U (d@sUddlZddlmZddlmZddlmZmZmZm Z m Z m Z ddl Z ddl m Z mZddlmZmZddlmZmZdd lmZd d lmZmZd d lmZd d lmZmZmZddddddddddddddddddd d!d"d#d$d%d&d'd(d)d*d+d,gZ Gd-d.d.eZ!Gd/d0d0e j"Z#Gd1d2d2e j$Z%Gd3d4d4e j"Z&Gd5d6d6Z'Gd7dde j$Z(e'e ee)ee(d8d9d:Z*d;ed<Z+ee,efe-d=<e+d>d?d@Z.GdAddeZ/GdBddeZ0GdCddeZ1GdDddeZ2GdEddeZ3GdFddeZ4GdGddeZ5GdHddeZ6GdIddeZ7GdJddeZ8GdKddeZ9GdLddeZ:GdMddeZ;GdNddeZfdQddRdSe e/e)ee(dTdUdZ?edPe0j>fdQddRdSe e0e)ee(dTdVdZ@edPe1j>fdQddRdSe e1e)ee(dTdWd ZAedPe2j>fdQddRdSe e2e)ee(dTdXd!ZBedPe3j>fdQddRdSe e3e)ee(dTdYd"ZCedPe4j>fdQddRdSe e4e)ee(dTdZd#ZDedPe5j>fdQddRdSe e5e)ee(dTd[d$ZEed\dQddRdSe e6e)ee(dTd]d%ZFedPe7j>fdQddRdSe e7e)ee(dTd^d&ZGedPe8j>fdQddRdSe e8e)ee(dTd_d'ZHedPe9j>fdQddRdSe e9e)ee(dTd`d(ZIedPe:j>fdQddRdSe e:e)ee(dTdad)ZJedPe;j>fdQddRdSe e;e)ee(dTdbd*ZKedPefdQddRdSe e<e)ee(dTdcd+ZLedPe=j>fdQddRdSe e=e)ee(dTddd,ZMd delmNZNeNe/j>jOe0j>jOe1j>jOe2j>jOe3j>jOe4j>jOe5j>jOe7j>jOe8j>jOe9j>jOe:j>jOe;j>jOejOe=j>jOdfZPdS)gN) OrderedDict)partial)AnyCallableDictListOptionalTuple)nnTensor)Conv2dNormActivationSqueezeExcitation)ImageClassificationInterpolationMode)_log_api_usage_once) WeightsEnumWeights)_IMAGENET_CATEGORIES)handle_legacy_interface_ovewrite_named_param_make_divisibleRegNetRegNet_Y_400MF_WeightsRegNet_Y_800MF_WeightsRegNet_Y_1_6GF_WeightsRegNet_Y_3_2GF_WeightsRegNet_Y_8GF_WeightsRegNet_Y_16GF_WeightsRegNet_Y_32GF_WeightsRegNet_Y_128GF_WeightsRegNet_X_400MF_WeightsRegNet_X_800MF_WeightsRegNet_X_1_6GF_WeightsRegNet_X_3_2GF_WeightsRegNet_X_8GF_WeightsRegNet_X_16GF_WeightsRegNet_X_32GF_Weightsregnet_y_400mfregnet_y_800mfregnet_y_1_6gfregnet_y_3_2gf regnet_y_8gf regnet_y_16gf regnet_y_32gfregnet_y_128gfregnet_x_400mfregnet_x_800mfregnet_x_1_6gfregnet_x_3_2gf regnet_x_8gf regnet_x_16gf regnet_x_32gfcsBeZdZdZeeedejfedejfddfdd ZZ S) SimpleStemINz(Simple stem for ImageNet: 3x3, BN, ReLU..N)width_in width_out norm_layeractivation_layerreturncstj||dd||ddS)Nr  kernel_sizestrider;r<)super__init__)selfr9r:r;r< __class__=/tmp/pip-unpacked-wheel-vx7f76es/torchvision/models/regnet.pyrC7szSimpleStemIN.__init__) __name__ __module__ __qualname____doc__intrr ModulerC __classcell__rGrGrErHr84s  r8c sNeZdZdZeeeedejfedejfeee edd fdd Z Z S)BottleneckTransformz/Bottleneck transformation: 1x1, 3x3 [+SE], 1x1..N r9r:rAr;r< group_widthbottleneck_multiplierse_ratior=c st} tt||} | |} t|| dd||d| d<t| | d|| ||d| d<|rvtt||} t| | |d| d<t| |dd|dd| d <t| dS) Nrr?ar>)r@rAgroupsr;r<b)Zinput_channelsZsqueeze_channels activationsec)rrMroundr rrBrC) rDr9r:rAr;r<rRrSrTZlayersZw_bgZ width_se_outrErGrHrCFsH     zBottleneckTransform.__init__ rIrJrKrLrMrr rNfloatrrCrOrGrGrErHrPCs  rPc s`eZdZdZd eeeedejfedejfeee edd fdd Z e e d d d Z Z S) ResBottleneckBlockz>Residual bottleneck block: x + F(x), F = bottleneck transform.r?N.rQc sbtd|_||kp|dk} | r:t||d||dd|_t|||||||||_|dd|_dS)Nrr?T)Zinplace)rBrCprojr rPfrX) rDr9r:rAr;r<rRrSrTZ should_projrErGrHrCos.  zResBottleneckBlock.__init__xr=cCs8|jdk r ||||}n|||}||SN)rarbrXrDrdrGrGrHforwards zResBottleneckBlock.forward)rr`N)rIrJrKrLrMrr rNr^rrCr rgrOrGrGrErHr_ls     r_cs`eZdZdZdeeeeedejfedejfedejfeee eedd fdd Z Z S) AnyStagez;AnyNet stage (sequence of blocks w/ the same output shape).Nr.) r9r:rAdepthblock_constructorr;r<rRrSrT stage_indexr=c  sbtt|D]J} || dkr$|n||| dkr4|nd|||| | } |d| d| | qdS)Nrrblock-)rBrCrangeZ add_module)rDr9r:rArirjr;r<rRrSrTrkirlrErGrHrCs   zAnyStage.__init__)Nrr]rGrGrErHrhs"    rhc @seZdZdeeeeeeeeeeeeddddZedeeeeeeeee dd ddZ d d Z e eeeeeee eeeefd d d ZdS) BlockParamsN)depthswidths group_widthsbottleneck_multipliersstridesrTr=cCs(||_||_||_||_||_||_dSrerqrrrsrtrurT)rDrqrrrsrtrurTrGrGrHrCs zBlockParams.__init__r`) riw_0w_aw_mrRrSrTkwargsr=c KsTd} d} |dks,|dks,|dks,|ddkr4tdt|||} tt| |t|} tt|t|| | |  } t t | }t | dgdg| | dgdg| }dd|D}ddt | |d d D}t td dt|D }| g|}|g|}|g|}||||\}}|||||||d S) a( Programatically compute all the per-block settings, given the RegNet parameters. The first step is to compute the quantized linear block parameters, in log space. Key parameters are: - `w_a` is the width progression slope - `w_0` is the initial width - `w_m` is the width stepping in the log space In other terms `log(block_width) = log(w_0) + w_m * block_capacity`, with `bock_capacity` ramping up following the w_0 and w_a params. This block width is finally quantized to multiples of 8. The second step is to compute the parameters per stage, taking into account the skip connection and the final 1x1 convolutions. We use the fact that the output width is constant within a stage. r rrzInvalid RegNet settingscSs$g|]\}}}}||kp||kqSrGrG).0wZwprrprGrGrH s z0BlockParams.from_init_params..cSsg|]\}}|r|qSrGrG)r|r}trGrGrHrsNcSsg|]\}}|r|qSrGrG)r|drrGrGrHrsrv) ValueErrortorchZaranger[logmathdividepowrMtolistlensetzipZdiffZtensor enumerate"_adjust_widths_groups_compatibilty)clsrirwrxryrRrSrTrzZQUANTZSTRIDEZ widths_contZblock_capacityZ block_widthsZ num_stagesZ split_helpersplits stage_widthsZ stage_depthsrurtrsrGrGrHfrom_init_paramssB $* &   zBlockParams.from_init_paramscCst|j|j|j|j|jSre)rrrrurqrsrt)rDrGrGrH_get_expanded_paramssz BlockParams._get_expanded_params)rbottleneck_ratiosrsr=cCsXddt||D}ddt||D}ddt||D}ddt||D}||fS)zl Adjusts the compatibility of widths and groups, depending on the bottleneck ratio. cSsg|]\}}t||qSrGrM)r|r}rWrGrGrHrszBBlockParams._adjust_widths_groups_compatibilty..cSsg|]\}}t||qSrG)min)r|r\w_botrGrGrHr scSsg|]\}}t||qSrG)r)r|rr\rGrGrHr#scSsg|]\}}t||qSrGr)r|rrWrGrGrHr$s)r)rrrsrrZgroup_widths_minZws_botrGrGrHrs z.BlockParams._adjust_widths_groups_compatibilty)N)r`N)rIrJrKrrMr^rrC classmethodrrr staticmethodr rrGrGrGrHrps> Hrpc szeZdZd eeeeedejfeedejfeedejfeedejfddfdd Z e e dd d Z Z S) r N.) block_params num_classes stem_width stem_type block_typer;rXr=cstt||dkrt}|dkr,tj}|dkr8t}|dkrFtj}|d||||_|}g} t | D]N\} \} } } }}| d| dt || | | ||||||j | dd f| }qjtt| |_td|_tj||d|_|D]}t|tjr:|jd|jd|j}tjj|jdtd |d qt|tjrftj|jtj |j!qt|tjrtjj|jdd d tj |j!qdS) Nr>rlr)rkrr)Z in_featuresZ out_featuresrg@)ZmeanZstdg{Gz?)"rBrCrr8r BatchNorm2dr_ZReLUstemrrappendrhrT Sequentialr trunk_outputZAdaptiveAvgPool2davgpoolZLinearfcmodules isinstanceZConv2dr@Z out_channelsinitZnormal_ZweightrsqrtZones_Zzeros_Zbias)rDrrrrrr;rXZ current_widthblocksror:rArirRrSmZfan_outrErGrHrC)sp       zRegNet.__init__rccCs8||}||}||}|jdd}||}|S)Nr)Z start_dim)rrrflattenrrfrGrGrHrgxs      zRegNet.forward)rrNNNN) rIrJrKrprMrrr rNrCr rgrOrGrGrErHr(s"O)rweightsprogressrzr=cKsh|dk rt|dt|jd|dttjddd}t|fd|i|}|dk rd||j |d|S)Nr categoriesr;gh㈵>g?)ZepsZmomentum)r) rrmetapoprr rrZload_state_dictZget_state_dict)rrrrzr;modelrGrGrH_regnetsrr)Zmin_sizer _COMMON_METAz(https://github.com/facebookresearch/SWAGz:https://github.com/facebookresearch/SWAG/blob/main/LICENSE)recipelicensec @sneZdZedeeddedddddd id d d Zed eedddedddddd idd d ZeZ dS)rz?https://download.pytorch.org/models/regnet_y_400mf-c65dace8.pth crop_sizeiPIBRhttps://github.com/pytorch/vision/tree/main/references/classification#small-models ImageNet-1KgmR@gV@zacc@1zacc@5XThese weights reproduce closely the results of the paper using a simple training recipe. num_paramsr_metrics_docsurlZ transformsrz?https://download.pytorch.org/models/regnet_y_400mf-e6988f5f.pthr resize_size8https://github.com/pytorch/vision/issues/3995#new-recipeg~jtR@gsh|/W@$ These weights improve upon the results of the original paper by using a modified version of TorchVision's `new training recipe `_. N rIrJrKrrrr IMAGENET1K_V1 IMAGENET1K_V2DEFAULTrGrGrGrHrs:  c @sneZdZedeeddedddddd id d d Zed eedddedddddd idd d ZeZ dS)rz?https://download.pytorch.org/models/regnet_y_800mf-1b27b58c.pthrri'brrg{GS@gbX9HW@rrrrz?https://download.pytorch.org/models/regnet_y_800mf-58fc7688.pthrrrgES@g㥛 W@rNrrGrGrGrHrs:  c @sneZdZedeeddedddddd id d d Zed eedddedddddd idd d ZeZ dS)rz?https://download.pytorch.org/models/regnet_y_1_6gf-b11a554e.pthrri~rrg|S@g}W@rrrrz?https://download.pytorch.org/models/regnet_y_1_6gf-0d7bc02a.pthrrrgMb8T@g#~jW@rNrrGrGrGrHrs:  c @sneZdZedeeddedddddd id d d Zed eedddedddddd idd d ZeZ dS)rz?https://download.pytorch.org/models/regnet_y_3_2gf-b5a9779c.pthrri2(Shttps://github.com/pytorch/vision/tree/main/references/classification#medium-modelsrg&1S@g/ݤW@rrrrz?https://download.pytorch.org/models/regnet_y_3_2gf-9180c971.pthrrrg+~T@g|?5W@rNrrGrGrGrHrs:  c @sneZdZedeeddedddddd id d d Zed eedddedddddd idd d ZeZ dS)rz=https://download.pytorch.org/models/regnet_y_8gf-d0d0e4a8.pthrriXrrg5^I T@gPnW@rrrrz=https://download.pytorch.org/models/regnet_y_8gf-dc2b1b54.pthrrrgET@gQX@rNrrGrGrGrHrAs:  c @seZdZedeeddedddddd id d d Zed eedddedddddd idd d Zedeedde j de ddddd iddd Z edeedde j de dddddd idd d Z eZd!S)"rz>https://download.pytorch.org/models/regnet_y_16gf-9e6ed7dd.pthrri{Rhttps://github.com/pytorch/vision/tree/main/references/classification#large-modelsrgB`"T@g(\W@rrrrz>https://download.pytorch.org/models/regnet_y_16gf-3e4a00f9.pthrrrgbX9T@gEX@rzChttps://download.pytorch.org/models/regnet_y_16gf_swag-43afe44d.pthrr interpolationgT㥛ĀU@g~jtX@ These weights are learnt via transfer learning by end-to-end fine-tuning the original `SWAG `_ weights on ImageNet-1K data. rrrzFhttps://download.pytorch.org/models/regnet_y_16gf_lc_swag-f3ec0043.pth+https://github.com/pytorch/vision/pull/5793gX9vT@gV-OX@ These weights are composed of the original frozen `SWAG `_ trunk weights and a linear classifier learnt on top of them trained on ImageNet-1K data. rrrrNrIrJrKrrrrrrrBICUBIC_COMMON_SWAG_METAIMAGENET1K_SWAG_E2E_V1IMAGENET1K_SWAG_LINEAR_V1rrGrGrGrHris  c @seZdZedeeddedddddd id d d Zed eedddedddddd idd d Zedeedde j de ddddd iddd Z edeedde j de dddddd idd d Z eZd!S)"r z>https://download.pytorch.org/models/regnet_y_32gf-4dee3f7a.pthrri<rrgx&18T@g(\W@rrrrz>https://download.pytorch.org/models/regnet_y_32gf-8db6d4b5.pthrrrgd;OT@gZd;X@rzChttps://download.pytorch.org/models/regnet_y_32gf_swag-04fdfa75.pthrrgʡU@gI +X@rrzFhttps://download.pytorch.org/models/regnet_y_32gf_lc_swag-e1583746.pthrg+'U@gQ^X@rrNrrGrGrGrHr s  c @sveZdZedeeddejdedddddid d d Z ed eed d ejdeddddddiddd Z e Z dS)r!zDhttps://download.pytorch.org/models/regnet_y_128gf_swag-c8ce3e52.pthrri^o&rg;OV@gS㥫X@rrrrzGhttps://download.pytorch.org/models/regnet_y_128gf_lc_swag-cbe8ce12.pthrrg1ZU@gtvX@rrN) rIrJrKrrrrrrrrrrGrGrGrHr! sLc @sneZdZedeeddedddddd id d d Zed eedddedddddd idd d ZeZ dS)r"z?https://download.pytorch.org/models/regnet_x_400mf-adf1edd5.pthrriSrrgL7A`5R@g̼V@rrrrz?https://download.pytorch.org/models/regnet_x_400mf-62229a5f.pthrrDhttps://github.com/pytorch/vision/issues/3995#new-recipe-with-fixresgKR@gS㥛W@rNrrGrGrGrHr"8s:  c @sneZdZedeeddedddddd id d d Zed eedddedddddd idd d ZeZ dS)r#z?https://download.pytorch.org/models/regnet_x_800mf-ad17e45c.pthrrinrrg!rhR@gʡEW@rrrrz?https://download.pytorch.org/models/regnet_x_800mf-94a99ebd.pthrrrg rhaS@g/tW@rNrrGrGrGrHr#`s:  c @sneZdZedeeddedddddd id d d Zed eedddedddddd idd d ZeZ dS)r$z?https://download.pytorch.org/models/regnet_x_1_6gf-e3633e7f.pthrri:rrg(\BS@g\(\W@rrrrz?https://download.pytorch.org/models/regnet_x_1_6gf-a12f2b72.pthrrrgnS@g^I W@rNrrGrGrGrHr$s:  c @sneZdZedeeddedddddd id d d Zed eedddedddddd idd d ZeZ dS)r%z?https://download.pytorch.org/models/regnet_x_3_2gf-f342aeae.pthrri(hrrgKS@gsh|W@rrrrz?https://download.pytorch.org/models/regnet_x_3_2gf-7071aa85.pthrrrgCLT@gQW@rNrrGrGrGrHr%s:  c @sneZdZedeeddedddddd id d d Zed eedddedddddd idd d ZeZ dS)r&z=https://download.pytorch.org/models/regnet_x_8gf-03ceed89.pthrri[rrgtS@gClW@rrrrz=https://download.pytorch.org/models/regnet_x_8gf-2b70d774.pthrrrgSkT@gZdW@rNrrGrGrGrHr&s:  c @sneZdZedeeddedddddd id d d Zed eedddedddddd idd d ZeZ dS)r'z>https://download.pytorch.org/models/regnet_x_16gf-2007eb11.pthrri9<rrgʡET@g#~jW@rrrrz>https://download.pytorch.org/models/regnet_x_16gf-ba3796d7.pthrrrgҭT@gC X@rNrrGrGrGrHr's:  c @sneZdZedeeddedddddd id d d Zed eedddedddddd idd d ZeZ dS)r(z>https://download.pytorch.org/models/regnet_x_32gf-9d47f8d0.pthrrimrrg+'T@gZd;W@rrrrz>https://download.pytorch.org/models/regnet_x_32gf-6eb8fdc6.pthrrrg7A`T@gOnX@rNrrGrGrGrHr((s:   pretrained)rT)rr)rrrzr=c Ks8t|}tjfddddddd|}t|||f|S)a Constructs a RegNetY_400MF architecture from `Designing Network Design Spaces `_. Args: weights (:class:`~torchvision.models.RegNet_Y_400MF_Weights`, optional): The pretrained weights to use. See :class:`~torchvision.models.RegNet_Y_400MF_Weights` below for more details and possible values. By default, no pretrained weights are used. progress (bool, optional): If True, displays a progress bar of the download to stderr. Default is True. **kwargs: parameters passed to either ``torchvision.models.regnet.RegNet`` or ``torchvision.models.regnet.BlockParams`` class. Please refer to the `source code `_ for more detail about the classes. .. autoclass:: torchvision.models.RegNet_Y_400MF_Weights :members: 0gp= ;@gQ@r{?rirwrxryrRrT)rverifyrprrrrrzparamsrGrGrHr)Ps c Ks8t|}tjfddddddd|}t|||f|S)a Constructs a RegNetY_800MF architecture from `Designing Network Design Spaces `_. Args: weights (:class:`~torchvision.models.RegNet_Y_800MF_Weights`, optional): The pretrained weights to use. See :class:`~torchvision.models.RegNet_Y_800MF_Weights` below for more details and possible values. By default, no pretrained weights are used. progress (bool, optional): If True, displays a progress bar of the download to stderr. Default is True. **kwargs: parameters passed to either ``torchvision.models.regnet.RegNet`` or ``torchvision.models.regnet.BlockParams`` class. Please refer to the `source code `_ for more detail about the classes. .. autoclass:: torchvision.models.RegNet_Y_800MF_Weights :members: 8gQkC@g333333@rrr)rrrprrrrGrGrHr*is c Ks8t|}tjfddddddd|}t|||f|S)a Constructs a RegNetY_1.6GF architecture from `Designing Network Design Spaces `_. Args: weights (:class:`~torchvision.models.RegNet_Y_1_6GF_Weights`, optional): The pretrained weights to use. See :class:`~torchvision.models.RegNet_Y_1_6GF_Weights` below for more details and possible values. By default, no pretrained weights are used. progress (bool, optional): If True, displays a progress bar of the download to stderr. Default is True. **kwargs: parameters passed to either ``torchvision.models.regnet.RegNet`` or ``torchvision.models.regnet.BlockParams`` class. Please refer to the `source code `_ for more detail about the classes. .. autoclass:: torchvision.models.RegNet_Y_1_6GF_Weights :members: rg(\µ4@g333333@rr)rrrprrrrGrGrHr+s c Ks8t|}tjfddddddd|}t|||f|S)a Constructs a RegNetY_3.2GF architecture from `Designing Network Design Spaces `_. Args: weights (:class:`~torchvision.models.RegNet_Y_3_2GF_Weights`, optional): The pretrained weights to use. See :class:`~torchvision.models.RegNet_Y_3_2GF_Weights` below for more details and possible values. By default, no pretrained weights are used. progress (bool, optional): If True, displays a progress bar of the download to stderr. Default is True. **kwargs: parameters passed to either ``torchvision.models.regnet.RegNet`` or ``torchvision.models.regnet.BlockParams`` class. Please refer to the `source code `_ for more detail about the classes. .. autoclass:: torchvision.models.RegNet_Y_3_2GF_Weights :members: Pgq= ףPE@gHzG@rrr)rrrprrrrGrGrHr,s c Ks8t|}tjfddddddd|}t|||f|S)a Constructs a RegNetY_8GF architecture from `Designing Network Design Spaces `_. Args: weights (:class:`~torchvision.models.RegNet_Y_8GF_Weights`, optional): The pretrained weights to use. See :class:`~torchvision.models.RegNet_Y_8GF_Weights` below for more details and possible values. By default, no pretrained weights are used. progress (bool, optional): If True, displays a progress bar of the download to stderr. Default is True. **kwargs: parameters passed to either ``torchvision.models.regnet.RegNet`` or ``torchvision.models.regnet.BlockParams`` class. Please refer to the `source code `_ for more detail about the classes. .. autoclass:: torchvision.models.RegNet_Y_8GF_Weights :members: gGz4S@gQ@rrr)rrrprrrrGrGrHr-s c Ks8t|}tjfddddddd|}t|||f|S)a Constructs a RegNetY_16GF architecture from `Designing Network Design Spaces `_. Args: weights (:class:`~torchvision.models.RegNet_Y_16GF_Weights`, optional): The pretrained weights to use. See :class:`~torchvision.models.RegNet_Y_16GF_Weights` below for more details and possible values. By default, no pretrained weights are used. progress (bool, optional): If True, displays a progress bar of the download to stderr. Default is True. **kwargs: parameters passed to either ``torchvision.models.regnet.RegNet`` or ``torchvision.models.regnet.BlockParams`` class. Please refer to the `source code `_ for more detail about the classes. .. autoclass:: torchvision.models.RegNet_Y_16GF_Weights :members: gQZ@gףp= @prr)rrrprrrrGrGrHr.s c Ks8t|}tjfddddddd|}t|||f|S)a Constructs a RegNetY_32GF architecture from `Designing Network Design Spaces `_. Args: weights (:class:`~torchvision.models.RegNet_Y_32GF_Weights`, optional): The pretrained weights to use. See :class:`~torchvision.models.RegNet_Y_32GF_Weights` below for more details and possible values. By default, no pretrained weights are used. progress (bool, optional): If True, displays a progress bar of the download to stderr. Default is True. **kwargs: parameters passed to either ``torchvision.models.regnet.RegNet`` or ``torchvision.models.regnet.BlockParams`` class. Please refer to the `source code `_ for more detail about the classes. .. autoclass:: torchvision.models.RegNet_Y_32GF_Weights :members: rg)\\@g= ףp=@rr)r rrprrrrGrGrHr/s )rNc Ks8t|}tjfddddddd|}t|||f|S)a Constructs a RegNetY_128GF architecture from `Designing Network Design Spaces `_. Args: weights (:class:`~torchvision.models.RegNet_Y_128GF_Weights`, optional): The pretrained weights to use. See :class:`~torchvision.models.RegNet_Y_128GF_Weights` below for more details and possible values. By default, no pretrained weights are used. progress (bool, optional): If True, displays a progress bar of the download to stderr. Default is True. **kwargs: parameters passed to either ``torchvision.models.regnet.RegNet`` or ``torchvision.models.regnet.BlockParams`` class. Please refer to the `source code `_ for more detail about the classes. .. autoclass:: torchvision.models.RegNet_Y_128GF_Weights :members: rig(\d@g)\(@irr)r!rrprrrrGrGrHr0 s cKs6t|}tjfdddddd|}t|||f|S)a Constructs a RegNetX_400MF architecture from `Designing Network Design Spaces `_. Args: weights (:class:`~torchvision.models.RegNet_X_400MF_Weights`, optional): The pretrained weights to use. See :class:`~torchvision.models.RegNet_X_400MF_Weights` below for more details and possible values. By default, no pretrained weights are used. progress (bool, optional): If True, displays a progress bar of the download to stderr. Default is True. **kwargs: parameters passed to either ``torchvision.models.regnet.RegNet`` or ``torchvision.models.regnet.BlockParams`` class. Please refer to the `source code `_ for more detail about the classes. .. autoclass:: torchvision.models.RegNet_X_400MF_Weights :members: rg{Gz8@gRQ@rrirwrxryrR)r"rrprrrrGrGrHr1$s cKs6t|}tjfdddddd|}t|||f|S)a Constructs a RegNetX_800MF architecture from `Designing Network Design Spaces `_. Args: weights (:class:`~torchvision.models.RegNet_X_800MF_Weights`, optional): The pretrained weights to use. See :class:`~torchvision.models.RegNet_X_800MF_Weights` below for more details and possible values. By default, no pretrained weights are used. progress (bool, optional): If True, displays a progress bar of the download to stderr. Default is True. **kwargs: parameters passed to either ``torchvision.models.regnet.RegNet`` or ``torchvision.models.regnet.BlockParams`` class. Please refer to the `source code `_ for more detail about the classes. .. autoclass:: torchvision.models.RegNet_X_800MF_Weights :members: rrg= ףpA@g= ףp=@r)r#rrprrrrGrGrHr2=s cKs6t|}tjfdddddd|}t|||f|S)aX Constructs a RegNetX_1.6GF architecture from `Designing Network Design Spaces `_. Args: weights (:class:`~torchvision.models.RegNet_X_1_6GF_Weights`, optional): The pretrained weights to use. See :class:`~torchvision.models.RegNet_X_1_6GF_Weights` below for more details and possible values. By default, no pretrained weights are used. progress (bool, optional): If True, displays a progress bar of the download to stderr. Default is True. **kwargs: parameters passed to either ``torchvision.models.regnet.RegNet`` or ``torchvision.models.regnet.BlockParams`` class. Please refer to the `source code `_ for more detail about the classes. .. autoclass:: torchvision.models.RegNet_X_1_6GF_Weights :members: Args: weights (RegNet_X_1_6GF_Weights, optional): The pretrained weights for the model progress (bool): If True, displays a progress bar of the download to stderr rrgzGA@@rr)r$rrprrrrGrGrHr3Vs cKs6t|}tjfdddddd|}t|||f|S)aX Constructs a RegNetX_3.2GF architecture from `Designing Network Design Spaces `_. Args: weights (:class:`~torchvision.models.RegNet_X_3_2GF_Weights`, optional): The pretrained weights to use. See :class:`~torchvision.models.RegNet_X_3_2GF_Weights` below for more details and possible values. By default, no pretrained weights are used. progress (bool, optional): If True, displays a progress bar of the download to stderr. Default is True. **kwargs: parameters passed to either ``torchvision.models.regnet.RegNet`` or ``torchvision.models.regnet.BlockParams`` class. Please refer to the `source code `_ for more detail about the classes. .. autoclass:: torchvision.models.RegNet_X_3_2GF_Weights :members: Args: weights (RegNet_X_3_2GF_Weights, optional): The pretrained weights for the model progress (bool): If True, displays a progress bar of the download to stderr Xg(\O:@rrr)r%rrprrrrGrGrHr4ss cKs6t|}tjfdddddd|}t|||f|S)aN Constructs a RegNetX_8GF architecture from `Designing Network Design Spaces `_. Args: weights (:class:`~torchvision.models.RegNet_X_8GF_Weights`, optional): The pretrained weights to use. See :class:`~torchvision.models.RegNet_X_8GF_Weights` below for more details and possible values. By default, no pretrained weights are used. progress (bool, optional): If True, displays a progress bar of the download to stderr. Default is True. **kwargs: parameters passed to either ``torchvision.models.regnet.RegNet`` or ``torchvision.models.regnet.BlockParams`` class. Please refer to the `source code `_ for more detail about the classes. .. autoclass:: torchvision.models.RegNet_X_8GF_Weights :members: Args: weights (RegNet_X_8GF_Weights, optional): The pretrained weights for the model progress (bool): If True, displays a progress bar of the download to stderr rgHzH@g ףp= @xr)r&rrprrrrGrGrHr5s cKs6t|}tjfdddddd|}t|||f|S)aS Constructs a RegNetX_16GF architecture from `Designing Network Design Spaces `_. Args: weights (:class:`~torchvision.models.RegNet_X_16GF_Weights`, optional): The pretrained weights to use. See :class:`~torchvision.models.RegNet_X_16GF_Weights` below for more details and possible values. By default, no pretrained weights are used. progress (bool, optional): If True, displays a progress bar of the download to stderr. Default is True. **kwargs: parameters passed to either ``torchvision.models.regnet.RegNet`` or ``torchvision.models.regnet.BlockParams`` class. Please refer to the `source code `_ for more detail about the classes. .. autoclass:: torchvision.models.RegNet_X_16GF_Weights :members: Args: weights (RegNet_X_16GF_Weights, optional): The pretrained weights for the model progress (bool): If True, displays a progress bar of the download to stderr rgQK@g@r)r'rrprrrrGrGrHr6s cKs6t|}tjfdddddd|}t|||f|S)aS Constructs a RegNetX_32GF architecture from `Designing Network Design Spaces `_. Args: weights (:class:`~torchvision.models.RegNet_X_32GF_Weights`, optional): The pretrained weights to use. See :class:`~torchvision.models.RegNet_X_32GF_Weights` below for more details and possible values. By default, no pretrained weights are used. progress (bool, optional): If True, displays a progress bar of the download to stderr. Default is True. **kwargs: parameters passed to either ``torchvision.models.regnet.RegNet`` or ``torchvision.models.regnet.BlockParams`` class. Please refer to the `source code `_ for more detail about the classes. .. autoclass:: torchvision.models.RegNet_X_32GF_Weights :members: Args: weights (RegNet_X_32GF_Weights, optional): The pretrained weights for the model progress (bool): If True, displays a progress bar of the download to stderr ri@gףp= wQ@rr)r(rrprrrrGrGrHr7s ) _ModelURLs)r)r*r+r,r-r.r/r1r2r3r4r5r6r7)Qr collectionsr functoolsrtypingrrrrrr rr r Zops.miscr rZtransforms._presetsrrutilsrZ_apirrZ_metar_utilsrrr__all__r8rrPrNr_rhrprboolrrstr__annotations__rrrrrrrr r!r"r#r$r%r&r'r(rr)r*r+r,r-r.r/r0r1r2r3r4r5r6r7rrZ model_urlsrGrGrGrHs     #)+"o\ (((((QQ-((((((("""""""""""""""