ڧg; ddlmZddlmZmZmZmZmZddlZddlm Z m Z ddl m Z ddlmZmZddlmZdd lmZdd lmZd d lmZmZmZd d lmZd dlmZmZgdZ Gdde j!Z"Gdde j#Z$GddZ%Gdde j#Z&dee%de'deede(dede&f dZ)ded d!d"Z*Gd#d$eZ+Gd%d&eZ,Gd'd(eZ-Gd)d*eZ.eed+e+j/f,dd-d.dee+de(dede&fd/Z0eed+e,j/f,dd-d.dee,de(dede&fd0Z1eed+e-j/f,dd-d.dee-de(dede&fd1Z2eed+e.j/f,dd-d.dee.de(dede&fd2Z3dS)3)partial)AnyCallableListOptionalSequenceN)nnTensor) functional)Conv2dNormActivationPermute)StochasticDepth)ImageClassification)_log_api_usage_once)register_modelWeights WeightsEnum)_IMAGENET_CATEGORIES)_ovewrite_named_paramhandle_legacy_interface) ConvNeXtConvNeXt_Tiny_WeightsConvNeXt_Small_WeightsConvNeXt_Base_WeightsConvNeXt_Large_Weights convnext_tinyconvnext_small convnext_baseconvnext_largeceZdZdedefdZdS) LayerNorm2dxreturnc|dddd}tj||j|j|j|j}|dddd}|S)Nrr r)permuteF layer_normnormalized_shapeweightbiasepsselfr$s W/var/www/html/ai-engine/env/lib/python3.11/site-packages/torchvision/models/convnext.pyforwardzLayerNorm2d.forwardsU IIaAq ! ! LD14; 48 T T IIaAq ! !N)__name__ __module__ __qualname__r r2r3r1r#r#s6Fr3r#c jeZdZ d dededeedejfddffd Zde de fd Z xZ S) CNBlockN layer_scalestochastic_depth_prob norm_layer.r%c Jt|ttjd}tjtj||dd|dtgd||tj|d|zd tj tjd|z|d tgd |_ tj tj |d d |z|_t|d |_dS) Nư>r.r'T) kernel_sizepaddinggroupsr-)rr r'r) in_features out_featuresr-)rr'rr rrow)super__init__rr LayerNorm SequentialConv2drLinearGELUblock Parametertorchonesr:rstochastic_depth)r0dimr:r;r< __class__s r1rIzCNBlock.__init__'s    4888J] Ic3Aq4 P P P LLL ! ! JsOO I#AG$ G G G GII I!c'$ G G G LLL ! !   < 31(=(= (KLL /0Eu M Mr3inputct|j||z}||}||z }|SN)r:rOrS)r0rVresults r1r2zCNBlock.forward>s<!DJJu$5$55&&v..% r3rX) r4r5r6floatrrr ModulerIr r2 __classcell__rUs@r1r9r9&s :> NNN % N Xc29n56 N  NNNNNN.Vr3r9c>eZdZdedeededdfdZdefdZdS) CNBlockConfiginput_channels out_channels num_layersr%Nc0||_||_||_dSrX)r`rarb)r0r`rarbs r1rIzCNBlockConfig.__init__Gs -($r3cl|jjdz}|dz }|dz }|dz }|dz }|jdi|jS)N(zinput_channels={input_channels}z, out_channels={out_channels}z, num_layers={num_layers})r7)rUr4format__dict__)r0ss r1__repr__zCNBlockConfig.__repr__QsU N #c ) .. ,, (( Sqx(($-(((r3)r4r5r6intrrIstrrjr7r3r1r_r_Esm%%sm% %  %%%%)#))))))r3r_ceZdZ ddeedededed eed e j fd eed e j fd e d dffd Z de d e fdZde d e fdZxZS)rr>N block_settingr;r: num_classesrO.r<kwargsr%c tt||stdt |t rt d|Dstd|t}|ttd}g}|dj } | td| ddd|dd td |D} d} |D]} g} t| jD]7}|| z| d z z }| || j ||| d z } 8| t#j| | jV| t#j|| j t#j| j | jddt#j||_t#jd |_|d}|j|jn|j }t#j||t#jd t#j|||_|D]y}t |t"jt"jfrQt"j|jd|j$t"j |jzdS)Nz%The block_setting should not be emptyc8g|]}t|tSr7) isinstancer_).0ris r1 z%ConvNeXt.__init__..js#>s>s>s`az!]?[?[>s>s>sr3z/The block_setting should be List[CNBlockConfig]r>r?rr'rDT)rAstriderBr<activation_layerr-c3$K|] }|jV dSrX)rb)rvcnfs r1 z$ConvNeXt.__init__..s$ I IC I I I I I Ir3g?rr )rArxg{Gz?)std)!rHrIr ValueErrorrurall TypeErrorr9rr#r`appendr sumrangerbr rKrarLfeaturesAdaptiveAvgPool2davgpoolFlattenrM classifiermodulesinit trunc_normal_r,r-zeros_)r0rpr;r:rqrOr<rrlayersfirstconv_output_channelstotal_stage_blocksstage_block_idr{stage_sd_prob lastblocklastconv_output_channelsmrUs r1rIzConvNeXt.__init__[s D!!! ODEE E]H55 O#>s>ser>s>s>s:t:t OMNN N =E   $777J"$%2!$4$C! )%!%    ! I I= I I III   C%'E3>** $ $/.@DVY\D\] UU3#5{GLLMMM!# MM"-/ 0 0 0+ M" 3#566 #"4c6FTU^_``` v. +A.. !"% &/&<&HI " "iNf !- J/ 0 0"*Q--KcepAqAq   + +A!bi344 +%%ahD%9996%GNN16***  + +r3r$c||}||}||}|SrX)rrrr/s r1 _forward_implzConvNeXt._forward_impls8 MM!   LLOO OOA  r3c,||SrX)rr/s r1r2zConvNeXt.forwards!!!$$$r3)rnr>roNN)r4r5r6rr_rZrkrrr r[rrIr rr2r\r]s@r1rrZs(+!489=L+L+M*L+ %L+ L+  L+ bi01 L+Xc29n56L+L+ L+L+L+L+L+L+\v& %%F%%%%%%%%r3rrpr;weightsprogressrrr%c |)t|dt|jdt|fd|i|}|*|||d|S)Nrq categoriesr;T)r check_hash)rlenmetarload_state_dictget_state_dict)rpr;rrrrmodels r1 _convnextrszfmSl9S5T5TUUU ] Z Z:O ZSY Z ZE g44hSW4XXYYY Lr3) rzNhttps://github.com/pytorch/vision/tree/main/references/classification#convnexta These weights improve upon the results of the original paper by using a modified version of TorchVision's `new training recipe `_. )min_sizerrecipe_docsc beZdZedeeddieddddd id d d  ZeZdS)rz>https://download.pytorch.org/models/convnext_tiny-983f1562.pth crop_size resize_sizeiH< ImageNet-1KgzGT@gMbX X@zacc@1zacc@5gm@gV-G[@ num_params_metrics_ops _file_sizeurl transformsrN r4r5r6rrr _COMMON_META IMAGENET1K_V1DEFAULTr7r3r1rrsG L7.#3OOO    "##   !    M GGGr3rc beZdZedeeddieddddd id d d  ZeZdS)rz?https://download.pytorch.org/models/convnext_small-0c510722.pthrriHZrgClT@g)X@rg|?5^!@g"~g@rrNrr7r3r1rrsG M7.#3OOO    "##   !    M GGGr3rc beZdZedeeddieddddd id d d  ZeZdS)rz>https://download.pytorch.org/models/convnext_base-6075fbad.pthrrihGrgU@gHz7X@rg(\µ.@g/$!u@rrNrr7r3r1rrsG L7.#3OOO    "##   !    M GGGr3rc beZdZedeeddieddddd id d d  ZeZdS)rz?https://download.pytorch.org/models/convnext_large-ea097f82.pthrrri rg"~U@gX9v>X@rg|?5.A@gK@rrNrr7r3r1rr sG M7.#3OOO    ###   !    M GGGr3r pretrained)rT)rrc t|}tdddtdddtdddtdddg}|dd }t ||||fi|S) aConvNeXt Tiny model architecture from the `A ConvNet for the 2020s `_ paper. Args: weights (:class:`~torchvision.models.convnext.ConvNeXt_Tiny_Weights`, optional): The pretrained weights to use. See :class:`~torchvision.models.convnext.ConvNeXt_Tiny_Weights` below for more details and possible values. By default, no pre-trained weights are used. progress (bool, optional): If True, displays a progress bar of the download to stderr. Default is True. **kwargs: parameters passed to the ``torchvision.models.convnext.ConvNext`` base class. Please refer to the `source code `_ for more details about this class. .. autoclass:: torchvision.models.ConvNeXt_Tiny_Weights :members: `r' Nr;g?)rverifyr_poprrrrrrpr;s r1rr!s&$**733G b#q!!c3""c3""c4## M #JJ'>DD ]$97H W WPV W WWr3c t|}tdddtdddtdddtdddg}|dd }t ||||fi|S) aConvNeXt Small model architecture from the `A ConvNet for the 2020s `_ paper. Args: weights (:class:`~torchvision.models.convnext.ConvNeXt_Small_Weights`, optional): The pretrained weights to use. See :class:`~torchvision.models.convnext.ConvNeXt_Small_Weights` below for more details and possible values. By default, no pre-trained weights are used. progress (bool, optional): If True, displays a progress bar of the download to stderr. Default is True. **kwargs: parameters passed to the ``torchvision.models.convnext.ConvNext`` base class. Please refer to the `source code `_ for more details about this class. .. autoclass:: torchvision.models.ConvNeXt_Small_Weights :members: rrr'rrNr;g?)rrr_rrrs r1rr@s*%++G44G b#q!!c3""c3##c4## M #JJ'>DD ]$97H W WPV W WWr3c t|}tdddtdddtdddtdddg}|dd }t ||||fi|S) aConvNeXt Base model architecture from the `A ConvNet for the 2020s `_ paper. Args: weights (:class:`~torchvision.models.convnext.ConvNeXt_Base_Weights`, optional): The pretrained weights to use. See :class:`~torchvision.models.convnext.ConvNeXt_Base_Weights` below for more details and possible values. By default, no pre-trained weights are used. progress (bool, optional): If True, displays a progress bar of the download to stderr. Default is True. **kwargs: parameters passed to the ``torchvision.models.convnext.ConvNext`` base class. Please refer to the `source code `_ for more details about this class. .. autoclass:: torchvision.models.ConvNeXt_Base_Weights :members: r'iirNr;?)rrr_rrrs r1r r as&$**733G c3""c3""c4$$dD!$$ M #JJ'>DD ]$97H W WPV W WWr3c t|}tdddtdddtdddtdddg}|dd }t ||||fi|S) aConvNeXt Large model architecture from the `A ConvNet for the 2020s `_ paper. Args: weights (:class:`~torchvision.models.convnext.ConvNeXt_Large_Weights`, optional): The pretrained weights to use. See :class:`~torchvision.models.convnext.ConvNeXt_Large_Weights` below for more details and possible values. By default, no pre-trained weights are used. progress (bool, optional): If True, displays a progress bar of the download to stderr. Default is True. **kwargs: parameters passed to the ``torchvision.models.convnext.ConvNext`` base class. Please refer to the `source code `_ for more details about this class. .. autoclass:: torchvision.models.ConvNeXt_Large_Weights :members: rrr'rirNr;r)rrr_rrrs r1r!r!s*%++G44G c3""c3""c4$$dD!$$ M #JJ'>DD ]$97H W WPV W WWr3)4 functoolsrtypingrrrrrrQr r torch.nnr r)ops.miscr rops.stochastic_depthrtransforms._presetsrutilsr_apirrr_metar_utilsrr__all__rJr#r[r9r_rrZboolrrrrrrrrrr r!r7r3r1rs":::::::::::::: $$$$$$44444444222222555555''''''6666666666''''''BBBBBBBB   ",bi>))))))))*V%V%V%V%V%ryV%V%V%r & k "     &&^   K([(K([(,0E0S!TUUU@DW[XXXh'<=XPTXgjXowXXXVUX:,0F0T!UVVV37$XXX/0XCGXZ]X XXXWVX>,0E0S!TUUU@DW[XXXh'<=XPTXgjXowXXXVUX:,0F0T!UVVV37$XXX/0XCGXZ]X XXXWVXXXr3