мои потери начинаются с 0,65, затем уменьшаются до достижения 0,01 и прекращают изменяться, когда я проверяю результат, который все еще не хорош, должен ли я ждать или что мне нужно делать?
моя функция активации - сиглоид последнего слоя lrelu моя функция потерь двоичная кросс-энтропия
все слои полностью связаны слой
размер пакета = 4
('ep:', 0, 'i:', 105, 'train aeu loss:', 0.6398013, 'total loss', 0.6398013)
('ep:', 0, 'i:', 106, 'train aeu loss:', 0.63618463, 'total loss', 0.63618463)
('ep:', 0, 'i:', 107, 'train aeu loss:', 0.6323359, 'total loss', 0.6323359)
('ep:', 0, 'i:', 108, 'train aeu loss:', 0.628239, 'total loss', 0.628239)
('ep:', 0, 'i:', 109, 'train aeu loss:', 0.6238803, 'total loss', 0.6238803)
('ep:', 0, 'i:', 110, 'train aeu loss:', 0.61924565, 'total loss', 0.61924565)
('ep:', 0, 'i:', 111, 'train aeu loss:', 0.6143209, 'total loss', 0.6143209)
('ep:', 0, 'i:', 112, 'train aeu loss:', 0.609091, 'total loss', 0.609091)
('ep:', 0, 'i:', 113, 'train aeu loss:', 0.60354114, 'total loss', 0.60354114)
('ep:', 0, 'i:', 114, 'train aeu loss:', 0.597656, 'total loss', 0.597656)
('ep:', 0, 'i:', 115, 'train aeu loss:', 0.5914216, 'total loss', 0.5914216)
('ep:', 0, 'i:', 116, 'train aeu loss:', 0.58482814, 'total loss', 0.58482814)
('ep:', 0, 'i:', 117, 'train aeu loss:', 0.5778549, 'total loss', 0.5778549)
('ep:', 0, 'i:', 118, 'train aeu loss:', 0.5704867, 'total loss', 0.5704867)
('ep:', 0, 'i:', 119, 'train aeu loss:', 0.5627097, 'total loss', 0.5627097)
('ep:', 0, 'i:', 120, 'train aeu loss:', 0.5545108, 'total loss', 0.5545108)
('ep:', 0, 'i:', 121, 'train aeu loss:', 0.54587483, 'total loss', 0.54587483)
('ep:', 0, 'i:', 122, 'train aeu loss:', 0.5367902, 'total loss', 0.5367902)
('ep:', 0, 'i:', 123, 'train aeu loss:', 0.5272466, 'total loss', 0.5272466)
('ep:', 0, 'i:', 124, 'train aeu loss:', 0.51723653, 'total loss', 0.51723653)
('ep:', 0, 'i:', 125, 'train aeu loss:', 0.5067488, 'total loss', 0.5067488)
('ep:', 0, 'i:', 126, 'train aeu loss:', 0.49573797, 'total loss', 0.49573797)
('ep:', 0, 'i:', 127, 'train aeu loss:', 0.48423475, 'total loss', 0.48423475)
('ep:', 0, 'i:', 128, 'train aeu loss:', 0.47223735, 'total loss', 0.47223735)
('ep:', 0, 'i:', 129, 'train aeu loss:', 0.45974857, 'total loss', 0.45974857)
('ep:', 0, 'i:', 130, 'train aeu loss:', 0.44677627, 'total loss', 0.44677627)
('ep:', 0, 'i:', 131, 'train aeu loss:', 0.43333283, 'total loss', 0.43333283)
('ep:', 0, 'i:', 132, 'train aeu loss:', 0.41943556, 'total loss', 0.41943556)
('ep:', 0, 'i:', 133, 'train aeu loss:', 0.40510648, 'total loss', 0.40510648)
('ep:', 0, 'i:', 134, 'train aeu loss:', 0.3903734, 'total loss', 0.3903734)
('ep:', 0, 'i:', 135, 'train aeu loss:', 0.37527022, 'total loss', 0.37527022)
('ep:', 0, 'i:', 136, 'train aeu loss:', 0.35984012, 'total loss', 0.35984012)
('ep:', 0, 'i:', 137, 'train aeu loss:', 0.34412184, 'total loss', 0.34412184)
('ep:', 0, 'i:', 138, 'train aeu loss:', 0.32817447, 'total loss', 0.32817447)
('ep:', 0, 'i:', 139, 'train aeu loss:', 0.3120531, 'total loss', 0.3120531)
('ep:', 0, 'i:', 140, 'train aeu loss:', 0.29582202, 'total loss', 0.29582202)
('ep:', 0, 'i:', 141, 'train aeu loss:', 0.27955088, 'total loss', 0.27955088)
('ep:', 0, 'i:', 142, 'train aeu loss:', 0.2633138, 'total loss', 0.2633138)
('ep:', 0, 'i:', 143, 'train aeu loss:', 0.24718845, 'total loss', 0.24718845)
('ep:', 0, 'i:', 144, 'train aeu loss:', 0.23125465, 'total loss', 0.23125465)
('ep:', 0, 'i:', 145, 'train aeu loss:', 0.2155931, 'total loss', 0.2155931)
('ep:', 0, 'i:', 146, 'train aeu loss:', 0.20028353, 'total loss', 0.20028353)
('ep:', 0, 'i:', 147, 'train aeu loss:', 0.18540315, 'total loss', 0.18540315)
('ep:', 0, 'i:', 148, 'train aeu loss:', 0.17102464, 'total loss', 0.17102464)
('ep:', 0, 'i:', 149, 'train aeu loss:', 0.15721451, 'total loss', 0.15721451)
('ep:', 0, 'i:', 150, 'train aeu loss:', 0.14403132, 'total loss', 0.14403132)
('ep:', 0, 'i:', 151, 'train aeu loss:', 0.13152426, 'total loss', 0.13152426)
('ep:', 0, 'i:', 152, 'train aeu loss:', 0.11973214, 'total loss', 0.11973214)
('ep:', 0, 'i:', 153, 'train aeu loss:', 0.10868255, 'total loss', 0.10868255)
('ep:', 0, 'i:', 154, 'train aeu loss:', 0.09839133, 'total loss', 0.09839133)
('ep:', 0, 'i:', 155, 'train aeu loss:', 0.08886323, 'total loss', 0.08886323)
('ep:', 0, 'i:', 156, 'train aeu loss:', 0.08009201, 'total loss', 0.08009201)
('ep:', 0, 'i:', 157, 'train aeu loss:', 0.07206154, 'total loss', 0.07206154)
('ep:', 0, 'i:', 158, 'train aeu loss:', 0.06474711, 'total loss', 0.06474711)
('ep:', 0, 'i:', 159, 'train aeu loss:', 0.058116734, 'total loss', 0.058116734)
('ep:', 0, 'i:', 160, 'train aeu loss:', 0.052132864, 'total loss', 0.052132864)
('ep:', 0, 'i:', 161, 'train aeu loss:', 0.046753947, 'total loss', 0.046753947)
('ep:', 0, 'i:', 162, 'train aeu loss:', 0.04193723, 'total loss', 0.04193723)
('ep:', 0, 'i:', 163, 'train aeu loss:', 0.03763741, 'total loss', 0.03763741)
('ep:', 0, 'i:', 164, 'train aeu loss:', 0.03380844, 'total loss', 0.03380844)
('ep:', 0, 'i:', 165, 'train aeu loss:', 0.030407183, 'total loss', 0.030407183)
('ep:', 0, 'i:', 166, 'train aeu loss:', 0.027393982, 'total loss', 0.027393982)
('ep:', 0, 'i:', 167, 'train aeu loss:', 0.024747191, 'total loss', 0.024747191)
('ep:', 0, 'i:', 168, 'train aeu loss:', 0.022434363, 'total loss', 0.022434363)
('ep:', 0, 'i:', 169, 'train aeu loss:', 0.020445904, 'total loss', 0.020445904)
('ep:', 0, 'i:', 170, 'train aeu loss:', 0.0187451, 'total loss', 0.0187451)
('ep:', 0, 'i:', 171, 'train aeu loss:', 0.01729209, 'total loss', 0.01729209)
('ep:', 0, 'i:', 172, 'train aeu loss:', 0.01606659, 'total loss', 0.01606659)
('ep:', 0, 'i:', 173, 'train aeu loss:', 0.015032028, 'total loss', 0.015032028)
('ep:', 0, 'i:', 174, 'train aeu loss:', 0.014165167, 'total loss', 0.014165167)
('ep:', 0, 'i:', 175, 'train aeu loss:', 0.013441812, 'total loss', 0.013441812)
('ep:', 0, 'i:', 176, 'train aeu loss:', 0.012847084, 'total loss', 0.012847084)
('ep:', 0, 'i:', 177, 'train aeu loss:', 0.0123621, 'total loss', 0.0123621)
('ep:', 0, 'i:', 178, 'train aeu loss:', 0.011963224, 'total loss', 0.011963224)
('ep:', 0, 'i:', 179, 'train aeu loss:', 0.011638624, 'total loss', 0.011638624)
('ep:', 0, 'i:', 180, 'train aeu loss:', 0.0113756955, 'total loss', 0.0113756955)
('ep:', 0, 'i:', 181, 'train aeu loss:', 0.011161308, 'total loss', 0.011161308)
('ep:', 0, 'i:', 182, 'train aeu loss:', 0.01098865, 'total loss', 0.01098865)
('ep:', 0, 'i:', 183, 'train aeu loss:', 0.010849173, 'total loss', 0.010849173)
('ep:', 0, 'i:', 184, 'train aeu loss:', 0.010730466, 'total loss', 0.010730466)
('ep:', 0, 'i:', 185, 'train aeu loss:', 0.010630407, 'total loss', 0.010630407)
('ep:', 0, 'i:', 186, 'train aeu loss:', 0.010549757, 'total loss', 0.010549757)
('ep:', 0, 'i:', 187, 'train aeu loss:', 0.010485613, 'total loss', 0.010485613)
('ep:', 0, 'i:', 188, 'train aeu loss:', 0.010433215, 'total loss', 0.010433215)
('ep:', 0, 'i:', 189, 'train aeu loss:', 0.010389576, 'total loss', 0.010389576)
('ep:', 0, 'i:', 190, 'train aeu loss:', 0.010352248, 'total loss', 0.010352248)
('ep:', 0, 'i:', 191, 'train aeu loss:', 0.0103185065, 'total loss', 0.0103185065)
('ep:', 0, 'i:', 192, 'train aeu loss:', 0.010290246, 'total loss', 0.010290246)
('ep:', 0, 'i:', 193, 'train aeu loss:', 0.01026703, 'total loss', 0.01026703)
('ep:', 0, 'i:', 194, 'train aeu loss:', 0.010247082, 'total loss', 0.010247082)
('ep:', 0, 'i:', 195, 'train aeu loss:', 0.010229151, 'total loss', 0.010229151)
('ep:', 0, 'i:', 196, 'train aeu loss:', 0.010213042, 'total loss', 0.010213042)
('ep:', 0, 'i:', 197, 'train aeu loss:', 0.010198735, 'total loss', 0.010198735)
('ep:', 0, 'i:', 198, 'train aeu loss:', 0.010185955, 'total loss', 0.010185955)
('ep:', 0, 'i:', 199, 'train aeu loss:', 0.010174904, 'total loss', 0.010174904)
('ep:', 0, 'i:', 200, 'train aeu loss:', 0.010165406, 'total loss', 0.010165406)
('ep:', 0, 'i:', 201, 'train aeu loss:', 0.010156921, 'total loss', 0.010156921)
('ep:', 0, 'i:', 202, 'train aeu loss:', 0.010149105, 'total loss', 0.010149105)
('ep:', 0, 'i:', 203, 'train aeu loss:', 0.010142249, 'total loss', 0.010142249)
('ep:', 0, 'i:', 204, 'train aeu loss:', 0.010136051, 'total loss', 0.010136051)
('ep:', 0, 'i:', 205, 'train aeu loss:', 0.010130281, 'total loss', 0.010130281)
('ep:', 0, 'i:', 206, 'train aeu loss:', 0.010125015, 'total loss', 0.010125015)
('ep:', 0, 'i:', 207, 'train aeu loss:', 0.0101200305, 'total loss', 0.0101200305)
('ep:', 0, 'i:', 208, 'train aeu loss:', 0.010115301, 'total loss', 0.010115301)
('ep:', 0, 'i:', 209, 'train aeu loss:', 0.010111166, 'total loss', 0.010111166)
('ep:', 0, 'i:', 210, 'train aeu loss:', 0.010107304, 'total loss', 0.010107304)
('ep:', 0, 'i:', 211, 'train aeu loss:', 0.010103617, 'total loss', 0.010103617)
('ep:', 0, 'i:', 212, 'train aeu loss:', 0.010100296, 'total loss', 0.010100296)
('ep:', 0, 'i:', 213, 'train aeu loss:', 0.010097368, 'total loss', 0.010097368)
('ep:', 0, 'i:', 214, 'train aeu loss:', 0.010094685, 'total loss', 0.010094685)
('ep:', 0, 'i:', 215, 'train aeu loss:', 0.010092108, 'total loss', 0.010092108)
('ep:', 0, 'i:', 216, 'train aeu loss:', 0.010089636, 'total loss', 0.010089636)
('ep:', 0, 'i:', 217, 'train aeu loss:', 0.0100874165, 'total loss', 0.0100874165)
('ep:', 0, 'i:', 218, 'train aeu loss:', 0.010085274, 'total loss', 0.010085274)
('ep:', 0, 'i:', 219, 'train aeu loss:', 0.010083204, 'total loss', 0.010083204)
('ep:', 0, 'i:', 219, 'model saved!')
('ep:', 0, 'i:', 220, 'train aeu loss:', 0.010081198, 'total loss', 0.010081198)
('ep:', 0, 'i:', 238, 'train aeu loss:', 0.010060191, 'total loss', 0.010060191)
('ep:', 0, 'i:', 239, 'train aeu loss:', 0.010059458, 'total loss', 0.010059458)
('ep:', 0, 'i:', 240, 'train aeu loss:', 0.010058737, 'total loss', 0.010058737)
('ep:', 0, 'i:', 241, 'train aeu loss:', 0.010058028, 'total loss', 0.010058028)
('ep:', 0, 'i:', 242, 'train aeu loss:', 0.010057329, 'total loss', 0.010057329)
('ep:', 0, 'i:', 243, 'train aeu loss:', 0.010056752, 'total loss', 0.010056752)
('ep:', 0, 'i:', 244, 'train aeu loss:', 0.010056196, 'total loss', 0.010056196)
('ep:', 0, 'i:', 245, 'train aeu loss:', 0.010055648, 'total loss', 0.010055648)
('ep:', 0, 'i:', 246, 'train aeu loss:', 0.010055143, 'total loss', 0.010055143)
('ep:', 0, 'i:', 247, 'train aeu loss:', 0.010054855, 'total loss', 0.010054855)
('ep:', 0, 'i:', 248, 'train aeu loss:', 0.010054571, 'total loss', 0.010054571)
('ep:', 0, 'i:', 249, 'train aeu loss:', 0.010054291, 'total loss', 0.010054291)
('ep:', 0, 'i:', 250, 'train aeu loss:', 0.0100540165, 'total loss', 0.0100540165)
('ep:', 0, 'i:', 251, 'train aeu loss:', 0.010053745, 'total loss', 0.010053745)
('ep:', 0, 'i:', 252, 'train aeu loss:', 0.010053477, 'total loss', 0.010053477)
('ep:', 0, 'i:', 253, 'train aeu loss:', 0.010053213, 'total loss', 0.010053213)
('ep:', 0, 'i:', 254, 'train aeu loss:', 0.010052953, 'total loss', 0.010052953)
('ep:', 0, 'i:', 255, 'train aeu loss:', 0.010052712, 'total loss', 0.010052712)
('ep:', 0, 'i:', 256, 'train aeu loss:', 0.010052575, 'total loss', 0.010052575)
('ep:', 0, 'i:', 257, 'train aeu loss:', 0.01005244, 'total loss', 0.01005244)
('ep:', 0, 'i:', 258, 'train aeu loss:', 0.010052307, 'total loss', 0.010052307)
('ep:', 0, 'i:', 259, 'train aeu loss:', 0.010052175, 'total loss', 0.010052175)
('ep:', 0, 'i:', 260, 'train aeu loss:', 0.010052046, 'total loss', 0.010052046)
('ep:', 0, 'i:', 261, 'train aeu loss:', 0.010051917, 'total loss', 0.010051917)
('ep:', 0, 'i:', 262, 'train aeu loss:', 0.010051791, 'total loss', 0.010051791)
('ep:', 0, 'i:', 263, 'train aeu loss:', 0.010051665, 'total loss', 0.010051665)
('ep:', 0, 'i:', 264, 'train aeu loss:', 0.010051541, 'total loss', 0.010051541)
('ep:', 0, 'i:', 265, 'train aeu loss:', 0.010051419, 'total loss', 0.010051419)
('ep:', 0, 'i:', 266, 'train aeu loss:', 0.010051298, 'total loss', 0.010051298)
('ep:', 0, 'i:', 267, 'train aeu loss:', 0.010051178, 'total loss', 0.010051178)
('ep:', 0, 'i:', 268, 'train aeu loss:', 0.0100510605, 'total loss', 0.0100510605)
('ep:', 0, 'i:', 269, 'train aeu loss:', 0.010050943, 'total loss', 0.010050943)
('ep:', 0, 'i:', 270, 'train aeu loss:', 0.010050828, 'total loss', 0.010050828)
('ep:', 0, 'i:', 271, 'train aeu loss:', 0.010050712, 'total loss', 0.010050712)
('ep:', 0, 'i:', 272, 'train aeu loss:', 0.010050599, 'total loss', 0.010050599)
('ep:', 0, 'i:', 273, 'train aeu loss:', 0.010050487, 'total loss', 0.010050487)
('ep:', 0, 'i:', 274, 'train aeu loss:', 0.010050376, 'total loss', 0.010050376)
('ep:', 0, 'i:', 275, 'train aeu loss:', 0.010050327, 'total loss', 0.010050327)
('ep:', 0, 'i:', 276, 'train aeu loss:', 0.010050327, 'total loss', 0.010050327)
('ep:', 0, 'i:', 277, 'train aeu loss:', 0.010050327, 'total loss', 0.010050327)
('ep:', 0, 'i:', 278, 'train aeu loss:', 0.010050327, 'total loss', 0.010050327)
('ep:', 0, 'i:', 279, 'train aeu loss:', 0.010050327, 'total loss', 0.010050327)
('ep:', 0, 'i:', 280, 'train aeu loss:', 0.010050327, 'total loss', 0.010050327)
('ep:', 0, 'i:', 281, 'train aeu loss:', 0.010050327, 'total loss', 0.010050327)
('ep:', 0, 'i:', 282, 'train aeu loss:', 0.010050327, 'total loss', 0.010050327)
('ep:', 0, 'i:', 283, 'train aeu loss:', 0.010050327, 'total loss', 0.010050327)
('ep:', 0, 'i:', 284, 'train aeu loss:', 0.010050327, 'total loss', 0.010050327)
('ep:', 0, 'i:', 285, 'train aeu loss:', 0.010050327, 'total loss', 0.010050327)
('ep:', 0, 'i:', 286, 'train aeu loss:', 0.010050327, 'total loss', 0.010050327)
если я уменьшу набор данных, у меня есть 7000 выборок для обучения или увеличения обучения ставка