568e-06,
-      "loss": 0.158,
-      "step": 1068
-    },
-    {
-      "epoch": 9.63,
-      "learning_rate": 7.387387387387387e-06,
-      "loss": 0.1679,
-      "step": 1069
+      "learning_rate": 1.8867924528301887e-06,
+      "loss": 0.6691,
+      "step": 510
     },
     {
       "epoch": 9.64,
-      "learning_rate": 7.207207207207208e-06,
-      "loss": 0.158,
-      "step": 1070
-    },
-    {
-      "epoch": 9.65,
-      "learning_rate": 7.027027027027028e-06,
-      "loss": 0.1688,
-      "step": 1071
+      "learning_rate": 1.7924528301886793e-06,
+      "loss": 0.5512,
+      "step": 511
     },
     {
       "epoch": 9.66,
-      "learning_rate": 6.846846846846847e-06,
-      "loss": 0.1714,
-      "step": 1072
-    },
-    {
-      "epoch": 9.67,
-      "learning_rate": 6.666666666666667e-06,
-      "loss": 0.1693,
-      "step": 1073
-    },
-    {
-      "epoch": 9.68,
-      "learning_rate": 6.486486486486487e-06,
-      "loss": 0.1465,
-      "step": 1074
+      "learning_rate": 1.6981132075471698e-06,
+      "loss": 0.598,
+      "step": 512
     },
     {
       "epoch": 9.68,
-      "learning_rate": 6.306306306306306e-06,
-      "loss": 0.1723,
-      "step": 1075
-    },
-    {
-      "epoch": 9.69,
-      "learning_rate": 6.126126126126126e-06,
-      "loss": 0.1704,
-      "step": 1076
+      "learning_rate": 1.6037735849056604e-06,
+      "loss": 0.5527,
+      "step": 513
     },
     {
       "epoch": 9.7,
-      "learning_rate": 5.945945945945946e-06,
-      "loss": 0.143,
-      "step": 1077
-    },
-    {
-      "epoch": 9.71,
-      "learning_rate": 5.765765765765766e-06,
-      "loss": 0.1696,
-      "step": 1078
+      "learning_rate": 1.509433962264151e-06,
+      "loss": 0.6116,
+      "step": 514
     },
     {
       "epoch": 9.72,
-      "learning_rate": 5.585585585585586e-06,
-      "loss": 0.1596,
-      "step": 1079
-    },
-    {
-      "epoch": 9.73,
-      "learning_rate": 5.405405405405406e-06,
-      "loss": 0.2013,
-      "step": 1080
+      "learning_rate": 1.4150943396226415e-06,
+      "loss": 0.6351,
+      "step": 515
     },
     {
       "epoch": 9.74,
-      "learning_rate": 5.225225225225226e-06,
-      "loss": 0.191,
-      "step": 1081
+      "learning_rate": 1.3207547169811322e-06,
+      "loss": 0.5633,
+      "step": 516
     },
     {
       "epoch": 9.75,
-      "learning_rate": 5.045045045045045e-06,
-      "loss": 0.177,
-      "step": 1082
-    },
-    {
-      "epoch": 9.76,
-      "learning_rate": 4.864864864864866e-06,
-      "loss": 0.1849,
-      "step": 1083
-    },
-    {
-      "epoch": 9.77,
-      "learning_rate": 4.684684684684685e-06,
-      "loss": 0.1549,
-      "step": 1084
+      "learning_rate": 1.2264150943396229e-06,
+      "loss": 0.5586,
+      "step": 517
     },
     {
       "epoch": 9.77,
-      "learning_rate": 4.504504504504505e-06,
-      "loss": 0.1875,
-      "step": 1085
-    },
-    {
-      "epoch": 9.78,
-      "learning_rate": 4.324324324324325e-06,
-      "loss": 0.1544,
-      "step": 1086
+      "learning_rate": 1.1320754716981131e-06,
+      "loss": 0.5783,
+      "step": 518
     },
     {
       "epoch": 9.79,
-      "learning_rate": 4.1441441441441446e-06,
-      "loss": 0.1816,
-      "step": 1087
-    },
-    {
-      "epoch": 9.8,
-      "learning_rate": 3.963963963963964e-06,
-      "loss": 0.1707,
-      "step": 1088
+      "learning_rate": 1.0377358490566038e-06,
+      "loss": 0.638,
+      "step": 519
     },
     {
       "epoch": 9.81,
-      "learning_rate": 3.783783783783784e-06,
-      "loss": 0.1575,
-      "step": 1089
-    },
-    {
-      "epoch": 9.82,
-      "learning_rate": 3.603603603603604e-06,
-      "loss": 0.1607,
-      "step": 1090
+      "learning_rate": 9.433962264150943e-07,
+      "loss": 0.5432,
+      "step": 520
     },
     {
       "epoch": 9.83,
-      "learning_rate": 3.4234234234234234e-06,
-      "loss": 0.1771,
-      "step": 1091
-    },
-    {
-      "epoch": 9.84,
-      "learning_rate": 3.2432432432432437e-06,
-      "loss": 0.1688,
-      "step": 1092
+      "learning_rate": 8.490566037735849e-07,
+      "loss": 0.5232,
+      "step": 521
     },
     {
       "epoch": 9.85,
-      "learning_rate": 3.063063063063063e-06,
-      "loss": 0.1689,
-      "step": 1093
-    },
-    {
-      "epoch": 9.86,
-      "learning_rate": 2.882882882882883e-06,
-      "loss": 0.1711,
-      "step": 1094
-    },
-    {
-      "epoch": 9.86,
-      "learning_rate": 2.702702702702703e-06,
-      "loss": 0.1696,
-      "step": 1095
+      "learning_rate": 7.547169811320755e-07,
+      "loss": 0.6154,
+      "step": 522
     },
     {
       "epoch": 9.87,
-      "learning_rate": 2.5225225225225225e-06,
-      "loss": 0.1863,
-      "step": 1096
-    },
-    {
-      "epoch": 9.88,
-      "learning_rate": 2.3423423423423424e-06,
-      "loss": 0.1648,
-      "step": 1097
+      "learning_rate": 6.603773584905661e-07,
+      "loss": 0.637,
+      "step": 523
     },
     {
       "epoch": 9.89,
-      "learning_rate": 2.1621621621621623e-06,
-      "loss": 0.1883,
-      "step": 1098
-    },
-    {
-      "epoch": 9.9,
-      "learning_rate": 1.981981981981982e-06,
-      "loss": 0.1593,
-      "step": 1099
+      "learning_rate": 5.660377358490566e-07,
+      "loss": 0.5593,
+      "step": 524
     },
     {
       "epoch": 9.91,
-      "learning_rate": 1.801801801801802e-06,
-      "loss": 0.1602,
-      "step": 1100
+      "learning_rate": 4.7169811320754717e-07,
+      "loss": 0.6447,
+      "step": 525
     },
     {
       "epoch": 9.92,
-      "learning_rate": 1.6216216216216219e-06,
-      "loss": 0.2187,
-      "step": 1101
-    },
-    {
-      "epoch": 9.93,
-      "learning_rate": 1.4414414414414416e-06,
-      "loss": 0.1863,
-      "step": 1102
+      "learning_rate": 3.773584905660378e-07,
+      "loss": 0.6056,
+      "step": 526
     },
     {
       "epoch": 9.94,
-      "learning_rate": 1.2612612612612613e-06,
-      "loss": 0.1656,
-      "step": 1103
-    },
-    {
-      "epoch": 9.95,
-      "learning_rate": 1.0810810810810812e-06,
-      "loss": 0.1939,
-      "step": 1104
-    },
-    {
-      "epoch": 9.95,
-      "learning_rate": 9.00900900900901e-07,
-      "loss": 0.1407,
-      "step": 1105
+      "learning_rate": 2.830188679245283e-07,
+      "loss": 0.6928,
+      "step": 527
     },
     {
       "epoch": 9.96,
-      "learning_rate": 7.207207207207208e-07,
-      "loss": 0.1753,
-      "step": 1106
-    },
-    {
-      "epoch": 9.97,
-      "learning_rate": 5.405405405405406e-07,
-      "loss": 0.1542,
-      "step": 1107
+      "learning_rate": 1.886792452830189e-07,
+      "loss": 0.609,
+      "step": 528
     },
     {
       "epoch": 9.98,
-      "learning_rate": 3.603603603603604e-07,
-      "loss": 0.1609,
-      "step": 1108
-    },
-    {
-      "epoch": 9.99,
-      "learning_rate": 1.801801801801802e-07,
-      "loss": 0.1702,
-      "step": 1109
+      "learning_rate": 9.433962264150944e-08,
+      "loss": 0.6045,
+      "step": 529
     },
     {
       "epoch": 10.0,
       "learning_rate": 0.0,
-      "loss": 0.1462,
-      "step": 1110
+      "loss": 0.5314,
+      "step": 530
     },
     {
       "epoch": 10.0,
-      "step": 1110,
-      "total_flos": 6398509737984000.0,
-      "train_loss": 0.5453134933838973,
-      "train_runtime": 627.0712,
-      "train_samples_per_second": 56.246,
-      "train_steps_per_second": 1.77
+      "step": 530,
+      "total_flos": 6755526882816000.0,
+      "train_loss": 1.0295860317518126,
+      "train_runtime": 364.5885,
+      "train_samples_per_second": 91.912,
+      "train_steps_per_second": 1.454
     }
   ],
-  "max_steps": 1110,
+  "max_steps": 530,
   "num_train_epochs": 10,
-  "total_flos": 6398509737984000.0,
+  "total_flos": 6755526882816000.0,
   "trial_name": null,
   "trial_params": null
 }