U ʹj[u@sBdZddlZddlmZddlmZmZmZmZm Z m Z m Z ddl Z ddl mZddlmmZddlmZmZddlmZmZmZmZmZmZmZmZmZmZm Z ddl!m"Z"dd l#m$Z$dd l%m&Z&m'Z'm(Z(d d d gZ)de*e*e*e*dddZ+de ej,e*e*e-ej,dddZ.Gdd d ej,Z/Gdd d ej,Z0de*e*e*e*e*ee*ee ej,ej,dddZ1de*e*e*e*e*ee*ee ej,ej,dddZ2de3ddd Z4de e/e0fee*ee*e*e*e*e*e-e3e3e ee e5ej,feee5effd# d$d%Z6Gd&d d ej,Z7de-e7d'd(d)Z8dd+d,Z9dd-d.Z:dd/d0Z;dd1d2Z<dd3d4Z=dd5d6Z>e'e;d7d8d9d:d;de;d7d?d9d:d;ded7ddAe>d7ddAe>d7ddAe>d7ddAe>d7ddAe>d7dd=dEe>d7dd=dEe>d7dd=dEe>d7dd=dEe>d7dd=dEe>d7dd=dEe>d7dd=dEe>d7dd=dEe>d7dd=dEe>d7ddAe>d7ddAe>d7ddAe>d7ddAe>d7ddAe>d7ddAe>d7dd=dEdZ?e&de-e7d'ddZ@e&de-e7d'ddZAe&de-e7d'ddZBe&de-e7d'ddZCe&de-e7d'ddZDe&de-e7d'ddZEe&de-e7d'ddZFe&de-e7d'ddZGe&de-e7d'ddZHe&de-e7d'ddZIe&de-e7d'd d ZJe&de-e7d'd d ZKe&de-e7d'd dZLe&de-e7d'ddZMe&de-e7d'ddZNe&de-e7d'ddZOe&de-e7d'ddZPe&de-e7d'ddZQe&de-e7d'ddZRe&de-e7d'ddZSe&de-e7d'ddZTe&de-e7d'dd ZUe&de-e7d'd!d"ZVe&de-e7d'd#d$ZWe&de-e7d'd%d&ZXe&de-e7d'd'd(ZYe&de-e7d'd)d*ZZe&de-e7d'd+d,Z[e&de-e7d'd-d.Z\e&de-e7d'd/d0Z]e&de-e7d'd1d2Z^e&de-e7d'd3d4Z_e&de-e7d'd5d6Z`e&de-e7d'd7d8Zae&de-e7d'd9d:Zbe&de-e7d'd;d<Zce&de-e7d'd=d>Zde&de-e7d'd?d@Zee&de-e7d'dAdBZfe&de-e7d'dCdDZge&de-e7d'dEdFZhe&de-e7d'dGdHZie&de-e7d'dIdJZje&de-e7d'dKdLZke&de-e7d'dMdNZle&de-e7d'dOdPZme&de-e7d'dQdRZne&de-e7d'dSdTZoe&de-e7d'dUdVZpe&de-e7d'dWdXZqe&de-e7d'dYdZZre&de-e7d'd[d\Zse&de-e7d'd]d^Zte&de-e7d'd_d`Zue&de-e7d'dadbZve&de-e7d'dcddZwe&de-e7d'dedfZxe&de-e7d'dgdhZye&de-e7d'didjZze&de-e7d'dkdlZ{e&de-e7d'dmdnZ|e&de-e7d'dodpZ}e&d e-e7d'dqdrZ~e&d e-e7d'dsdtZe&d e-e7d'dudvZe&d e-e7d'dwdxZe&d e-e7d'dydzZe&de-e7d'd{d|Ze&de-e7d'd}d~Ze&de-e7d'ddZe&de-e7d'ddZe&de-e7d'ddZe&de-dddZe&de-e7d'ddZe&de-e7d'ddZe&de-e7d'ddZe&de-e7d'ddZe&de-e7d'ddZe&de-e7d'ddZe&de-e7d'ddZe(eddddddddddddddddddddddddddddddddddddddddddddd+dS(a*PyTorch ResNet This started as a copy of https://github.com/pytorch/vision 'resnet.py' (BSD-3-Clause) with additional dropout and dynamic global avg/max pool. ResNeXt, SE-ResNeXt, SENet, and MXNet Gluon stem/downsample variants, tiered stems added by Ross Wightman Copyright 2019, Ross Wightman N)partial)AnyDictListOptionalTupleTypeUnionIMAGENET_DEFAULT_MEANIMAGENET_DEFAULT_STD) DropBlock2dDropPath AvgPool2dSame BlurPool2d GroupNorm LayerType create_attnget_attn get_act_layerget_norm_layercreate_classifier)build_model_with_cfg)checkpoint_seq)register_modelgenerate_default_cfgsregister_model_deprecationsResNet BasicBlock Bottleneck) kernel_sizestridedilationreturncCs|d||dd}|S)Nr)r!r"r#paddingr&r&Q/var/www/html/bwave/python/venv/lib/python3.8/site-packages/timm/models/resnet.py get_paddingsr)r%T)aa_layerchannelsr"enabler$cCs4|r|stSt|tjr$||S|||dSdS)Nr+r")nnIdentity issubclass AvgPool2d)r*r+r"r,r&r&r( create_aa!s  r2cseZdZdZdddddddejejddddf eeeeej eeeeeee ej e ej ee ej ee ej ee ej eej dfdd Z ddZ e je jd d d ZZS) rrN@inplanesplanesr" downsample cardinality base_width reduce_firstr#first_dilation act_layer norm_layer attn_layerr* drop_block drop_pathc stt||dkstd|dks.td||}||j}| pF|} | dk o^|dkp^| |k}tj||d|rrdn|| | dd |_| ||_|dk r|nt |_ | d d |_ t | |||d |_ tj||d||dd |_| ||_t| ||_| d d |_||_||_||_||_dS) Args: inplanes: Input channel dimensionality. planes: Used to determine output channel dimensionalities. stride: Stride used in convolution layers. downsample: Optional downsample layer for residual path. cardinality: Number of convolution groups. base_width: Base width used to determine output channel dimensionality. reduce_first: Reduction factor for first convolution output width of residual blocks. dilation: Dilation rate for convolution layers. first_dilation: Dilation rate for first convolution layer. act_layer: Activation layer. norm_layer: Normalization layer. attn_layer: Attention layer. aa_layer: Anti-aliasing layer. drop_block: Class for DropBlock layer. drop_path: Optional DropPath layer. rz)BasicBlock only supports cardinality of 1r3z/BasicBlock does not support changing base widthNr%F)r!r"r'r#biasTZinplacer+r"r,)r!r'r#rC)superr__init__AssertionError expansionr.Conv2dconv1bn1r/r?act1r2aaconv2bn2rseact2r7r"r#r@)selfr5r6r"r7r8r9r:r#r;r<r=r>r*r?r@ first_planes outplanesuse_aa __class__r&r(rG-sF$       zBasicBlock.__init__cCs&t|jdddk r"tj|jjdSNweight)getattrrPr.initzeros_rZrSr&r&r(zero_init_lastnszBasicBlock.zero_init_lastxr$cCs|}||}||}||}||}||}||}||}|jdk r^||}|jdk rr||}|j dk r| |}||7}| |}|SN) rKrLr?rMrNrOrPrQr@r7rRrSraZshortcutr&r&r(forwardrs"              zBasicBlock.forward__name__ __module__ __qualname__rIr.ReLU BatchNorm2dintrModulerrGr_torchTensorrd __classcell__r&r&rWr(r*sB   AcseZdZdZdddddddejejddddf eeeeej eeeeeee ej e ej ee ej ee ej ee ej eej dfdd Z dd Z e je jd d d ZZS) r rNr3r4c s:tt|tt||d|}||}||j}| p@|} | dk oX|dkpX| |k}tj||ddd|_ | ||_ | dd|_ tj||d |rdn|| | |dd |_ | ||_ |dk r|nt|_| dd|_t| |||d |_tj||ddd|_| ||_t| ||_| dd|_||_||_||_||_dS) rAr3Nr%rF)r!rCTrDrB)r!r"r'r#groupsrCrE)rFr rGrkmathfloorrIr.rJrKrLrMrOrPr/r?rRr2rNconv3bn3rrQact3r7r"r#r@)rSr5r6r"r7r8r9r:r#r;r<r=r>r*r?r@widthrTrUrVrWr&r(rGs>$         zBottleneck.__init__cCs&t|jdddk r"tj|jjdSrY)r[rur.r\r]rZr^r&r&r(r_szBottleneck.zero_init_lastr`cCs|}||}||}||}||}||}||}||}||}||}| |}|j dk r|| |}|j dk r| |}|j dk r| |}||7}| |}|Srb)rKrLrMrOrPr?rRrNrtrurQr@r7rvrcr&r&r(rds(                 zBottleneck.forwardrer&r&rWr(r sB   C) in_channels out_channelsr!r"r#r;r=r$c Csh|ptj}|dkr|dkrdn|}|dkr2|p4|nd}t|||}tjtj||||||dd||gS)NrF)r"r'r#rC)r.rjr) SequentialrJ)rxryr!r"r#r;r=pr&r&r(downsample_convs  r|c Cs|ptj}|dkr|nd}|dkr4|dkr4t}n*|dkrH|dkrHtntj} | d|ddd}tj|tj||ddddd||gS)Nrr%TF)Z ceil_modeZcount_include_padrr"r'rC)r.rjr/rr1rzrJ) rxryr!r"r#r;r=Z avg_stridepoolZ avg_pool_fnr&r&r(downsample_avg s  r drop_probc Cs4dd|rtt|dddnd|r.tt|dddndgS)N?)r block_sizeZ gamma_scalerB?)rr rr&r&r( drop_blocks!s r F) block_fnr+ block_repeatsr5r: output_stridedown_kernel_sizeavg_downdrop_block_ratedrop_path_rater$c  Ksg} g} t|} d}d}d}}tt||t|D]l\}\}}}d|d}|dkr^dnd}||krx||9}d}n||9}d}|dks|||jkrt|||j||||| dd}|rtf|ntf|}tf|||d| }g}t |D]}|dkr|nd}|dkr|nd}| || d}| |||||f||d krRt |ndd ||}||j}|d7}q| |t j |f| t|||d q4| | fS) Nrrprlayerr%r=)rxryr!r"r#r;r=)r:r#r?r)r;r@Znum_chsZ reductionmodule)sum enumerateziprrIdictgetrr|rangeappendrr.rz)rr+rr5r:rrrrrkwargsZstages feature_infoZnet_num_blocksZ net_block_idxZ net_strider#Z prev_dilationZ stage_idxr6Z num_blocksdbZ stage_namer"r7Z down_kwargsZ block_kwargsblocksZ block_idxZ block_dprr&r&r( make_blocks(s` $     rcsJeZdZdZddddddddd ddd ejejd d d d d d feee fe e e e e e e e e e e e e e eeeeejeeee eee efd fdd Zejjd&e dddZejjd'e dddZejjd(e dddZejjd)e dddZd*ddZejejddd Zd+eje ejd!d"d#Z ejejdd$d%Z!Z"S),raResNet / ResNeXt / SE-ResNeXt / SE-Net This class implements all variants of ResNet, ResNeXt, SE-ResNeXt, and SENet that * have > 1 stride in the 3x3 conv layer of bottleneck * have conv-bn-act ordering This ResNet impl supports a number of stem and downsample options based on the v1c, v1d, v1e, and v1s variants included in the MXNet Gluon ResNetV1b model. The C and D variants are also discussed in the 'Bag of Tricks' paper: https://arxiv.org/pdf/1812.01187. The B variant is equivalent to torchvision default. ResNet variants (the same modifications can be used in SE/ResNeXt models as well): * normal, b - 7x7 stem, stem_width = 64, same as torchvision ResNet, NVIDIA ResNet 'v1.5', Gluon v1b * c - 3 layer deep 3x3 stem, stem_width = 32 (32, 32, 64) * d - 3 layer deep 3x3 stem, stem_width = 32 (32, 32, 64), average pool in downsample * e - 3 layer deep 3x3 stem, stem_width = 64 (64, 64, 128), average pool in downsample * s - 3 layer deep 3x3 stem, stem_width = 64 (64, 64, 128) * t - 3 layer deep 3x3 stem, stem width = 32 (24, 48, 64), average pool in downsample * tn - 3 layer deep 3x3 stem, stem width = 32 (24, 32, 64), average pool in downsample ResNeXt * normal - 7x7 stem, stem_width = 64, standard cardinality and base widths * same c,d, e, s variants as ResNet can be enabled SE-ResNeXt * normal - 7x7 stem, stem_width = 64 * same c, d, e, s variants as ResNet can be enabled SENet-154 - 3 layer deep 3x3 stem (same as v1c-v1s), stem_width = 64, cardinality=64, reduction by 2 on width of first bottleneck convolution, 3x3 downsample convs after first block rBravgrr3FNrT)blocklayers num_classesin_chansr global_poolr8r9 stem_width stem_typereplace_stem_poolblock_reduce_firstrrr<r=r* drop_raterrr_ block_argscstt||pt}|dks$t||_||_d|_t|}t |}d| k}|rZ| dnd}|r| | f}d| krd| d| f}t j t j ||d ddd dd ||d |d d t j |d |d dd d dd ||d |d d t j |d |dd d dd g|_ nt j ||ddddd|_ |||_|d d |_t|dddg|_| rt j tdt j ||d|rdd ndd dd |dk rt||ddnd|||d d g|_n\|dk rt|t jr|d|_n$t j t jdd d d||ddg|_nt jddd d|_ddddg}t||||f|||| || |||||d |\}}|D]}|j|q@|j|d|j|_t|j|j|d\|_|_|j|ddS)a Args: block (nn.Module): class for the residual block. Options are BasicBlock, Bottleneck. layers (List[int]) : number of layers in each block num_classes (int): number of classification classes (default 1000) in_chans (int): number of input (color) channels. (default 3) output_stride (int): output stride of the network, 32, 16, or 8. (default 32) global_pool (str): Global pooling type. One of 'avg', 'max', 'avgmax', 'catavgmax' (default 'avg') cardinality (int): number of convolution groups for 3x3 conv in Bottleneck. (default 1) base_width (int): bottleneck channels factor. `planes * base_width / 64 * cardinality` (default 64) stem_width (int): number of channels in stem convolutions (default 64) stem_type (str): The type of stem (default ''): * '', default - a single 7x7 conv with a width of stem_width * 'deep' - three 3x3 convolution layers of widths stem_width, stem_width, stem_width * 2 * 'deep_tiered' - three 3x3 conv layers of widths stem_width//4 * 3, stem_width, stem_width * 2 replace_stem_pool (bool): replace stem max-pooling layer with a 3x3 stride-2 convolution block_reduce_first (int): Reduction factor for first convolution output width of residual blocks, 1 for all archs except senets, where 2 (default 1) down_kernel_size (int): kernel size of residual block downsample path, 1x1 for most, 3x3 for senets (default: 1) avg_down (bool): use avg pooling for projection skip connection between stages/downsample (default False) act_layer (str, nn.Module): activation layer norm_layer (str, nn.Module): normalization layer aa_layer (nn.Module): anti-aliasing layer drop_rate (float): Dropout probability before classifier, for training (default 0.) drop_path_rate (float): Stochastic depth drop-path rate (default 0.) drop_block_rate (float): Drop block rate (default 0.) zero_init_last (bool): zero-init the last weight in residual path (usually last BN affine weight) block_args (dict): Extra kwargs to pass through to block module )rFdeepr%r3ZtieredrBrprrr}TrD)r!r"r'rCrMrNr-)r!r"r'i) r8r9rr:rrr<r=r*rrZ pool_typer_) rFrrGrrHrrgrad_checkpointingrrr.rzrJrKrLrMrfilterr2maxpoolr0r1Z MaxPool2drZ add_moduleextendrI num_featuresrrfc init_weights)rSrrrrrrr8r9rrrrrrr<r=r*rrrr_rZ deep_stemr5Zstem_chsr+Z stage_modulesZstage_feature_infoZstagerWr&r(rGs7               zResNet.__init__rcCsZ|D](\}}t|tjrtjj|jdddq|rV|D]}t|dr>| q>dS)NZfan_outZrelu)modeZ nonlinearityr_) Z named_modules isinstancer.rJr\Zkaiming_normal_rZmoduleshasattrr_)rSr_nmr&r&r(rs   zResNet.init_weights)coarsecCstd|r dndd}|S)Nz^conv1|bn1|maxpoolz ^layer(\d+)z^layer(\d+)\.(\d+))stemr)r)rSrZmatcherr&r&r( group_matcherszResNet.group_matcher)r,cCs ||_dSrb)r)rSr,r&r&r(set_grad_checkpointing szResNet.set_grad_checkpointing) name_onlycCs|rdS|jS)Nr)r)rSrr&r&r(get_classifier$szResNet.get_classifiercCs$||_t|j|j|d\|_|_dS)Nr)rrrrr)rSrrr&r&r(reset_classifier(szResNet.reset_classifierr`cCs||}||}||}||}|jrXtjsXt|j |j |j |j g|dd}n(| |}| |}| |}| |}|S)NT)flatten) rKrLrMrrrmjitZ is_scriptingrZlayer1Zlayer2Zlayer3Zlayer4rSrar&r&r(forward_features,s         zResNet.forward_features)ra pre_logitsr$cCs:||}|jr(tj|t|j|jd}|r0|S||S)N)r{training)rrFZdropoutfloatrr)rSrarr&r&r( forward_head;s zResNet.forward_headcCs||}||}|Srb)rrrr&r&r(rdAs  zResNet.forward)T)F)T)F)r)F)#rfrgrh__doc__r.rirjr rr rrkstrboolrrrrlrrrrGrmrignorerrrrrrnrrrdror&r&rWr(rjst#    ) pretrainedr$cKstt||f|Srb)rr)variantrrr&r&r(_create_resnetGsrrc Ks|dddddttddd |S) NrrBr)rrg?ZbilinearrKr) urlr input_size pool_sizecrop_pct interpolationZmeanZstd first_conv classifierr rrr&r&r(_cfgKsrcKstfd|itddif|S)Nrrbicubicrrrr&r&r(_tcfgVsrc Ks$tfd|itdddddf|S)NrrrB rffffff?3https://github.com/huggingface/pytorch-image-models)rtest_input_size test_crop_pct origin_urlrrr&r&r(_ttcfgZs rc Ks(tfd|itdddddddf|S) NrrrrrrarXiv:2110.00476)rrrrr paper_idsrrr&r&r(_rcfgas rc Ks,tfd|itddddddddd f|S) NrrrBrrrrrrr)rrrrrrrrrrr&r&r(_r3cfghs rcKs tfd|itdddf|S)Nrrz1https://cv.gluon.ai/model_zoo/classification.html)rrrrr&r&r(_gcfgps  rztimm/zrhttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-rsb-weights/resnet10t_176_c3-f3215ab1.pth)rBr)rrrzconv1.0) hf_hub_idrrrrrrzrhttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-rsb-weights/resnet14t_176_c3-c4ed2c37.pthzqhttps://github.com/huggingface/pytorch-image-models/releases/download/v0.1-rsb-weights/resnet18_a1_0-d63eafa0.pth)rrzqhttps://github.com/huggingface/pytorch-image-models/releases/download/v0.1-rsb-weights/resnet18_a2_0-b61bd467.pthzqhttps://github.com/huggingface/pytorch-image-models/releases/download/v0.1-rsb-weights/resnet18_a3_0-40c531c8.pthzkhttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-weights/resnet18d_ra2-48a79e06.pth)rrrzqhttps://github.com/huggingface/pytorch-image-models/releases/download/v0.1-rsb-weights/resnet34_a1_0-46f8f793.pthzqhttps://github.com/huggingface/pytorch-image-models/releases/download/v0.1-rsb-weights/resnet34_a2_0-82d47d71.pthzqhttps://github.com/huggingface/pytorch-image-models/releases/download/v0.1-rsb-weights/resnet34_a3_0-a20cabb6.pth)rrrzfhttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-weights/resnet34-43635321.pthzkhttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-weights/resnet34d_ra2-f8dcfcaf.pthzfhttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-weights/resnet26-9aa10e23.pthzghttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-weights/resnet26d-69e92c46.pthzthttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-attn-weights/resnet26t_256_ra2-6f6fa748.pth)rBrr)rrgGz?)rB@rr)rrrrrrrrzohttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-rsb-weights/resnet50_a1_0-14fe96d1.pthzuhttps://github.com/huggingface/pytorch-image-models/releases/download/v0.1-rsb-weights/resnet50_a1h2_176-001a1197.pthg?)rrrrrrrzqhttps://github.com/huggingface/pytorch-image-models/releases/download/v0.1-rsb-weights/resnet50_a2_0-a2746f79.pthzqhttps://github.com/huggingface/pytorch-image-models/releases/download/v0.1-rsb-weights/resnet50_a3_0-59cae1ef.pthzphttps://github.com/huggingface/pytorch-image-models/releases/download/v0.1-rsb-weights/resnet50_b1k-532a802a.pthzphttps://github.com/huggingface/pytorch-image-models/releases/download/v0.1-rsb-weights/resnet50_b2k-1ba180c1.pthzohttps://github.com/huggingface/pytorch-image-models/releases/download/v0.1-rsb-weights/resnet50_c1-5ba5e060.pthzohttps://github.com/huggingface/pytorch-image-models/releases/download/v0.1-rsb-weights/resnet50_c2-d01e05b2.pthznhttps://github.com/huggingface/pytorch-image-models/releases/download/v0.1-rsb-weights/resnet50_d-f39db8af.pthzlhttps://github.com/huggingface/pytorch-image-models/releases/download/v0.1-weights/resnet50_ram-a26f946b.pthzkhttps://github.com/huggingface/pytorch-image-models/releases/download/v0.1-weights/resnet50_am-6c502b37.pthzkhttps://github.com/huggingface/pytorch-image-models/releases/download/v0.1-weights/resnet50_ra-85ebb6e5.pthzkhttps://github.com/huggingface/pytorch-image-models/releases/download/v0.1-weights/rw_resnet50-86acaeed.pthzkhttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-weights/resnet50d_ra2-464e36ba.pthzrhttps://github.com/huggingface/pytorch-image-models/releases/download/v0.1-rsb-weights/resnet50d_a1_0-e20cff14.pthzrhttps://github.com/huggingface/pytorch-image-models/releases/download/v0.1-rsb-weights/resnet50d_a2_0-a3adc64d.pthzrhttps://github.com/huggingface/pytorch-image-models/releases/download/v0.1-rsb-weights/resnet50d_a3_0-403fdfad.pth)rzohttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-rsb-weights/resnet101_a1h-36d3f2aa.pthzrhttps://github.com/huggingface/pytorch-image-models/releases/download/v0.1-rsb-weights/resnet101_a1_0-cdcb52a9.pthzrhttps://github.com/huggingface/pytorch-image-models/releases/download/v0.1-rsb-weights/resnet101_a2_0-6edb36c7.pthzrhttps://github.com/huggingface/pytorch-image-models/releases/download/v0.1-rsb-weights/resnet101_a3_0-1db14157.pthzlhttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-weights/resnet101d_ra2-2803ffab.pth)rrrrrrrrzohttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-rsb-weights/resnet152_a1h-dc400468.pthzrhttps://github.com/huggingface/pytorch-image-models/releases/download/v0.1-rsb-weights/resnet152_a1_0-2eee8a7a.pthzrhttps://github.com/huggingface/pytorch-image-models/releases/download/v0.1-rsb-weights/resnet152_a2_0-b4c6978f.pthzrhttps://github.com/huggingface/pytorch-image-models/releases/download/v0.1-rsb-weights/resnet152_a3_0-134d4688.pthzlhttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-weights/resnet152d_ra2-5cac0439.pthzlhttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-weights/resnet200d_ra2-bdba9bf9.pthzphttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-weights/wide_resnet50_racm-8234f177.pthz9https://download.pytorch.org/models/resnet18-5c106cde.pthz bsd-3-clausez!https://github.com/pytorch/vision)rrlicenserz9https://download.pytorch.org/models/resnet34-333f7ec4.pthz9https://download.pytorch.org/models/resnet50-19c8e357.pthz9https://download.pytorch.org/models/resnet50-11ad3fa6.pthgzG?)rrrrrrrrz:https://download.pytorch.org/models/resnet101-5d3b4d8f.pthz:https://download.pytorch.org/models/resnet101-cd907fc2.pthz:https://download.pytorch.org/models/resnet152-b121ed2d.pthz:https://download.pytorch.org/models/resnet152-f82ba261.pthz@https://download.pytorch.org/models/wide_resnet50_2-95faca4d.pthz@https://download.pytorch.org/models/wide_resnet50_2-9ba9bcbe.pthzAhttps://download.pytorch.org/models/wide_resnet101_2-32ee1156.pthzAhttps://download.pytorch.org/models/wide_resnet101_2-d733dc28.pthzrhttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-rsb-weights/resnet50_gn_a1h2-8fe6c4d0.pthzuhttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-rsb-weights/resnext50_32x4d_a1h-0146ab0a.pthzxhttps://github.com/huggingface/pytorch-image-models/releases/download/v0.1-rsb-weights/resnext50_32x4d_a1_0-b5a91a1d.pthzxhttps://github.com/huggingface/pytorch-image-models/releases/download/v0.1-rsb-weights/resnext50_32x4d_a2_0-efc76add.pthzxhttps://github.com/huggingface/pytorch-image-models/releases/download/v0.1-rsb-weights/resnext50_32x4d_a3_0-3e450271.pthzrhttps://github.com/huggingface/pytorch-image-models/releases/download/v0.1-weights/resnext50_32x4d_ra-d733960d.pthznhttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-weights/resnext50d_32x4d-103e99f8.pthzthttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-tpu-weights/resnext101_64x4d_c-0d0e0cc0.pthz@https://download.pytorch.org/models/resnext50_32x4d-7cdf4587.pthzAhttps://download.pytorch.org/models/resnext101_32x8d-8ba56ff5.pthzAhttps://download.pytorch.org/models/resnext101_64x4d-173b62eb.pthz@https://download.pytorch.org/models/resnext50_32x4d-1a0047aa.pthzAhttps://download.pytorch.org/models/resnext101_32x8d-110c445d.pthzChttps://download.pytorch.org/models/ig_resnext101_32x8-c38310e5.pthz cc-by-nc-4.0z.https://github.com/facebookresearch/WSL-ImageszDhttps://download.pytorch.org/models/ig_resnext101_32x16-c6f796b0.pthzDhttps://download.pytorch.org/models/ig_resnext101_32x32-e4b90b00.pthzDhttps://download.pytorch.org/models/ig_resnext101_32x48-3e41cc8a.pthzdhttps://dl.fbaipublicfiles.com/semiweaksupervision/model_files/semi_supervised_resnet18-d92f0530.pthzEhttps://github.com/facebookresearch/semi-supervised-ImageNet1K-modelszdhttps://dl.fbaipublicfiles.com/semiweaksupervision/model_files/semi_supervised_resnet50-08389792.pthzjhttps://dl.fbaipublicfiles.com/semiweaksupervision/model_files/semi_supervised_resnext50_32x4-ddb3e555.pthzkhttps://dl.fbaipublicfiles.com/semiweaksupervision/model_files/semi_supervised_resnext101_32x4-dc43570a.pthzkhttps://dl.fbaipublicfiles.com/semiweaksupervision/model_files/semi_supervised_resnext101_32x8-2cfe2f8b.pthzlhttps://dl.fbaipublicfiles.com/semiweaksupervision/model_files/semi_supervised_resnext101_32x16-15fffa57.pthzkhttps://dl.fbaipublicfiles.com/semiweaksupervision/model_files/semi_weakly_supervised_resnet18-118f1556.pthzkhttps://dl.fbaipublicfiles.com/semiweaksupervision/model_files/semi_weakly_supervised_resnet50-16a12f1b.pthzqhttps://dl.fbaipublicfiles.com/semiweaksupervision/model_files/semi_weakly_supervised_resnext50_32x4-72679e44.pthzrhttps://dl.fbaipublicfiles.com/semiweaksupervision/model_files/semi_weakly_supervised_resnext101_32x4-3f87e46b.pthzrhttps://dl.fbaipublicfiles.com/semiweaksupervision/model_files/semi_weakly_supervised_resnext101_32x8-b4712904.pthzshttps://dl.fbaipublicfiles.com/semiweaksupervision/model_files/semi_weakly_supervised_resnext101_32x16-f3559a9c.pthznhttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-weights/ecaresnet26t_ra2-46609757.pth)rrrrrrrzlhttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-tresnet/ecaresnetlight-75a9c627.pthr)rrrrzjhttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-tresnet/ecaresnet50d-93c81e3b.pth)rrrrrzlhttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-tresnet/ecaresnet50d_p-e4fa23c2.pthznhttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-weights/ecaresnet50t_ra2-f7ac63c4.pthzuhttps://github.com/huggingface/pytorch-image-models/releases/download/v0.1-rsb-weights/ecaresnet50t_a1_0-99bd76a8.pthzuhttps://github.com/huggingface/pytorch-image-models/releases/download/v0.1-rsb-weights/ecaresnet50t_a2_0-b1c7b745.pthzuhttps://github.com/huggingface/pytorch-image-models/releases/download/v0.1-rsb-weights/ecaresnet50t_a3_0-8cc311f1.pthzkhttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-tresnet/ecaresnet101d-153dad65.pthzmhttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-tresnet/ecaresnet101d_p-9e74cb91.pth)rrrrzshttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-weights/ecaresnet269d_320_ra2-7baa55cb.pth) r)rB`rzshttps://github.com/huggingface/pytorch-image-models/releases/download/v0.1-rsb-weights/seresnet50_a1_0-ffa00869.pthzshttps://github.com/huggingface/pytorch-image-models/releases/download/v0.1-rsb-weights/seresnet50_a2_0-850de0d9.pthzshttps://github.com/huggingface/pytorch-image-models/releases/download/v0.1-rsb-weights/seresnet50_a3_0-317ecd56.pthzohttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-weights/seresnet50_ra_224-8efdb4bb.pthznhttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-weights/seresnet152d_ra2-04464dd2.pth)rrrzphttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-weights/seresnext26d_32x4d-80fa48a3.pthzqhttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-weights/seresnext26tn_32x4d-569cb627.pthzthttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-weights/seresnext50_32x4d_racm-a304a460.pthzwhttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-tpu-weights/seresnext101_32x8d_ah-e6bc4c0a.pthzxhttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-tpu-weights/seresnext101d_32x8d_ah-191d7b94.pth)rrrr) r)rrrrrrr)rrrr) r)rBr)rrrrrri-.) rrrrrrrrr)rrrrrzjhttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-weights/resnetblur50-84f4748f.pthzphttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-rsb-weights/resnetaa50_a1h-4cf422b3.pthzzhttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-tpu-weights/seresnextaa101d_32x8d_ah-83c8ae12.pthzohttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-rs-weights/resnetrs50_ema-6b53758b.pthrrgQ?)rrrrrrrrzuhttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-rs-weights/resnetrs101_i192_ema-1509bbf6.pth)rBrzuhttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-rs-weights/resnetrs152_i256_ema-a9aff7f9.pthzohttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-tpu-weights/resnetrs200_c-6b698b88.pthzphttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-rs-weights/resnetrs270_ema-b40e674c.pthzuhttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-rs-weights/resnetrs350_i256_ema-5a1aa8f1.pthzphttps://github.com/rwightman/pytorch-image-models/releases/download/v0.1-rs-weights/resnetrs420_ema-972dee69.pth)rBrzrhttps://github.com/rwightman/pytorch-pretrained-gluonresnet/releases/download/v0.1/gluon_resnet18_v1b-0757602b.pthzrhttps://github.com/rwightman/pytorch-pretrained-gluonresnet/releases/download/v0.1/gluon_resnet34_v1b-c6d82d59.pthzrhttps://github.com/rwightman/pytorch-pretrained-gluonresnet/releases/download/v0.1/gluon_resnet50_v1b-0ebe02e2.pthzshttps://github.com/rwightman/pytorch-pretrained-gluonresnet/releases/download/v0.1/gluon_resnet101_v1b-3b017079.pthzshttps://github.com/rwightman/pytorch-pretrained-gluonresnet/releases/download/v0.1/gluon_resnet152_v1b-c1edb0dd.pthzrhttps://github.com/rwightman/pytorch-pretrained-gluonresnet/releases/download/v0.1/gluon_resnet50_v1c-48092f55.pthzshttps://github.com/rwightman/pytorch-pretrained-gluonresnet/releases/download/v0.1/gluon_resnet101_v1c-1f26822a.pthzshttps://github.com/rwightman/pytorch-pretrained-gluonresnet/releases/download/v0.1/gluon_resnet152_v1c-a3bb0b98.pthzrhttps://github.com/rwightman/pytorch-pretrained-gluonresnet/releases/download/v0.1/gluon_resnet50_v1d-818a1b1b.pthzshttps://github.com/rwightman/pytorch-pretrained-gluonresnet/releases/download/v0.1/gluon_resnet101_v1d-0f9c8644.pthzshttps://github.com/rwightman/pytorch-pretrained-gluonresnet/releases/download/v0.1/gluon_resnet152_v1d-bd354e12.pthzrhttps://github.com/rwightman/pytorch-pretrained-gluonresnet/releases/download/v0.1/gluon_resnet50_v1s-1762acc0.pthzshttps://github.com/rwightman/pytorch-pretrained-gluonresnet/releases/download/v0.1/gluon_resnet101_v1s-60fe0cc1.pthzshttps://github.com/rwightman/pytorch-pretrained-gluonresnet/releases/download/v0.1/gluon_resnet152_v1s-dcc41b81.pthzuhttps://github.com/rwightman/pytorch-pretrained-gluonresnet/releases/download/v0.1/gluon_resnext50_32x4d-e6a097c1.pthzvhttps://github.com/rwightman/pytorch-pretrained-gluonresnet/releases/download/v0.1/gluon_resnext101_32x4d-b253c8c4.pthzvhttps://github.com/rwightman/pytorch-pretrained-gluonresnet/releases/download/v0.1/gluon_resnext101_64x4d-f9a8e184.pthzwhttps://github.com/rwightman/pytorch-pretrained-gluonresnet/releases/download/v0.1/gluon_seresnext50_32x4d-90cf2d6e.pthzxhttps://github.com/rwightman/pytorch-pretrained-gluonresnet/releases/download/v0.1/gluon_seresnext101_32x4d-cf52900d.pthzxhttps://github.com/rwightman/pytorch-pretrained-gluonresnet/releases/download/v0.1/gluon_seresnext101_64x4d-f9926f93.pthznhttps://github.com/rwightman/pytorch-pretrained-gluonresnet/releases/download/v0.1/gluon_senet154-70a1a3c0.pth)zresnet10t.c3_in1kzresnet14t.c3_in1kzresnet18.a1_in1kzresnet18.a2_in1kzresnet18.a3_in1kzresnet18d.ra2_in1kzresnet34.a1_in1kzresnet34.a2_in1kzresnet34.a3_in1kzresnet34.bt_in1kzresnet34d.ra2_in1kzresnet26.bt_in1kzresnet26d.bt_in1kzresnet26t.ra2_in1kzresnet50.a1_in1kzresnet50.a1h_in1kzresnet50.a2_in1kzresnet50.a3_in1kzresnet50.b1k_in1kzresnet50.b2k_in1kzresnet50.c1_in1kzresnet50.c2_in1kzresnet50.d_in1kzresnet50.ram_in1kzresnet50.am_in1kzresnet50.ra_in1kzresnet50.bt_in1kzresnet50d.ra2_in1kzresnet50d.a1_in1kzresnet50d.a2_in1kzresnet50d.a3_in1kzresnet50t.untrainedzresnet101.a1h_in1kzresnet101.a1_in1kzresnet101.a2_in1kzresnet101.a3_in1kzresnet101d.ra2_in1kzresnet152.a1h_in1kzresnet152.a1_in1kzresnet152.a2_in1kzresnet152.a3_in1kzresnet152d.ra2_in1kzresnet200.untrainedzresnet200d.ra2_in1kzwide_resnet50_2.racm_in1kzresnet18.tv_in1kresnet34.tv_in1kresnet50.tv_in1kzresnet50.tv2_in1kresnet101.tv_in1kzresnet101.tv2_in1kresnet152.tv_in1kzresnet152.tv2_in1kzwide_resnet50_2.tv_in1kzwide_resnet50_2.tv2_in1kzwide_resnet101_2.tv_in1kzwide_resnet101_2.tv2_in1kzresnet50_gn.a1h_in1kzresnext50_32x4d.a1h_in1kzresnext50_32x4d.a1_in1kzresnext50_32x4d.a2_in1kzresnext50_32x4d.a3_in1kzresnext50_32x4d.ra_in1kzresnext50d_32x4d.bt_in1kzresnext101_32x4d.untrainedzresnext101_64x4d.c1_in1kresnext50_32x4d.tv_in1kzresnext101_32x8d.tv_in1kzresnext101_64x4d.tv_in1kzresnext50_32x4d.tv2_in1kzresnext101_32x8d.tv2_in1k$resnext101_32x8d.fb_wsl_ig1b_ft_in1kz%resnext101_32x16d.fb_wsl_ig1b_ft_in1kz%resnext101_32x32d.fb_wsl_ig1b_ft_in1kz%resnext101_32x48d.fb_wsl_ig1b_ft_in1k resnet18.fb_ssl_yfcc100m_ft_in1k resnet50.fb_ssl_yfcc100m_ft_in1k'resnext50_32x4d.fb_ssl_yfcc100m_ft_in1k(resnext101_32x4d.fb_ssl_yfcc100m_ft_in1k(resnext101_32x8d.fb_ssl_yfcc100m_ft_in1k)resnext101_32x16d.fb_ssl_yfcc100m_ft_in1kresnet18.fb_swsl_ig1b_ft_in1kresnet50.fb_swsl_ig1b_ft_in1k$resnext50_32x4d.fb_swsl_ig1b_ft_in1k%resnext101_32x4d.fb_swsl_ig1b_ft_in1k%resnext101_32x8d.fb_swsl_ig1b_ft_in1k&resnext101_32x16d.fb_swsl_ig1b_ft_in1kzecaresnet26t.ra2_in1kzecaresnetlight.miil_in1kzecaresnet50d.miil_in1kzecaresnet50d_pruned.miil_in1kzecaresnet50t.ra2_in1kzecaresnet50t.a1_in1kzecaresnet50t.a2_in1kzecaresnet50t.a3_in1kzecaresnet101d.miil_in1kzecaresnet101d_pruned.miil_in1kzecaresnet200d.untrainedzecaresnet269d.ra2_in1kzecaresnext26t_32x4d.untrainedzecaresnext50t_32x4d.untrainedzseresnet18.untrainedzseresnet34.untrainedzseresnet50.a1_in1kzseresnet50.a2_in1kzseresnet50.a3_in1kzseresnet50.ra2_in1kzseresnet50t.untrainedzseresnet101.untrainedzseresnet152.untrainedzseresnet152d.ra2_in1kzseresnet200d.untrainedzseresnet269d.untrainedzseresnext26d_32x4d.bt_in1kzseresnext26t_32x4d.bt_in1kzseresnext50_32x4d.racm_in1kzseresnext101_32x4d.untrainedzseresnext101_32x8d.ah_in1kzseresnext101d_32x8d.ah_in1kzresnetaa50d.sw_in12k_ft_in1kzresnetaa101d.sw_in12k_ft_in1kz*seresnextaa101d_32x8d.sw_in12k_ft_in1k_288z&seresnextaa101d_32x8d.sw_in12k_ft_in1kz*seresnextaa201d_32x8d.sw_in12k_ft_in1k_384zseresnextaa201d_32x8d.sw_in12kzresnetaa50d.sw_in12kzresnetaa50d.d_in12kzresnetaa101d.sw_in12kzseresnextaa101d_32x8d.sw_in12kzresnetblur18.untrainedzresnetblur50.bt_in1kzresnetblur50d.untrainedzresnetblur101d.untrainedzresnetaa34d.untrainedzresnetaa50.a1h_in1kzseresnetaa50d.untrainedzseresnextaa101d_32x8d.ah_in1kzresnetrs50.tf_in1kzresnetrs101.tf_in1kzresnetrs152.tf_in1kzresnetrs200.tf_in1kzresnetrs270.tf_in1kzresnetrs350.tf_in1kzresnetrs420.tf_in1kresnet18.gluon_in1kresnet34.gluon_in1kresnet50.gluon_in1kresnet101.gluon_in1kresnet152.gluon_in1kresnet50c.gluon_in1kresnet101c.gluon_in1kresnet152c.gluon_in1kresnet50d.gluon_in1kresnet101d.gluon_in1kresnet152d.gluon_in1kresnet50s.gluon_in1kresnet101s.gluon_in1kresnet152s.gluon_in1kresnext50_32x4d.gluon_in1kresnext101_32x4d.gluon_in1kresnext101_64x4d.gluon_in1kseresnext50_32x4d.gluon_in1kseresnext101_32x4d.gluon_in1kseresnext101_64x4d.gluon_in1ksenet154.gluon_in1kcKs0ttddddgdddd}td|ft|f|S)z$Constructs a ResNet-10-T model. rr deep_tieredTrrrrr resnet10trrrrr model_argsr&r&r(r,sr,cKs0ttddddgdddd}td|ft|f|S)z$Constructs a ResNet-14-T model. rrr*Tr+ resnet14trr rr.r&r&r(r0sr0cKs*ttddddgd}td|ft|f|S)z"Constructs a ResNet-18 model. r%rrresnet18r-r.r&r&r(r3sr3cKs0ttddddgdddd}td|ft|f|S)z$Constructs a ResNet-18-D model. r%rrTr+ resnet18dr-r.r&r&r(r4sr4cKs*ttddddgd}td|ft|f|S)z"Constructs a ResNet-34 model. rBrprr2resnet34r-r.r&r&r(r5sr5cKs0ttddddgdddd}td|ft|f|S) z$Constructs a ResNet-34-D model. rBrprrrTr+ resnet34dr-r.r&r&r(r6sr6cKs*ttddddgd}td|ft|f|S)z"Constructs a ResNet-26 model. r%r2resnet26r1r.r&r&r(r7 sr7cKs0ttddddgdddd}td|ft|f|S)z$Constructs a ResNet-26-T model. r%rr*Tr+ resnet26tr1r.r&r&r(r8sr8cKs0ttddddgdddd}td|ft|f|S)z$Constructs a ResNet-26-D model. r%rrTr+ resnet26dr1r.r&r&r(r9sr9cKs*ttddddgd}td|ft|f|S)z"Constructs a ResNet-50 model. rBrprr2resnet50r1r.r&r&r(r:%sr:cKs.ttddddgddd}td|ft|f|S)z$Constructs a ResNet-50-C model. rBrprrrrrrr resnet50cr1r.r&r&r(r<-sr<cKs0ttddddgdddd}td|ft|f|S) z$Constructs a ResNet-50-D model. rBrprrrTr+ resnet50dr1r.r&r&r(r=5sr=cKs.ttddddgddd}td|ft|f|S)z$Constructs a ResNet-50-S model. rBrprr3rr; resnet50sr1r.r&r&r(r>=sr>cKs0ttddddgdddd}td|ft|f|S) z$Constructs a ResNet-50-T model. rBrprrr*Tr+ resnet50tr1r.r&r&r(r?Esr?cKs*ttddddgd}td|ft|f|S)z#Constructs a ResNet-101 model. rBrpr2 resnet101r1r.r&r&r(rAMsrAcKs.ttddddgddd}td|ft|f|S)z%Constructs a ResNet-101-C model. rBrpr@rrr; resnet101cr1r.r&r&r(rBUsrBcKs0ttddddgdddd}td|ft|f|S) z%Constructs a ResNet-101-D model. rBrpr@rrTr+ resnet101dr1r.r&r&r(rC]srCcKs.ttddddgddd}td|ft|f|S)z%Constructs a ResNet-101-S model. rBrpr@r3rr; resnet101sr1r.r&r&r(rDesrDcKs*ttddddgd}td|ft|f|S)z#Constructs a ResNet-152 model. rBr$r2 resnet152r1r.r&r&r(rFmsrFcKs.ttddddgddd}td|ft|f|S)z%Constructs a ResNet-152-C model. rBrrErrr; resnet152cr1r.r&r&r(rGusrGcKs0ttddddgdddd}td|ft|f|S) z%Constructs a ResNet-152-D model. rBrrErrTr+ resnet152dr1r.r&r&r(rH}srHcKs.ttddddgddd}td|ft|f|S)z%Constructs a ResNet-152-S model. rBrrEr3rr; resnet152sr1r.r&r&r(rIsrIcKs*ttddddgd}td|ft|f|S)z#Constructs a ResNet-200 model. rBrEr2 resnet200r1r.r&r&r(rKsrKcKs0ttddddgdddd}td|ft|f|S) z%Constructs a ResNet-200-D model. rBrJrErrTr+ resnet200dr1r.r&r&r(rLsrLcKs,ttddddgdd}td|ft|f|S)aOConstructs a Wide ResNet-50-2 model. The model is the same as ResNet except for the bottleneck number of channels which is twice larger in every block. The number of channels in outer 1x1 convolutions is the same, e.g. last block in ResNet-50 has 2048-512-2048 channels, and in Wide ResNet-50-2 has 2048-1024-2048. rBrprrrrr9wide_resnet50_2r1r.r&r&r(rNsrNcKs,ttddddgdd}td|ft|f|S)zConstructs a Wide ResNet-101-2 model. The model is the same as ResNet except for the bottleneck number of channels which is twice larger in every block. The number of channels in outer 1x1 convolutions is the same. rBrpr@rrMwide_resnet101_2r1r.r&r&r(rOsrOcKs,ttddddgdd}td|ft|f|S)z.Constructs a ResNet-50 model w/ GroupNorm rBrprZ groupnorm)rrr= resnet50_gnr1r.r&r&r(rPsrPcKs.ttddddgddd}td|ft|f|S)z(Constructs a ResNeXt50-32x4d model. rBrprrrrr8r9resnext50_32x4dr1r.r&r&r(rRsrRc Ks4ttddddgdddddd}td|ft|f|S) zVConstructs a ResNeXt50d-32x4d model. ResNext50 w/ deep stem & avg pool downsample rBrprrrT)rrr8r9rrrresnext50d_32x4dr1r.r&r&r(rSs rScKs.ttddddgddd}td|ft|f|S)z*Constructs a ResNeXt-101 32x4d model. rBrpr@rrQresnext101_32x4dr1r.r&r&r(rTsrTcKs.ttddddgddd}td|ft|f|S)z*Constructs a ResNeXt-101 32x8d model. rBrpr@rrrQresnext101_32x8dr1r.r&r&r(rUsrUcKs.ttddddgddd}td|ft|f|S)z*Constructs a ResNeXt-101 32x16d model rBrpr@rrrQresnext101_32x16dr1r.r&r&r(rVsrVcKs.ttddddgddd}td|ft|f|S)z*Constructs a ResNeXt-101 32x32d model rBrpr@rrQresnext101_32x32dr1r.r&r&r(rWsrWcKs.ttddddgddd}td|ft|f|S)z)Constructs a ResNeXt101-64x4d model. rBrpr@r3rQresnext101_64x4dr1r.r&r&r(rXsrXc Ks8ttddddgdddtddd}td|ft|f|S) zConstructs an ECA-ResNeXt-26-T model. This is technically a 28 layer ResNet, like a 'D' bag-of-tricks model but with tiered 24, 32, 64 channels in the deep stem and ECA attn. r%rr*Tecar>rrrrrr ecaresnet26tr1r.r&r&r(r\s r\c Ks8ttddddgdddtddd }td |ft|f|S) z-Constructs a ResNet-50-D model with eca. rBrprrrTrYrZr[ ecaresnet50dr1r.r&r&r(r]s r]c Ks@ttddddgdddtddd }td |fd dit|f|S) zConstructs a ResNet-50-D model pruned with eca. The pruning has been obtained using https://arxiv.org/pdf/2002.08258.pdf rBrprrrTrYrZr[ecaresnet50d_prunedprunedr1r.r&r&r(r^ s r^c Ks8ttddddgdddtddd }td |ft|f|S) zConstructs an ECA-ResNet-50-T model. Like a 'D' bag-of-tricks model but with tiered 24, 32, 64 channels in the deep stem and ECA attn. rBrprrr*TrYrZr[ ecaresnet50tr1r.r&r&r(r`s r`cKs6ttddddgddtddd}td |ft|f|S) z3Constructs a ResNet-50-D light model with eca. r rBrTrYrZ)rrrrrecaresnetlightr1r.r&r&r(rb"s rbc Ks8ttddddgdddtddd }td |ft|f|S) z.Constructs a ResNet-101-D model with eca. rBrpr@rrTrYrZr[ ecaresnet101dr1r.r&r&r(rc,s rcc Ks@ttddddgdddtddd }td |fd dit|f|S) zConstructs a ResNet-101-D model pruned with eca. The pruning has been obtained using https://arxiv.org/pdf/2002.08258.pdf rBrpr@rrTrYrZr[ecaresnet101d_prunedr_r1r.r&r&r(rd6s rdc Ks8ttddddgdddtddd }td |ft|f|S) z.Constructs a ResNet-200-D model with ECA. rBrJrErrTrYrZr[ ecaresnet200dr1r.r&r&r(reAs rec Ks8ttddddgdddtdd d }td |ft|f|S) z.Constructs a ResNet-269-D model with ECA. rB0rrrTrYrZr[ ecaresnet269dr1r.r&r&r(rhKs rhc Ks<ttddddgdddddtddd}td |ft|f|S) zConstructs an ECA-ResNeXt-26-T model. This is technically a 28 layer ResNet, like a 'D' bag-of-tricks model but with tiered 24, 32, 64 channels in the deep stem. This model replaces SE module with the ECA module r%rrpr*TrYrZrrr8r9rrrrecaresnext26t_32x4dr1r.r&r&r(rjUs rjc Ks<ttddddgdddddtddd}td |ft|f|S) zConstructs an ECA-ResNeXt-50-T model. This is technically a 28 layer ResNet, like a 'D' bag-of-tricks model but with tiered 24, 32, 64 channels in the deep stem. This model replaces SE module with the ECA module r%rrpr*TrYrZriecaresnext50t_32x4dr1r.r&r&r(rkas rkcKs2ttddddgtddd}td|ft|f|S)Nr%rQrZrrr seresnet18r-r.r&r&r(rmmsrmcKs2ttddddgtddd}td|ft|f|S)NrBrprrQrZrl seresnet34r-r.r&r&r(rnssrncKs2ttddddgtddd}td|ft|f|S)NrBrprrQrZrl seresnet50r1r.r&r&r(roysroc Ks8ttddddgdddtddd }td |ft|f|S) NrBrprrr*TrQrZr[ seresnet50tr1r.r&r&r(rps rpcKs2ttddddgtddd}td|ft|f|S)NrBrpr@rQrZrl seresnet101r1r.r&r&r(rqsrqcKs2ttddddgtddd}td|ft|f|S)NrBrrErQrZrl seresnet152r1r.r&r&r(rrsrrc Ks8ttddddgdddtddd }td |ft|f|S) NrBrrErrTrQrZr[ seresnet152dr1r.r&r&r(rss rsc Ks8ttddddgdddtddd }td |ft|f|S) z2Constructs a ResNet-200-D model with SE attn. rBrJrErrTrQrZr[ seresnet200dr1r.r&r&r(rts rtc Ks8ttddddgdddtdd d }td |ft|f|S) z2Constructs a ResNet-269-D model with SE attn. rBrfrgrrrTrQrZr[ seresnet269dr1r.r&r&r(rus ruc Ks<ttddddgdddddtddd}td |ft|f|S) zConstructs a SE-ResNeXt-26-D model.` This is technically a 28 layer ResNet, using the 'D' modifier from Gluon / bag-of-tricks for combination of deep stem and avg_pool in downsample. r%rrprTrQrZriseresnext26d_32x4dr1r.r&r&r(rvs rvc Ks<ttddddgdddddtddd}td |ft|f|S) zConstructs a SE-ResNet-26-T model. This is technically a 28 layer ResNet, like a 'D' bag-of-tricks model but with tiered 24, 32, 64 channels in the deep stem. r%rrpr*TrQrZriseresnext26t_32x4dr1r.r&r&r(rws rwcKs6ttddddgddtddd}td|ft|f|S) NrBrprrrQrZrrr8r9rseresnext50_32x4dr1r.r&r&r(rys rycKs6ttddddgddtddd}td|ft|f|S) NrBrpr@rrQrZrxseresnext101_32x4dr1r.r&r&r(rzs rzcKs6ttddddgddtddd}td |ft|f|S) NrBrpr@rrrQrZrxseresnext101_32x8dr1r.r&r&r(r{s r{c Ks<ttddddgdddddtdd d }td |ft|f|S) NrBrpr@rrrTrQrZriseresnext101d_32x8dr1r.r&r&r(r|s r|cKs6ttddddgddtddd}td|ft|f|S) NrBrpr@r3rQrZrxseresnext101_64x4dr1r.r&r&r(r}s r}c Ks<ttddddgdddddtdd d }td |ft|f|S) NrBrrEr3rprr%rQrZ)rrr8r9rrrrsenet154r1r.r&r&r(r~s r~cKs,ttddddgtd}td|ft|f|S)z9Constructs a ResNet-18 model with blur anti-aliasing r%rrr* resnetblur18)rrrrr.r&r&r(rsrcKs,ttddddgtd}td|ft|f|S)z9Constructs a ResNet-50 model with blur anti-aliasing rBrprr resnetblur50rr rrr.r&r&r(rsrcKs2ttddddgtdddd}td|ft|f|S) z;Constructs a ResNet-50-D model with blur anti-aliasing rBrprrrTrrr*rrr resnetblur50drr.r&r&r(rs rcKs2ttddddgtdddd}td|ft|f|S) zConstructs a ResNet-50-D model with avgpool anti-aliasing rBrprrrTr resnetaa50drr.r&r&r(r-s rcKs4ttddddgtjdddd}td|ft|f|S) z?Constructs a ResNet-101-D model with avgpool anti-aliasing rBrpr@rrTr resnetaa101drr.r&r&r(r7s rc Ks<ttddddgtjdddtddd }td |ft|f|S) zAConstructs a SE=ResNet-50-D model with avgpool anti-aliasing rBrprrrTrQrZ)rrr*rrrr seresnetaa50drr.r&r&r(rAs rc Ks@ttddddgdddddtjtdd d }td |ft|f|S) IConstructs a SE=ResNeXt-101-D 32x8d model with avgpool anti-aliasing rBrpr@rrrTrQrZ rrr8r9rrrr*rseresnextaa101d_32x8drr.r&r&r(rKs r)rc Ks@ttddddgddddd tjtd d d }td |ft|f|S)rrBrJrErprrr3rTrQrZrseresnextaa201d_32x8drr.r&r&r(rVs rc KsJttddd}ttddddgddd d t|d d }td |ft|f|S) zConstructs a ResNet-RS-50 model. Paper: Revisiting ResNets - https://arxiv.org/abs/2103.07579 Pretrained weights from https://github.com/tensorflow/tpu/tree/bee9c4f6/models/official/resnet/resnet_rs rQrZrd_ratiorBrprrrTrZrrrrrrr resnetrs50rrrr rrrr>r/r&r&r(ras rc KsJttddd}ttddddgddd d t|d d }td |ft|f|S) zConstructs a ResNet-RS-101 model. Paper: Revisiting ResNets - https://arxiv.org/abs/2103.07579 Pretrained weights from https://github.com/tensorflow/tpu/tree/bee9c4f6/models/official/resnet/resnet_rs rQrrrBrpr@rrTrZr resnetrs101rrr&r&r(rns rc KsJttddd}ttddddgddd d t|d d }td |ft|f|S) zConstructs a ResNet-RS-152 model. Paper: Revisiting ResNets - https://arxiv.org/abs/2103.07579 Pretrained weights from https://github.com/tensorflow/tpu/tree/bee9c4f6/models/official/resnet/resnet_rs rQrrrBrrErrTrZr resnetrs152rrr&r&r(r{s rc KsJttddd}ttddddgddd d t|d d }td |ft|f|S) zConstructs a ResNet-RS-200 model. Paper: Revisiting ResNets - https://arxiv.org/abs/2103.07579 Pretrained weights from https://github.com/tensorflow/tpu/tree/bee9c4f6/models/official/resnet/resnet_rs rQrrrBrJrErrTrZr resnetrs200rrr&r&r(rs rc KsJttddd}ttddddgddd d t|d d }td |ft|f|S) zConstructs a ResNet-RS-270 model. Paper: Revisiting ResNets - https://arxiv.org/abs/2103.07579 Pretrained weights from https://github.com/tensorflow/tpu/tree/bee9c4f6/models/official/resnet/resnet_rs rQrrrp5rrTrZr resnetrs270rrr&r&r(rs rc KsJttddd}ttddddgddd d t|d d }td |ft|f|S) zConstructs a ResNet-RS-350 model. Paper: Revisiting ResNets - https://arxiv.org/abs/2103.07579 Pretrained weights from https://github.com/tensorflow/tpu/tree/bee9c4f6/models/official/resnet/resnet_rs rQrrrprEHrrTrZr resnetrs350rrr&r&r(rs rc KsJttddd}ttddddgddd d t|d d }td |ft|f|S) zConstructs a ResNet-RS-420 model Paper: Revisiting ResNets - https://arxiv.org/abs/2103.07579 Pretrained weights from https://github.com/tensorflow/tpu/tree/bee9c4f6/models/official/resnet/resnet_rs rQrrrp,WrrTrZr resnetrs420rrr&r&r(rs rrrrrrrr r r r r rrrrrrrrrrrrrrrrrrr r!r"r#r$r%r&r'r(r))+Z tv_resnet34Z tv_resnet50Z tv_resnet101Z tv_resnet152Ztv_resnext50_32x4dZig_resnext101_32x8dZig_resnext101_32x16dZig_resnext101_32x32dZig_resnext101_32x48dZ ssl_resnet18Z ssl_resnet50Zssl_resnext50_32x4dZssl_resnext101_32x4dZssl_resnext101_32x8dZssl_resnext101_32x16dZ swsl_resnet18Z swsl_resnet50Zswsl_resnext50_32x4dZswsl_resnext101_32x4dZswsl_resnext101_32x8dZswsl_resnext101_32x16dZgluon_resnet18_v1bZgluon_resnet34_v1bZgluon_resnet50_v1bZgluon_resnet101_v1bZgluon_resnet152_v1bZgluon_resnet50_v1cZgluon_resnet101_v1cZgluon_resnet152_v1cZgluon_resnet50_v1dZgluon_resnet101_v1dZgluon_resnet152_v1dZgluon_resnet50_v1sZgluon_resnet101_v1sZgluon_resnet152_v1sZgluon_resnext50_32x4dZgluon_resnext101_32x4dZgluon_resnext101_64x4dZgluon_seresnext50_32x4dZgluon_seresnext101_32x4dZgluon_seresnext101_64x4dZgluon_senet154Zseresnext26tn_32x4d)r)r%T)rrNN)rrNN)r)rrrFrr)F)r)r)r)r)r)r)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)F)rrr functoolsrtypingrrrrrrr rmZtorch.nnr.Ztorch.nn.functionalZ functionalrZ timm.datar r Z timm.layersr rrrrrrrrrrZ_builderrZ _manipulater _registryrrr__all__rkr)rlrr2rr r|rrrrrrrrrrrrrZ default_cfgsr,r0r3r4r5r6r7r8r9r:r<r=r>r?rArBrCrDrFrGrHrIrKrLrNrOrPrRrSrTrUrVrWrXr\r]r^r`rbrcrdrerhrjrkrmrnrorprqrrrsrtrurvrwryrzr{r|r}r~rrrrrrrrrrrrrrrrrrrfr&r&r&r(s8   $ 4   " bl      $ B^      j