ossaili/27_Architectural_Styles_Classifier
0
1EfficientNet(2 (features): Sequential(3 (0): Conv2dNormActivation(4 (0): Conv2d(3, 32, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)5 (1): BatchNorm2d(32, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)6 (2): SiLU(inplace=True)7 )8 (1): Sequential(9 (0): FusedMBConv(10 (block): Sequential(11 (0): Conv2dNormActivation(12 (0): Conv2d(32, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)13 (1): BatchNorm2d(32, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)14 (2): SiLU(inplace=True)15 )16 )17 (stochastic_depth): StochasticDepth(p=0.0, mode=row)18 )19 (1): FusedMBConv(20 (block): Sequential(21 (0): Conv2dNormActivation(22 (0): Conv2d(32, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)23 (1): BatchNorm2d(32, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)24 (2): SiLU(inplace=True)25 )26 )27 (stochastic_depth): StochasticDepth(p=0.002531645569620253, mode=row)28 )29 (2): FusedMBConv(30 (block): Sequential(31 (0): Conv2dNormActivation(32 (0): Conv2d(32, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)33 (1): BatchNorm2d(32, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)34 (2): SiLU(inplace=True)35 )36 )37 (stochastic_depth): StochasticDepth(p=0.005063291139240506, mode=row)38 )39 (3): FusedMBConv(40 (block): Sequential(41 (0): Conv2dNormActivation(42 (0): Conv2d(32, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)43 (1): BatchNorm2d(32, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)44 (2): SiLU(inplace=True)45 )46 )47 (stochastic_depth): StochasticDepth(p=0.007594936708860761, mode=row)48 )49 )50 (2): Sequential(51 (0): FusedMBConv(52 (block): Sequential(53 (0): Conv2dNormActivation(54 (0): Conv2d(32, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)55 (1): BatchNorm2d(128, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)56 (2): SiLU(inplace=True)57 )58 (1): Conv2dNormActivation(59 (0): Conv2d(128, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)60 (1): BatchNorm2d(64, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)61 )62 )63 (stochastic_depth): StochasticDepth(p=0.010126582278481013, mode=row)64 )65 (1): FusedMBConv(66 (block): Sequential(67 (0): Conv2dNormActivation(68 (0): Conv2d(64, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)69 (1): BatchNorm2d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)70 (2): SiLU(inplace=True)71 )72 (1): Conv2dNormActivation(73 (0): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)74 (1): BatchNorm2d(64, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)75 )76 )77 (stochastic_depth): StochasticDepth(p=0.012658227848101266, mode=row)78 )79 (2): FusedMBConv(80 (block): Sequential(81 (0): Conv2dNormActivation(82 (0): Conv2d(64, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)83 (1): BatchNorm2d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)84 (2): SiLU(inplace=True)85 )86 (1): Conv2dNormActivation(87 (0): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)88 (1): BatchNorm2d(64, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)89 )90 )91 (stochastic_depth): StochasticDepth(p=0.015189873417721522, mode=row)92 )93 (3): FusedMBConv(94 (block): Sequential(95 (0): Conv2dNormActivation(96 (0): Conv2d(64, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)97 (1): BatchNorm2d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)98 (2): SiLU(inplace=True)99 )100 (1): Conv2dNormActivation(101 (0): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)102 (1): BatchNorm2d(64, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)103 )104 )105 (stochastic_depth): StochasticDepth(p=0.017721518987341773, mode=row)106 )107 (4): FusedMBConv(108 (block): Sequential(109 (0): Conv2dNormActivation(110 (0): Conv2d(64, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)111 (1): BatchNorm2d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)112 (2): SiLU(inplace=True)113 )114 (1): Conv2dNormActivation(115 (0): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)116 (1): BatchNorm2d(64, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)117 )118 )119 (stochastic_depth): StochasticDepth(p=0.020253164556962026, mode=row)120 )121 (5): FusedMBConv(122 (block): Sequential(123 (0): Conv2dNormActivation(124 (0): Conv2d(64, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)125 (1): BatchNorm2d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)126 (2): SiLU(inplace=True)127 )128 (1): Conv2dNormActivation(129 (0): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)130 (1): BatchNorm2d(64, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)131 )132 )133 (stochastic_depth): StochasticDepth(p=0.02278481012658228, mode=row)134 )135 (6): FusedMBConv(136 (block): Sequential(137 (0): Conv2dNormActivation(138 (0): Conv2d(64, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)139 (1): BatchNorm2d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)140 (2): SiLU(inplace=True)141 )142 (1): Conv2dNormActivation(143 (0): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)144 (1): BatchNorm2d(64, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)145 )146 )147 (stochastic_depth): StochasticDepth(p=0.02531645569620253, mode=row)148 )149 )150 (3): Sequential(151 (0): FusedMBConv(152 (block): Sequential(153 (0): Conv2dNormActivation(154 (0): Conv2d(64, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)155 (1): BatchNorm2d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)156 (2): SiLU(inplace=True)157 )158 (1): Conv2dNormActivation(159 (0): Conv2d(256, 96, kernel_size=(1, 1), stride=(1, 1), bias=False)160 (1): BatchNorm2d(96, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)161 )162 )163 (stochastic_depth): StochasticDepth(p=0.027848101265822787, mode=row)164 )165 (1): FusedMBConv(166 (block): Sequential(167 (0): Conv2dNormActivation(168 (0): Conv2d(96, 384, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)169 (1): BatchNorm2d(384, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)170 (2): SiLU(inplace=True)171 )172 (1): Conv2dNormActivation(173 (0): Conv2d(384, 96, kernel_size=(1, 1), stride=(1, 1), bias=False)174 (1): BatchNorm2d(96, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)175 )176 )177 (stochastic_depth): StochasticDepth(p=0.030379746835443044, mode=row)178 )179 (2): FusedMBConv(180 (block): Sequential(181 (0): Conv2dNormActivation(182 (0): Conv2d(96, 384, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)183 (1): BatchNorm2d(384, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)184 (2): SiLU(inplace=True)185 )186 (1): Conv2dNormActivation(187 (0): Conv2d(384, 96, kernel_size=(1, 1), stride=(1, 1), bias=False)188 (1): BatchNorm2d(96, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)189 )190 )191 (stochastic_depth): StochasticDepth(p=0.03291139240506329, mode=row)192 )193 (3): FusedMBConv(194 (block): Sequential(195 (0): Conv2dNormActivation(196 (0): Conv2d(96, 384, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)197 (1): BatchNorm2d(384, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)198 (2): SiLU(inplace=True)199 )200 (1): Conv2dNormActivation(201 (0): Conv2d(384, 96, kernel_size=(1, 1), stride=(1, 1), bias=False)202 (1): BatchNorm2d(96, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)203 )204 )205 (stochastic_depth): StochasticDepth(p=0.035443037974683546, mode=row)206 )207 (4): FusedMBConv(208 (block): Sequential(209 (0): Conv2dNormActivation(210 (0): Conv2d(96, 384, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)211 (1): BatchNorm2d(384, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)212 (2): SiLU(inplace=True)213 )214 (1): Conv2dNormActivation(215 (0): Conv2d(384, 96, kernel_size=(1, 1), stride=(1, 1), bias=False)216 (1): BatchNorm2d(96, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)217 )218 )219 (stochastic_depth): StochasticDepth(p=0.0379746835443038, mode=row)220 )221 (5): FusedMBConv(222 (block): Sequential(223 (0): Conv2dNormActivation(224 (0): Conv2d(96, 384, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)225 (1): BatchNorm2d(384, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)226 (2): SiLU(inplace=True)227 )228 (1): Conv2dNormActivation(229 (0): Conv2d(384, 96, kernel_size=(1, 1), stride=(1, 1), bias=False)230 (1): BatchNorm2d(96, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)231 )232 )233 (stochastic_depth): StochasticDepth(p=0.04050632911392405, mode=row)234 )235 (6): FusedMBConv(236 (block): Sequential(237 (0): Conv2dNormActivation(238 (0): Conv2d(96, 384, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)239 (1): BatchNorm2d(384, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)240 (2): SiLU(inplace=True)241 )242 (1): Conv2dNormActivation(243 (0): Conv2d(384, 96, kernel_size=(1, 1), stride=(1, 1), bias=False)244 (1): BatchNorm2d(96, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)245 )246 )247 (stochastic_depth): StochasticDepth(p=0.04303797468354431, mode=row)248 )249 )250 (4): Sequential(251 (0): MBConv(252 (block): Sequential(253 (0): Conv2dNormActivation(254 (0): Conv2d(96, 384, kernel_size=(1, 1), stride=(1, 1), bias=False)255 (1): BatchNorm2d(384, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)256 (2): SiLU(inplace=True)257 )258 (1): Conv2dNormActivation(259 (0): Conv2d(384, 384, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), groups=384, bias=False)260 (1): BatchNorm2d(384, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)261 (2): SiLU(inplace=True)262 )263 (2): SqueezeExcitation(264 (avgpool): AdaptiveAvgPool2d(output_size=1)265 (fc1): Conv2d(384, 24, kernel_size=(1, 1), stride=(1, 1))266 (fc2): Conv2d(24, 384, kernel_size=(1, 1), stride=(1, 1))267 (activation): SiLU(inplace=True)268 (scale_activation): Sigmoid()269 )270 (3): Conv2dNormActivation(271 (0): Conv2d(384, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)272 (1): BatchNorm2d(192, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)273 )274 )275 (stochastic_depth): StochasticDepth(p=0.04556962025316456, mode=row)276 )277 (1): MBConv(278 (block): Sequential(279 (0): Conv2dNormActivation(280 (0): Conv2d(192, 768, kernel_size=(1, 1), stride=(1, 1), bias=False)281 (1): BatchNorm2d(768, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)282 (2): SiLU(inplace=True)283 )284 (1): Conv2dNormActivation(285 (0): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768, bias=False)286 (1): BatchNorm2d(768, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)287 (2): SiLU(inplace=True)288 )289 (2): SqueezeExcitation(290 (avgpool): AdaptiveAvgPool2d(output_size=1)291 (fc1): Conv2d(768, 48, kernel_size=(1, 1), stride=(1, 1))292 (fc2): Conv2d(48, 768, kernel_size=(1, 1), stride=(1, 1))293 (activation): SiLU(inplace=True)294 (scale_activation): Sigmoid()295 )296 (3): Conv2dNormActivation(297 (0): Conv2d(768, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)298 (1): BatchNorm2d(192, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)299 )300 )301 (stochastic_depth): StochasticDepth(p=0.04810126582278482, mode=row)302 )303 (2): MBConv(304 (block): Sequential(305 (0): Conv2dNormActivation(306 (0): Conv2d(192, 768, kernel_size=(1, 1), stride=(1, 1), bias=False)307 (1): BatchNorm2d(768, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)308 (2): SiLU(inplace=True)309 )310 (1): Conv2dNormActivation(311 (0): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768, bias=False)312 (1): BatchNorm2d(768, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)313 (2): SiLU(inplace=True)314 )315 (2): SqueezeExcitation(316 (avgpool): AdaptiveAvgPool2d(output_size=1)317 (fc1): Conv2d(768, 48, kernel_size=(1, 1), stride=(1, 1))318 (fc2): Conv2d(48, 768, kernel_size=(1, 1), stride=(1, 1))319 (activation): SiLU(inplace=True)320 (scale_activation): Sigmoid()321 )322 (3): Conv2dNormActivation(323 (0): Conv2d(768, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)324 (1): BatchNorm2d(192, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)325 )326 )327 (stochastic_depth): StochasticDepth(p=0.05063291139240506, mode=row)328 )329 (3): MBConv(330 (block): Sequential(331 (0): Conv2dNormActivation(332 (0): Conv2d(192, 768, kernel_size=(1, 1), stride=(1, 1), bias=False)333 (1): BatchNorm2d(768, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)334 (2): SiLU(inplace=True)335 )336 (1): Conv2dNormActivation(337 (0): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768, bias=False)338 (1): BatchNorm2d(768, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)339 (2): SiLU(inplace=True)340 )341 (2): SqueezeExcitation(342 (avgpool): AdaptiveAvgPool2d(output_size=1)343 (fc1): Conv2d(768, 48, kernel_size=(1, 1), stride=(1, 1))344 (fc2): Conv2d(48, 768, kernel_size=(1, 1), stride=(1, 1))345 (activation): SiLU(inplace=True)346 (scale_activation): Sigmoid()347 )348 (3): Conv2dNormActivation(349 (0): Conv2d(768, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)350 (1): BatchNorm2d(192, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)351 )352 )353 (stochastic_depth): StochasticDepth(p=0.053164556962025315, mode=row)354 )355 (4): MBConv(356 (block): Sequential(357 (0): Conv2dNormActivation(358 (0): Conv2d(192, 768, kernel_size=(1, 1), stride=(1, 1), bias=False)359 (1): BatchNorm2d(768, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)360 (2): SiLU(inplace=True)361 )362 (1): Conv2dNormActivation(363 (0): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768, bias=False)364 (1): BatchNorm2d(768, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)365 (2): SiLU(inplace=True)366 )367 (2): SqueezeExcitation(368 (avgpool): AdaptiveAvgPool2d(output_size=1)369 (fc1): Conv2d(768, 48, kernel_size=(1, 1), stride=(1, 1))370 (fc2): Conv2d(48, 768, kernel_size=(1, 1), stride=(1, 1))371 (activation): SiLU(inplace=True)372 (scale_activation): Sigmoid()373 )374 (3): Conv2dNormActivation(375 (0): Conv2d(768, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)376 (1): BatchNorm2d(192, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)377 )378 )379 (stochastic_depth): StochasticDepth(p=0.055696202531645575, mode=row)380 )381 (5): MBConv(382 (block): Sequential(383 (0): Conv2dNormActivation(384 (0): Conv2d(192, 768, kernel_size=(1, 1), stride=(1, 1), bias=False)385 (1): BatchNorm2d(768, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)386 (2): SiLU(inplace=True)387 )388 (1): Conv2dNormActivation(389 (0): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768, bias=False)390 (1): BatchNorm2d(768, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)391 (2): SiLU(inplace=True)392 )393 (2): SqueezeExcitation(394 (avgpool): AdaptiveAvgPool2d(output_size=1)395 (fc1): Conv2d(768, 48, kernel_size=(1, 1), stride=(1, 1))396 (fc2): Conv2d(48, 768, kernel_size=(1, 1), stride=(1, 1))397 (activation): SiLU(inplace=True)398 (scale_activation): Sigmoid()399 )400 (3): Conv2dNormActivation(401 (0): Conv2d(768, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)402 (1): BatchNorm2d(192, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)403 )404 )405 (stochastic_depth): StochasticDepth(p=0.05822784810126583, mode=row)406 )407 (6): MBConv(408 (block): Sequential(409 (0): Conv2dNormActivation(410 (0): Conv2d(192, 768, kernel_size=(1, 1), stride=(1, 1), bias=False)411 (1): BatchNorm2d(768, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)412 (2): SiLU(inplace=True)413 )414 (1): Conv2dNormActivation(415 (0): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768, bias=False)416 (1): BatchNorm2d(768, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)417 (2): SiLU(inplace=True)418 )419 (2): SqueezeExcitation(420 (avgpool): AdaptiveAvgPool2d(output_size=1)421 (fc1): Conv2d(768, 48, kernel_size=(1, 1), stride=(1, 1))422 (fc2): Conv2d(48, 768, kernel_size=(1, 1), stride=(1, 1))423 (activation): SiLU(inplace=True)424 (scale_activation): Sigmoid()425 )426 (3): Conv2dNormActivation(427 (0): Conv2d(768, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)428 (1): BatchNorm2d(192, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)429 )430 )431 (stochastic_depth): StochasticDepth(p=0.06075949367088609, mode=row)432 )433 (7): MBConv(434 (block): Sequential(435 (0): Conv2dNormActivation(436 (0): Conv2d(192, 768, kernel_size=(1, 1), stride=(1, 1), bias=False)437 (1): BatchNorm2d(768, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)438 (2): SiLU(inplace=True)439 )440 (1): Conv2dNormActivation(441 (0): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768, bias=False)442 (1): BatchNorm2d(768, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)443 (2): SiLU(inplace=True)444 )445 (2): SqueezeExcitation(446 (avgpool): AdaptiveAvgPool2d(output_size=1)447 (fc1): Conv2d(768, 48, kernel_size=(1, 1), stride=(1, 1))448 (fc2): Conv2d(48, 768, kernel_size=(1, 1), stride=(1, 1))449 (activation): SiLU(inplace=True)450 (scale_activation): Sigmoid()451 )452 (3): Conv2dNormActivation(453 (0): Conv2d(768, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)454 (1): BatchNorm2d(192, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)455 )456 )457 (stochastic_depth): StochasticDepth(p=0.06329113924050633, mode=row)458 )459 (8): MBConv(460 (block): Sequential(461 (0): Conv2dNormActivation(462 (0): Conv2d(192, 768, kernel_size=(1, 1), stride=(1, 1), bias=False)463 (1): BatchNorm2d(768, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)464 (2): SiLU(inplace=True)465 )466 (1): Conv2dNormActivation(467 (0): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768, bias=False)468 (1): BatchNorm2d(768, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)469 (2): SiLU(inplace=True)470 )471 (2): SqueezeExcitation(472 (avgpool): AdaptiveAvgPool2d(output_size=1)473 (fc1): Conv2d(768, 48, kernel_size=(1, 1), stride=(1, 1))474 (fc2): Conv2d(48, 768, kernel_size=(1, 1), stride=(1, 1))475 (activation): SiLU(inplace=True)476 (scale_activation): Sigmoid()477 )478 (3): Conv2dNormActivation(479 (0): Conv2d(768, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)480 (1): BatchNorm2d(192, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)481 )482 )483 (stochastic_depth): StochasticDepth(p=0.06582278481012659, mode=row)484 )485 (9): MBConv(486 (block): Sequential(487 (0): Conv2dNormActivation(488 (0): Conv2d(192, 768, kernel_size=(1, 1), stride=(1, 1), bias=False)489 (1): BatchNorm2d(768, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)490 (2): SiLU(inplace=True)491 )492 (1): Conv2dNormActivation(493 (0): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768, bias=False)494 (1): BatchNorm2d(768, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)495 (2): SiLU(inplace=True)496 )497 (2): SqueezeExcitation(498 (avgpool): AdaptiveAvgPool2d(output_size=1)499 (fc1): Conv2d(768, 48, kernel_size=(1, 1), stride=(1, 1))500 (fc2): Conv2d(48, 768, kernel_size=(1, 1), stride=(1, 1))501 (activation): SiLU(inplace=True)502 (scale_activation): Sigmoid()503 )504 (3): Conv2dNormActivation(505 (0): Conv2d(768, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)506 (1): BatchNorm2d(192, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)507 )508 )509 (stochastic_depth): StochasticDepth(p=0.06835443037974684, mode=row)510 )511 )512 (5): Sequential(513 (0): MBConv(514 (block): Sequential(515 (0): Conv2dNormActivation(516 (0): Conv2d(192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False)517 (1): BatchNorm2d(1152, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)518 (2): SiLU(inplace=True)519 )520 (1): Conv2dNormActivation(521 (0): Conv2d(1152, 1152, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=1152, bias=False)522 (1): BatchNorm2d(1152, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)523 (2): SiLU(inplace=True)524 )525 (2): SqueezeExcitation(526 (avgpool): AdaptiveAvgPool2d(output_size=1)527 (fc1): Conv2d(1152, 48, kernel_size=(1, 1), stride=(1, 1))528 (fc2): Conv2d(48, 1152, kernel_size=(1, 1), stride=(1, 1))529 (activation): SiLU(inplace=True)530 (scale_activation): Sigmoid()531 )532 (3): Conv2dNormActivation(533 (0): Conv2d(1152, 224, kernel_size=(1, 1), stride=(1, 1), bias=False)534 (1): BatchNorm2d(224, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)535 )536 )537 (stochastic_depth): StochasticDepth(p=0.07088607594936709, mode=row)538 )539 (1): MBConv(540 (block): Sequential(541 (0): Conv2dNormActivation(542 (0): Conv2d(224, 1344, kernel_size=(1, 1), stride=(1, 1), bias=False)543 (1): BatchNorm2d(1344, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)544 (2): SiLU(inplace=True)545 )546 (1): Conv2dNormActivation(547 (0): Conv2d(1344, 1344, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=1344, bias=False)548 (1): BatchNorm2d(1344, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)549 (2): SiLU(inplace=True)550 )551 (2): SqueezeExcitation(552 (avgpool): AdaptiveAvgPool2d(output_size=1)553 (fc1): Conv2d(1344, 56, kernel_size=(1, 1), stride=(1, 1))554 (fc2): Conv2d(56, 1344, kernel_size=(1, 1), stride=(1, 1))555 (activation): SiLU(inplace=True)556 (scale_activation): Sigmoid()557 )558 (3): Conv2dNormActivation(559 (0): Conv2d(1344, 224, kernel_size=(1, 1), stride=(1, 1), bias=False)560 (1): BatchNorm2d(224, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)561 )562 )563 (stochastic_depth): StochasticDepth(p=0.07341772151898734, mode=row)564 )565 (2): MBConv(566 (block): Sequential(567 (0): Conv2dNormActivation(568 (0): Conv2d(224, 1344, kernel_size=(1, 1), stride=(1, 1), bias=False)569 (1): BatchNorm2d(1344, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)570 (2): SiLU(inplace=True)571 )572 (1): Conv2dNormActivation(573 (0): Conv2d(1344, 1344, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=1344, bias=False)574 (1): BatchNorm2d(1344, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)575 (2): SiLU(inplace=True)576 )577 (2): SqueezeExcitation(578 (avgpool): AdaptiveAvgPool2d(output_size=1)579 (fc1): Conv2d(1344, 56, kernel_size=(1, 1), stride=(1, 1))580 (fc2): Conv2d(56, 1344, kernel_size=(1, 1), stride=(1, 1))581 (activation): SiLU(inplace=True)582 (scale_activation): Sigmoid()583 )584 (3): Conv2dNormActivation(585 (0): Conv2d(1344, 224, kernel_size=(1, 1), stride=(1, 1), bias=False)586 (1): BatchNorm2d(224, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)587 )588 )589 (stochastic_depth): StochasticDepth(p=0.0759493670886076, mode=row)590 )591 (3): MBConv(592 (block): Sequential(593 (0): Conv2dNormActivation(594 (0): Conv2d(224, 1344, kernel_size=(1, 1), stride=(1, 1), bias=False)595 (1): BatchNorm2d(1344, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)596 (2): SiLU(inplace=True)597 )598 (1): Conv2dNormActivation(599 (0): Conv2d(1344, 1344, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=1344, bias=False)600 (1): BatchNorm2d(1344, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)601 (2): SiLU(inplace=True)602 )603 (2): SqueezeExcitation(604 (avgpool): AdaptiveAvgPool2d(output_size=1)605 (fc1): Conv2d(1344, 56, kernel_size=(1, 1), stride=(1, 1))606 (fc2): Conv2d(56, 1344, kernel_size=(1, 1), stride=(1, 1))607 (activation): SiLU(inplace=True)608 (scale_activation): Sigmoid()609 )610 (3): Conv2dNormActivation(611 (0): Conv2d(1344, 224, kernel_size=(1, 1), stride=(1, 1), bias=False)612 (1): BatchNorm2d(224, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)613 )614 )615 (stochastic_depth): StochasticDepth(p=0.07848101265822785, mode=row)616 )617 (4): MBConv(618 (block): Sequential(619 (0): Conv2dNormActivation(620 (0): Conv2d(224, 1344, kernel_size=(1, 1), stride=(1, 1), bias=False)621 (1): BatchNorm2d(1344, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)622 (2): SiLU(inplace=True)623 )624 (1): Conv2dNormActivation(625 (0): Conv2d(1344, 1344, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=1344, bias=False)626 (1): BatchNorm2d(1344, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)627 (2): SiLU(inplace=True)628 )629 (2): SqueezeExcitation(630 (avgpool): AdaptiveAvgPool2d(output_size=1)631 (fc1): Conv2d(1344, 56, kernel_size=(1, 1), stride=(1, 1))632 (fc2): Conv2d(56, 1344, kernel_size=(1, 1), stride=(1, 1))633 (activation): SiLU(inplace=True)634 (scale_activation): Sigmoid()635 )636 (3): Conv2dNormActivation(637 (0): Conv2d(1344, 224, kernel_size=(1, 1), stride=(1, 1), bias=False)638 (1): BatchNorm2d(224, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)639 )640 )641 (stochastic_depth): StochasticDepth(p=0.0810126582278481, mode=row)642 )643 (5): MBConv(644 (block): Sequential(645 (0): Conv2dNormActivation(646 (0): Conv2d(224, 1344, kernel_size=(1, 1), stride=(1, 1), bias=False)647 (1): BatchNorm2d(1344, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)648 (2): SiLU(inplace=True)649 )650 (1): Conv2dNormActivation(651 (0): Conv2d(1344, 1344, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=1344, bias=False)652 (1): BatchNorm2d(1344, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)653 (2): SiLU(inplace=True)654 )655 (2): SqueezeExcitation(656 (avgpool): AdaptiveAvgPool2d(output_size=1)657 (fc1): Conv2d(1344, 56, kernel_size=(1, 1), stride=(1, 1))658 (fc2): Conv2d(56, 1344, kernel_size=(1, 1), stride=(1, 1))659 (activation): SiLU(inplace=True)660 (scale_activation): Sigmoid()661 )662 (3): Conv2dNormActivation(663 (0): Conv2d(1344, 224, kernel_size=(1, 1), stride=(1, 1), bias=False)664 (1): BatchNorm2d(224, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)665 )666 )667 (stochastic_depth): StochasticDepth(p=0.08354430379746836, mode=row)668 )669 (6): MBConv(670 (block): Sequential(671 (0): Conv2dNormActivation(672 (0): Conv2d(224, 1344, kernel_size=(1, 1), stride=(1, 1), bias=False)673 (1): BatchNorm2d(1344, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)674 (2): SiLU(inplace=True)675 )676 (1): Conv2dNormActivation(677 (0): Conv2d(1344, 1344, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=1344, bias=False)678 (1): BatchNorm2d(1344, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)679 (2): SiLU(inplace=True)680 )681 (2): SqueezeExcitation(682 (avgpool): AdaptiveAvgPool2d(output_size=1)683 (fc1): Conv2d(1344, 56, kernel_size=(1, 1), stride=(1, 1))684 (fc2): Conv2d(56, 1344, kernel_size=(1, 1), stride=(1, 1))685 (activation): SiLU(inplace=True)686 (scale_activation): Sigmoid()687 )688 (3): Conv2dNormActivation(689 (0): Conv2d(1344, 224, kernel_size=(1, 1), stride=(1, 1), bias=False)690 (1): BatchNorm2d(224, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)691 )692 )693 (stochastic_depth): StochasticDepth(p=0.08607594936708862, mode=row)694 )695 (7): MBConv(696 (block): Sequential(697 (0): Conv2dNormActivation(698 (0): Conv2d(224, 1344, kernel_size=(1, 1), stride=(1, 1), bias=False)699 (1): BatchNorm2d(1344, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)700 (2): SiLU(inplace=True)701 )702 (1): Conv2dNormActivation(703 (0): Conv2d(1344, 1344, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=1344, bias=False)704 (1): BatchNorm2d(1344, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)705 (2): SiLU(inplace=True)706 )707 (2): SqueezeExcitation(708 (avgpool): AdaptiveAvgPool2d(output_size=1)709 (fc1): Conv2d(1344, 56, kernel_size=(1, 1), stride=(1, 1))710 (fc2): Conv2d(56, 1344, kernel_size=(1, 1), stride=(1, 1))711 (activation): SiLU(inplace=True)712 (scale_activation): Sigmoid()713 )714 (3): Conv2dNormActivation(715 (0): Conv2d(1344, 224, kernel_size=(1, 1), stride=(1, 1), bias=False)716 (1): BatchNorm2d(224, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)717 )718 )719 (stochastic_depth): StochasticDepth(p=0.08860759493670886, mode=row)720 )721 (8): MBConv(722 (block): Sequential(723 (0): Conv2dNormActivation(724 (0): Conv2d(224, 1344, kernel_size=(1, 1), stride=(1, 1), bias=False)725 (1): BatchNorm2d(1344, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)726 (2): SiLU(inplace=True)727 )728 (1): Conv2dNormActivation(729 (0): Conv2d(1344, 1344, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=1344, bias=False)730 (1): BatchNorm2d(1344, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)731 (2): SiLU(inplace=True)732 )733 (2): SqueezeExcitation(734 (avgpool): AdaptiveAvgPool2d(output_size=1)735 (fc1): Conv2d(1344, 56, kernel_size=(1, 1), stride=(1, 1))736 (fc2): Conv2d(56, 1344, kernel_size=(1, 1), stride=(1, 1))737 (activation): SiLU(inplace=True)738 (scale_activation): Sigmoid()739 )740 (3): Conv2dNormActivation(741 (0): Conv2d(1344, 224, kernel_size=(1, 1), stride=(1, 1), bias=False)742 (1): BatchNorm2d(224, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)743 )744 )745 (stochastic_depth): StochasticDepth(p=0.09113924050632911, mode=row)746 )747 (9): MBConv(748 (block): Sequential(749 (0): Conv2dNormActivation(750 (0): Conv2d(224, 1344, kernel_size=(1, 1), stride=(1, 1), bias=False)751 (1): BatchNorm2d(1344, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)752 (2): SiLU(inplace=True)753 )754 (1): Conv2dNormActivation(755 (0): Conv2d(1344, 1344, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=1344, bias=False)756 (1): BatchNorm2d(1344, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)757 (2): SiLU(inplace=True)758 )759 (2): SqueezeExcitation(760 (avgpool): AdaptiveAvgPool2d(output_size=1)761 (fc1): Conv2d(1344, 56, kernel_size=(1, 1), stride=(1, 1))762 (fc2): Conv2d(56, 1344, kernel_size=(1, 1), stride=(1, 1))763 (activation): SiLU(inplace=True)764 (scale_activation): Sigmoid()765 )766 (3): Conv2dNormActivation(767 (0): Conv2d(1344, 224, kernel_size=(1, 1), stride=(1, 1), bias=False)768 (1): BatchNorm2d(224, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)769 )770 )771 (stochastic_depth): StochasticDepth(p=0.09367088607594937, mode=row)772 )773 (10): MBConv(774 (block): Sequential(775 (0): Conv2dNormActivation(776 (0): Conv2d(224, 1344, kernel_size=(1, 1), stride=(1, 1), bias=False)777 (1): BatchNorm2d(1344, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)778 (2): SiLU(inplace=True)779 )780 (1): Conv2dNormActivation(781 (0): Conv2d(1344, 1344, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=1344, bias=False)782 (1): BatchNorm2d(1344, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)783 (2): SiLU(inplace=True)784 )785 (2): SqueezeExcitation(786 (avgpool): AdaptiveAvgPool2d(output_size=1)787 (fc1): Conv2d(1344, 56, kernel_size=(1, 1), stride=(1, 1))788 (fc2): Conv2d(56, 1344, kernel_size=(1, 1), stride=(1, 1))789 (activation): SiLU(inplace=True)790 (scale_activation): Sigmoid()791 )792 (3): Conv2dNormActivation(793 (0): Conv2d(1344, 224, kernel_size=(1, 1), stride=(1, 1), bias=False)794 (1): BatchNorm2d(224, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)795 )796 )797 (stochastic_depth): StochasticDepth(p=0.09620253164556963, mode=row)798 )799 (11): MBConv(800 (block): Sequential(801 (0): Conv2dNormActivation(802 (0): Conv2d(224, 1344, kernel_size=(1, 1), stride=(1, 1), bias=False)803 (1): BatchNorm2d(1344, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)804 (2): SiLU(inplace=True)805 )806 (1): Conv2dNormActivation(807 (0): Conv2d(1344, 1344, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=1344, bias=False)808 (1): BatchNorm2d(1344, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)809 (2): SiLU(inplace=True)810 )811 (2): SqueezeExcitation(812 (avgpool): AdaptiveAvgPool2d(output_size=1)813 (fc1): Conv2d(1344, 56, kernel_size=(1, 1), stride=(1, 1))814 (fc2): Conv2d(56, 1344, kernel_size=(1, 1), stride=(1, 1))815 (activation): SiLU(inplace=True)816 (scale_activation): Sigmoid()817 )818 (3): Conv2dNormActivation(819 (0): Conv2d(1344, 224, kernel_size=(1, 1), stride=(1, 1), bias=False)820 (1): BatchNorm2d(224, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)821 )822 )823 (stochastic_depth): StochasticDepth(p=0.09873417721518989, mode=row)824 )825 (12): MBConv(826 (block): Sequential(827 (0): Conv2dNormActivation(828 (0): Conv2d(224, 1344, kernel_size=(1, 1), stride=(1, 1), bias=False)829 (1): BatchNorm2d(1344, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)830 (2): SiLU(inplace=True)831 )832 (1): Conv2dNormActivation(833 (0): Conv2d(1344, 1344, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=1344, bias=False)834 (1): BatchNorm2d(1344, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)835 (2): SiLU(inplace=True)836 )837 (2): SqueezeExcitation(838 (avgpool): AdaptiveAvgPool2d(output_size=1)839 (fc1): Conv2d(1344, 56, kernel_size=(1, 1), stride=(1, 1))840 (fc2): Conv2d(56, 1344, kernel_size=(1, 1), stride=(1, 1))841 (activation): SiLU(inplace=True)842 (scale_activation): Sigmoid()843 )844 (3): Conv2dNormActivation(845 (0): Conv2d(1344, 224, kernel_size=(1, 1), stride=(1, 1), bias=False)846 (1): BatchNorm2d(224, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)847 )848 )849 (stochastic_depth): StochasticDepth(p=0.10126582278481013, mode=row)850 )851 (13): MBConv(852 (block): Sequential(853 (0): Conv2dNormActivation(854 (0): Conv2d(224, 1344, kernel_size=(1, 1), stride=(1, 1), bias=False)855 (1): BatchNorm2d(1344, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)856 (2): SiLU(inplace=True)857 )858 (1): Conv2dNormActivation(859 (0): Conv2d(1344, 1344, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=1344, bias=False)860 (1): BatchNorm2d(1344, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)861 (2): SiLU(inplace=True)862 )863 (2): SqueezeExcitation(864 (avgpool): AdaptiveAvgPool2d(output_size=1)865 (fc1): Conv2d(1344, 56, kernel_size=(1, 1), stride=(1, 1))866 (fc2): Conv2d(56, 1344, kernel_size=(1, 1), stride=(1, 1))867 (activation): SiLU(inplace=True)868 (scale_activation): Sigmoid()869 )870 (3): Conv2dNormActivation(871 (0): Conv2d(1344, 224, kernel_size=(1, 1), stride=(1, 1), bias=False)872 (1): BatchNorm2d(224, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)873 )874 )875 (stochastic_depth): StochasticDepth(p=0.10379746835443039, mode=row)876 )877 (14): MBConv(878 (block): Sequential(879 (0): Conv2dNormActivation(880 (0): Conv2d(224, 1344, kernel_size=(1, 1), stride=(1, 1), bias=False)881 (1): BatchNorm2d(1344, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)882 (2): SiLU(inplace=True)883 )884 (1): Conv2dNormActivation(885 (0): Conv2d(1344, 1344, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=1344, bias=False)886 (1): BatchNorm2d(1344, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)887 (2): SiLU(inplace=True)888 )889 (2): SqueezeExcitation(890 (avgpool): AdaptiveAvgPool2d(output_size=1)891 (fc1): Conv2d(1344, 56, kernel_size=(1, 1), stride=(1, 1))892 (fc2): Conv2d(56, 1344, kernel_size=(1, 1), stride=(1, 1))893 (activation): SiLU(inplace=True)894 (scale_activation): Sigmoid()895 )896 (3): Conv2dNormActivation(897 (0): Conv2d(1344, 224, kernel_size=(1, 1), stride=(1, 1), bias=False)898 (1): BatchNorm2d(224, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)899 )900 )901 (stochastic_depth): StochasticDepth(p=0.10632911392405063, mode=row)902 )903 (15): MBConv(904 (block): Sequential(905 (0): Conv2dNormActivation(906 (0): Conv2d(224, 1344, kernel_size=(1, 1), stride=(1, 1), bias=False)907 (1): BatchNorm2d(1344, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)908 (2): SiLU(inplace=True)909 )910 (1): Conv2dNormActivation(911 (0): Conv2d(1344, 1344, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=1344, bias=False)912 (1): BatchNorm2d(1344, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)913 (2): SiLU(inplace=True)914 )915 (2): SqueezeExcitation(916 (avgpool): AdaptiveAvgPool2d(output_size=1)917 (fc1): Conv2d(1344, 56, kernel_size=(1, 1), stride=(1, 1))918 (fc2): Conv2d(56, 1344, kernel_size=(1, 1), stride=(1, 1))919 (activation): SiLU(inplace=True)920 (scale_activation): Sigmoid()921 )922 (3): Conv2dNormActivation(923 (0): Conv2d(1344, 224, kernel_size=(1, 1), stride=(1, 1), bias=False)924 (1): BatchNorm2d(224, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)925 )926 )927 (stochastic_depth): StochasticDepth(p=0.10886075949367088, mode=row)928 )929 (16): MBConv(930 (block): Sequential(931 (0): Conv2dNormActivation(932 (0): Conv2d(224, 1344, kernel_size=(1, 1), stride=(1, 1), bias=False)933 (1): BatchNorm2d(1344, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)934 (2): SiLU(inplace=True)935 )936 (1): Conv2dNormActivation(937 (0): Conv2d(1344, 1344, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=1344, bias=False)938 (1): BatchNorm2d(1344, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)939 (2): SiLU(inplace=True)940 )941 (2): SqueezeExcitation(942 (avgpool): AdaptiveAvgPool2d(output_size=1)943 (fc1): Conv2d(1344, 56, kernel_size=(1, 1), stride=(1, 1))944 (fc2): Conv2d(56, 1344, kernel_size=(1, 1), stride=(1, 1))945 (activation): SiLU(inplace=True)946 (scale_activation): Sigmoid()947 )948 (3): Conv2dNormActivation(949 (0): Conv2d(1344, 224, kernel_size=(1, 1), stride=(1, 1), bias=False)950 (1): BatchNorm2d(224, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)951 )952 )953 (stochastic_depth): StochasticDepth(p=0.11139240506329115, mode=row)954 )955 (17): MBConv(956 (block): Sequential(957 (0): Conv2dNormActivation(958 (0): Conv2d(224, 1344, kernel_size=(1, 1), stride=(1, 1), bias=False)959 (1): BatchNorm2d(1344, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)960 (2): SiLU(inplace=True)961 )962 (1): Conv2dNormActivation(963 (0): Conv2d(1344, 1344, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=1344, bias=False)964 (1): BatchNorm2d(1344, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)965 (2): SiLU(inplace=True)966 )967 (2): SqueezeExcitation(968 (avgpool): AdaptiveAvgPool2d(output_size=1)969 (fc1): Conv2d(1344, 56, kernel_size=(1, 1), stride=(1, 1))970 (fc2): Conv2d(56, 1344, kernel_size=(1, 1), stride=(1, 1))971 (activation): SiLU(inplace=True)972 (scale_activation): Sigmoid()973 )974 (3): Conv2dNormActivation(975 (0): Conv2d(1344, 224, kernel_size=(1, 1), stride=(1, 1), bias=False)976 (1): BatchNorm2d(224, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)977 )978 )979 (stochastic_depth): StochasticDepth(p=0.11392405063291139, mode=row)980 )981 (18): MBConv(982 (block): Sequential(983 (0): Conv2dNormActivation(984 (0): Conv2d(224, 1344, kernel_size=(1, 1), stride=(1, 1), bias=False)985 (1): BatchNorm2d(1344, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)986 (2): SiLU(inplace=True)987 )988 (1): Conv2dNormActivation(989 (0): Conv2d(1344, 1344, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=1344, bias=False)990 (1): BatchNorm2d(1344, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)991 (2): SiLU(inplace=True)992 )993 (2): SqueezeExcitation(994 (avgpool): AdaptiveAvgPool2d(output_size=1)995 (fc1): Conv2d(1344, 56, kernel_size=(1, 1), stride=(1, 1))996 (fc2): Conv2d(56, 1344, kernel_size=(1, 1), stride=(1, 1))997 (activation): SiLU(inplace=True)998 (scale_activation): Sigmoid()999 )1000 (3): Conv2dNormActivation(1001 (0): Conv2d(1344, 224, kernel_size=(1, 1), stride=(1, 1), bias=False)1002 (1): BatchNorm2d(224, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)1003 )1004 )1005 (stochastic_depth): StochasticDepth(p=0.11645569620253166, mode=row)1006 )1007 )1008 (6): Sequential(1009 (0): MBConv(1010 (block): Sequential(1011 (0): Conv2dNormActivation(1012 (0): Conv2d(224, 1344, kernel_size=(1, 1), stride=(1, 1), bias=False)1013 (1): BatchNorm2d(1344, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)1014 (2): SiLU(inplace=True)1015 )1016 (1): Conv2dNormActivation(1017 (0): Conv2d(1344, 1344, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), groups=1344, bias=False)1018 (1): BatchNorm2d(1344, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)1019 (2): SiLU(inplace=True)1020 )1021 (2): SqueezeExcitation(1022 (avgpool): AdaptiveAvgPool2d(output_size=1)1023 (fc1): Conv2d(1344, 56, kernel_size=(1, 1), stride=(1, 1))1024 (fc2): Conv2d(56, 1344, kernel_size=(1, 1), stride=(1, 1))1025 (activation): SiLU(inplace=True)1026 (scale_activation): Sigmoid()1027 )1028 (3): Conv2dNormActivation(1029 (0): Conv2d(1344, 384, kernel_size=(1, 1), stride=(1, 1), bias=False)1030 (1): BatchNorm2d(384, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)1031 )1032 )1033 (stochastic_depth): StochasticDepth(p=0.11898734177215191, mode=row)1034 )1035 (1): MBConv(1036 (block): Sequential(1037 (0): Conv2dNormActivation(1038 (0): Conv2d(384, 2304, kernel_size=(1, 1), stride=(1, 1), bias=False)1039 (1): BatchNorm2d(2304, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)1040 (2): SiLU(inplace=True)1041 )1042 (1): Conv2dNormActivation(1043 (0): Conv2d(2304, 2304, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=2304, bias=False)1044 (1): BatchNorm2d(2304, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)1045 (2): SiLU(inplace=True)1046 )1047 (2): SqueezeExcitation(1048 (avgpool): AdaptiveAvgPool2d(output_size=1)1049 (fc1): Conv2d(2304, 96, kernel_size=(1, 1), stride=(1, 1))1050 (fc2): Conv2d(96, 2304, kernel_size=(1, 1), stride=(1, 1))1051 (activation): SiLU(inplace=True)1052 (scale_activation): Sigmoid()1053 )1054 (3): Conv2dNormActivation(1055 (0): Conv2d(2304, 384, kernel_size=(1, 1), stride=(1, 1), bias=False)1056 (1): BatchNorm2d(384, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)1057 )1058 )1059 (stochastic_depth): StochasticDepth(p=0.12151898734177217, mode=row)1060 )1061 (2): MBConv(1062 (block): Sequential(1063 (0): Conv2dNormActivation(1064 (0): Conv2d(384, 2304, kernel_size=(1, 1), stride=(1, 1), bias=False)1065 (1): BatchNorm2d(2304, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)1066 (2): SiLU(inplace=True)1067 )1068 (1): Conv2dNormActivation(1069 (0): Conv2d(2304, 2304, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=2304, bias=False)1070 (1): BatchNorm2d(2304, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)1071 (2): SiLU(inplace=True)1072 )1073 (2): SqueezeExcitation(1074 (avgpool): AdaptiveAvgPool2d(output_size=1)1075 (fc1): Conv2d(2304, 96, kernel_size=(1, 1), stride=(1, 1))1076 (fc2): Conv2d(96, 2304, kernel_size=(1, 1), stride=(1, 1))1077 (activation): SiLU(inplace=True)1078 (scale_activation): Sigmoid()1079 )1080 (3): Conv2dNormActivation(1081 (0): Conv2d(2304, 384, kernel_size=(1, 1), stride=(1, 1), bias=False)1082 (1): BatchNorm2d(384, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)1083 )1084 )1085 (stochastic_depth): StochasticDepth(p=0.12405063291139241, mode=row)1086 )1087 (3): MBConv(1088 (block): Sequential(1089 (0): Conv2dNormActivation(1090 (0): Conv2d(384, 2304, kernel_size=(1, 1), stride=(1, 1), bias=False)1091 (1): BatchNorm2d(2304, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)1092 (2): SiLU(inplace=True)1093 )1094 (1): Conv2dNormActivation(1095 (0): Conv2d(2304, 2304, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=2304, bias=False)1096 (1): BatchNorm2d(2304, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)1097 (2): SiLU(inplace=True)1098 )1099 (2): SqueezeExcitation(1100 (avgpool): AdaptiveAvgPool2d(output_size=1)1101 (fc1): Conv2d(2304, 96, kernel_size=(1, 1), stride=(1, 1))1102 (fc2): Conv2d(96, 2304, kernel_size=(1, 1), stride=(1, 1))1103 (activation): SiLU(inplace=True)1104 (scale_activation): Sigmoid()1105 )1106 (3): Conv2dNormActivation(1107 (0): Conv2d(2304, 384, kernel_size=(1, 1), stride=(1, 1), bias=False)1108 (1): BatchNorm2d(384, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)1109 )1110 )1111 (stochastic_depth): StochasticDepth(p=0.12658227848101267, mode=row)1112 )1113 (4): MBConv(1114 (block): Sequential(1115 (0): Conv2dNormActivation(1116 (0): Conv2d(384, 2304, kernel_size=(1, 1), stride=(1, 1), bias=False)1117 (1): BatchNorm2d(2304, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)1118 (2): SiLU(inplace=True)1119 )1120 (1): Conv2dNormActivation(1121 (0): Conv2d(2304, 2304, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=2304, bias=False)1122 (1): BatchNorm2d(2304, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)1123 (2): SiLU(inplace=True)1124 )1125 (2): SqueezeExcitation(1126 (avgpool): AdaptiveAvgPool2d(output_size=1)1127 (fc1): Conv2d(2304, 96, kernel_size=(1, 1), stride=(1, 1))1128 (fc2): Conv2d(96, 2304, kernel_size=(1, 1), stride=(1, 1))1129 (activation): SiLU(inplace=True)1130 (scale_activation): Sigmoid()1131 )1132 (3): Conv2dNormActivation(1133 (0): Conv2d(2304, 384, kernel_size=(1, 1), stride=(1, 1), bias=False)1134 (1): BatchNorm2d(384, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)1135 )1136 )1137 (stochastic_depth): StochasticDepth(p=0.12911392405063293, mode=row)1138 )1139 (5): MBConv(1140 (block): Sequential(1141 (0): Conv2dNormActivation(1142 (0): Conv2d(384, 2304, kernel_size=(1, 1), stride=(1, 1), bias=False)1143 (1): BatchNorm2d(2304, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)1144 (2): SiLU(inplace=True)1145 )1146 (1): Conv2dNormActivation(1147 (0): Conv2d(2304, 2304, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=2304, bias=False)1148 (1): BatchNorm2d(2304, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)1149 (2): SiLU(inplace=True)1150 )1151 (2): SqueezeExcitation(1152 (avgpool): AdaptiveAvgPool2d(output_size=1)1153 (fc1): Conv2d(2304, 96, kernel_size=(1, 1), stride=(1, 1))1154 (fc2): Conv2d(96, 2304, kernel_size=(1, 1), stride=(1, 1))1155 (activation): SiLU(inplace=True)1156 (scale_activation): Sigmoid()1157 )1158 (3): Conv2dNormActivation(1159 (0): Conv2d(2304, 384, kernel_size=(1, 1), stride=(1, 1), bias=False)1160 (1): BatchNorm2d(384, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)1161 )1162 )1163 (stochastic_depth): StochasticDepth(p=0.13164556962025317, mode=row)1164 )1165 (6): MBConv(1166 (block): Sequential(1167 (0): Conv2dNormActivation(1168 (0): Conv2d(384, 2304, kernel_size=(1, 1), stride=(1, 1), bias=False)1169 (1): BatchNorm2d(2304, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)1170 (2): SiLU(inplace=True)1171 )1172 (1): Conv2dNormActivation(1173 (0): Conv2d(2304, 2304, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=2304, bias=False)1174 (1): BatchNorm2d(2304, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)1175 (2): SiLU(inplace=True)1176 )1177 (2): SqueezeExcitation(1178 (avgpool): AdaptiveAvgPool2d(output_size=1)1179 (fc1): Conv2d(2304, 96, kernel_size=(1, 1), stride=(1, 1))1180 (fc2): Conv2d(96, 2304, kernel_size=(1, 1), stride=(1, 1))1181 (activation): SiLU(inplace=True)1182 (scale_activation): Sigmoid()1183 )1184 (3): Conv2dNormActivation(1185 (0): Conv2d(2304, 384, kernel_size=(1, 1), stride=(1, 1), bias=False)1186 (1): BatchNorm2d(384, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)1187 )1188 )1189 (stochastic_depth): StochasticDepth(p=0.13417721518987344, mode=row)1190 )1191 (7): MBConv(1192 (block): Sequential(1193 (0): Conv2dNormActivation(1194 (0): Conv2d(384, 2304, kernel_size=(1, 1), stride=(1, 1), bias=False)1195 (1): BatchNorm2d(2304, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)1196 (2): SiLU(inplace=True)1197 )1198 (1): Conv2dNormActivation(1199 (0): Conv2d(2304, 2304, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=2304, bias=False)1200 (1): BatchNorm2d(2304, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)