timm/models/regnet.py

"""RegNet X, Y, Z, and more

Paper: `Designing Network Design Spaces` - https://arxiv.org/abs/2003.13678
Original Impl: https://github.com/facebookresearch/pycls/blob/master/pycls/models/regnet.py

Paper: `Fast and Accurate Model Scaling` - https://arxiv.org/abs/2103.06877
Original Impl: None

Based on original PyTorch impl linked above, but re-wrote to use my own blocks (adapted from ResNet here)
and cleaned up with more descriptive variable names.

Weights from original pycls impl have been modified:
* first layer from BGR -> RGB as most PyTorch models are
* removed training specific dict entries from checkpoints and keep model state_dict only
* remap names to match the ones here

Supports weight loading from torchvision and classy-vision (incl VISSL SEER)

A number of custom timm model definitions additions including:
* stochastic depth, gradient checkpointing, layer-decay, configurable dilation
* a pre-activation 'V' variant
* only known RegNet-Z model definitions with pretrained weights

Hacked together by / Copyright 2020 Ross Wightman
"""
import math
from dataclasses import dataclass, replace
from functools import partial
from typing import Optional, Union, Callable

import numpy as np
import torch
import torch.nn as nn

from timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD
from timm.layers import ClassifierHead, AvgPool2dSame, ConvNormAct, SEModule, DropPath, GroupNormAct
from timm.layers import get_act_layer, get_norm_act_layer, create_conv2d, make_divisible
from ._builder import build_model_with_cfg
from ._manipulate import checkpoint_seq, named_apply
from ._registry import generate_default_cfgs, register_model, register_model_deprecations

__all__ = ['RegNet', 'RegNetCfg']  # model_registry will add each entrypoint fn to this


@dataclass
class RegNetCfg:
    depth: int = 21
    w0: int = 80
    wa: float = 42.63
    wm: float = 2.66
    group_size: int = 24
    bottle_ratio: float = 1.
    se_ratio: float = 0.
    group_min_ratio: float = 0.
    stem_width: int = 32
    downsample: Optional[str] = 'conv1x1'
    linear_out: bool = False
    preact: bool = False
    num_features: int = 0
    act_layer: Union[str, Callable] = 'relu'
    norm_layer: Union[str, Callable] = 'batchnorm'


def quantize_float(f, q):
    """Converts a float to the closest non-zero int divisible by q."""
    return int(round(f / q) * q)


def adjust_widths_groups_comp(widths, bottle_ratios, groups, min_ratio=0.):
    """Adjusts the compatibility of widths and groups."""
    bottleneck_widths = [int(w * b) for w, b in zip(widths, bottle_ratios)]
    groups = [min(g, w_bot) for g, w_bot in zip(groups, bottleneck_widths)]
    if min_ratio:
        # torchvision uses a different rounding scheme for ensuring bottleneck widths divisible by group widths
        bottleneck_widths = [make_divisible(w_bot, g, min_ratio) for w_bot, g in zip(bottleneck_widths, groups)]
    else:
        bottleneck_widths = [quantize_float(w_bot, g) for w_bot, g in zip(bottleneck_widths, groups)]
    widths = [int(w_bot / b) for w_bot, b in zip(bottleneck_widths, bottle_ratios)]
    return widths, groups


def generate_regnet(width_slope, width_initial, width_mult, depth, group_size, quant=8):
    """Generates per block widths from RegNet parameters."""
    assert width_slope >= 0 and width_initial > 0 and width_mult > 1 and width_initial % quant == 0
    # TODO dWr scaling?
    # depth = int(depth * (scale ** 0.1))
    # width_scale = scale ** 0.4  # dWr scale, exp 0.8 / 2, applied to both group and layer widths
    widths_cont = np.arange(depth) * width_slope + width_initial
    width_exps = np.round(np.log(widths_cont / width_initial) / np.log(width_mult))
    widths = np.round(np.divide(width_initial * np.power(width_mult, width_exps), quant)) * quant
    num_stages, max_stage = len(np.unique(widths)), width_exps.max() + 1
    groups = np.array([group_size for _ in range(num_stages)])
    return widths.astype(int).tolist(), num_stages, groups.astype(int).tolist()


def downsample_conv(
        in_chs,
        out_chs,
        kernel_size=1,
        stride=1,
        dilation=1,
        norm_layer=None,
        preact=False,
):
    norm_layer = norm_layer or nn.BatchNorm2d
    kernel_size = 1 if stride == 1 and dilation == 1 else kernel_size
    dilation = dilation if kernel_size > 1 else 1
    if preact:
        return create_conv2d(
            in_chs,
            out_chs,
            kernel_size,
            stride=stride,
            dilation=dilation,
        )
    else:
        return ConvNormAct(
            in_chs,
            out_chs,
            kernel_size,
            stride=stride,
            dilation=dilation,
            norm_layer=norm_layer,
            apply_act=False,
        )


def downsample_avg(
        in_chs,
        out_chs,
        kernel_size=1,
        stride=1,
        dilation=1,
        norm_layer=None,
        preact=False,
):
    """ AvgPool Downsampling as in 'D' ResNet variants. This is not in RegNet space but I might experiment."""
    norm_layer = norm_layer or nn.BatchNorm2d
    avg_stride = stride if dilation == 1 else 1
    pool = nn.Identity()
    if stride > 1 or dilation > 1:
        avg_pool_fn = AvgPool2dSame if avg_stride == 1 and dilation > 1 else nn.AvgPool2d
        pool = avg_pool_fn(2, avg_stride, ceil_mode=True, count_include_pad=False)
    if preact:
        conv = create_conv2d(in_chs, out_chs, 1, stride=1)
    else:
        conv = ConvNormAct(in_chs, out_chs, 1, stride=1, norm_layer=norm_layer, apply_act=False)
    return nn.Sequential(*[pool, conv])


def create_shortcut(
        downsample_type,
        in_chs,
        out_chs,
        kernel_size,
        stride,
        dilation=(1, 1),
        norm_layer=None,
        preact=False,
):
    assert downsample_type in ('avg', 'conv1x1', '', None)
    if in_chs != out_chs or stride != 1 or dilation[0] != dilation[1]:
        dargs = dict(stride=stride, dilation=dilation[0], norm_layer=norm_layer, preact=preact)
        if not downsample_type:
            return None  # no shortcut, no downsample
        elif downsample_type == 'avg':
            return downsample_avg(in_chs, out_chs, **dargs)
        else:
            return downsample_conv(in_chs, out_chs, kernel_size=kernel_size, **dargs)
    else:
        return nn.Identity()  # identity shortcut (no downsample)


class Bottleneck(nn.Module):
    """ RegNet Bottleneck

    This is almost exactly the same as a ResNet Bottlneck. The main difference is the SE block is moved from
    after conv3 to after conv2. Otherwise, it's just redefining the arguments for groups/bottleneck channels.
    """

    def __init__(
            self,
            in_chs,
            out_chs,
            stride=1,
            dilation=(1, 1),
            bottle_ratio=1,
            group_size=1,
            se_ratio=0.25,
            downsample='conv1x1',
            linear_out=False,
            act_layer=nn.ReLU,
            norm_layer=nn.BatchNorm2d,
            drop_block=None,
            drop_path_rate=0.,
    ):
        super(Bottleneck, self).__init__()
        act_layer = get_act_layer(act_layer)
        bottleneck_chs = int(round(out_chs * bottle_ratio))
        groups = bottleneck_chs // group_size

        cargs = dict(act_layer=act_layer, norm_layer=norm_layer)
        self.conv1 = ConvNormAct(in_chs, bottleneck_chs, kernel_size=1, **cargs)
        self.conv2 = ConvNormAct(
            bottleneck_chs,
            bottleneck_chs,
            kernel_size=3,
            stride=stride,
            dilation=dilation[0],
            groups=groups,
            drop_layer=drop_block,
            **cargs,
        )
        if se_ratio:
            se_channels = int(round(in_chs * se_ratio))
            self.se = SEModule(bottleneck_chs, rd_channels=se_channels, act_layer=act_layer)
        else:
            self.se = nn.Identity()
        self.conv3 = ConvNormAct(bottleneck_chs, out_chs, kernel_size=1, apply_act=False, **cargs)
        self.act3 = nn.Identity() if linear_out else act_layer()
        self.downsample = create_shortcut(
            downsample,
            in_chs,
            out_chs,
            kernel_size=1,
            stride=stride,
            dilation=dilation,
            norm_layer=norm_layer,
        )
        self.drop_path = DropPath(drop_path_rate) if drop_path_rate > 0 else nn.Identity()

    def zero_init_last(self):
        nn.init.zeros_(self.conv3.bn.weight)

    def forward(self, x):
        shortcut = x
        x = self.conv1(x)
        x = self.conv2(x)
        x = self.se(x)
        x = self.conv3(x)
        if self.downsample is not None:
            # NOTE stuck with downsample as the attr name due to weight compatibility
            # now represents the shortcut, no shortcut if None, and non-downsample shortcut == nn.Identity()
            x = self.drop_path(x) + self.downsample(shortcut)
        x = self.act3(x)
        return x


class PreBottleneck(nn.Module):
    """ RegNet Bottleneck

    This is almost exactly the same as a ResNet Bottlneck. The main difference is the SE block is moved from
    after conv3 to after conv2. Otherwise, it's just redefining the arguments for groups/bottleneck channels.
    """

    def __init__(
            self,
            in_chs,
            out_chs,
            stride=1,
            dilation=(1, 1),
            bottle_ratio=1,
            group_size=1,
            se_ratio=0.25,
            downsample='conv1x1',
            linear_out=False,
            act_layer=nn.ReLU,
            norm_layer=nn.BatchNorm2d,
            drop_block=None,
            drop_path_rate=0.,
    ):
        super(PreBottleneck, self).__init__()
        norm_act_layer = get_norm_act_layer(norm_layer, act_layer)
        bottleneck_chs = int(round(out_chs * bottle_ratio))
        groups = bottleneck_chs // group_size

        self.norm1 = norm_act_layer(in_chs)
        self.conv1 = create_conv2d(in_chs, bottleneck_chs, kernel_size=1)
        self.norm2 = norm_act_layer(bottleneck_chs)
        self.conv2 = create_conv2d(
            bottleneck_chs,
            bottleneck_chs,
            kernel_size=3,
            stride=stride,
            dilation=dilation[0],
            groups=groups,
        )
        if se_ratio:
            se_channels = int(round(in_chs * se_ratio))
            self.se = SEModule(bottleneck_chs, rd_channels=se_channels, act_layer=act_layer)
        else:
            self.se = nn.Identity()
        self.norm3 = norm_act_layer(bottleneck_chs)
        self.conv3 = create_conv2d(bottleneck_chs, out_chs, kernel_size=1)
        self.downsample = create_shortcut(
            downsample,
            in_chs,
            out_chs,
            kernel_size=1,
            stride=stride,
            dilation=dilation,
            preact=True,
        )
        self.drop_path = DropPath(drop_path_rate) if drop_path_rate > 0 else nn.Identity()

    def zero_init_last(self):
        pass

    def forward(self, x):
        x = self.norm1(x)
        shortcut = x
        x = self.conv1(x)
        x = self.norm2(x)
        x = self.conv2(x)
        x = self.se(x)
        x = self.norm3(x)
        x = self.conv3(x)
        if self.downsample is not None:
            # NOTE stuck with downsample as the attr name due to weight compatibility
            # now represents the shortcut, no shortcut if None, and non-downsample shortcut == nn.Identity()
            x = self.drop_path(x) + self.downsample(shortcut)
        return x


class RegStage(nn.Module):
    """Stage (sequence of blocks w/ the same output shape)."""

    def __init__(
            self,
            depth,
            in_chs,
            out_chs,
            stride,
            dilation,
            drop_path_rates=None,
            block_fn=Bottleneck,
            **block_kwargs,
    ):
        super(RegStage, self).__init__()
        self.grad_checkpointing = False

        first_dilation = 1 if dilation in (1, 2) else 2
        for i in range(depth):
            block_stride = stride if i == 0 else 1
            block_in_chs = in_chs if i == 0 else out_chs
            block_dilation = (first_dilation, dilation)
            dpr = drop_path_rates[i] if drop_path_rates is not None else 0.
            name = "b{}".format(i + 1)
            self.add_module(
                name,
                block_fn(
                    block_in_chs,
                    out_chs,
                    stride=block_stride,
                    dilation=block_dilation,
                    drop_path_rate=dpr,
                    **block_kwargs,
                )
            )
            first_dilation = dilation

    def forward(self, x):
        if self.grad_checkpointing and not torch.jit.is_scripting():
            x = checkpoint_seq(self.children(), x)
        else:
            for block in self.children():
                x = block(x)
        return x


class RegNet(nn.Module):
    """RegNet-X, Y, and Z Models

    Paper: https://arxiv.org/abs/2003.13678
    Original Impl: https://github.com/facebookresearch/pycls/blob/master/pycls/models/regnet.py
    """

    def __init__(
            self,
            cfg: RegNetCfg,
            in_chans=3,
            num_classes=1000,
            output_stride=32,
            global_pool='avg',
            drop_rate=0.,
            drop_path_rate=0.,
            zero_init_last=True,
            **kwargs,
    ):
        """

        Args:
            cfg (RegNetCfg): Model architecture configuration
            in_chans (int): Number of input channels (default: 3)
            num_classes (int): Number of classifier classes (default: 1000)
            output_stride (int): Output stride of network, one of (8, 16, 32) (default: 32)
            global_pool (str): Global pooling type (default: 'avg')
            drop_rate (float): Dropout rate (default: 0.)
            drop_path_rate (float): Stochastic depth drop-path rate (default: 0.)
            zero_init_last (bool): Zero-init last weight of residual path
            kwargs (dict): Extra kwargs overlayed onto cfg
        """
        super().__init__()
        self.num_classes = num_classes
        self.drop_rate = drop_rate
        assert output_stride in (8, 16, 32)
        cfg = replace(cfg, **kwargs)  # update cfg with extra passed kwargs

        # Construct the stem
        stem_width = cfg.stem_width
        na_args = dict(act_layer=cfg.act_layer, norm_layer=cfg.norm_layer)
        if cfg.preact:
            self.stem = create_conv2d(in_chans, stem_width, 3, stride=2)
        else:
            self.stem = ConvNormAct(in_chans, stem_width, 3, stride=2, **na_args)
        self.feature_info = [dict(num_chs=stem_width, reduction=2, module='stem')]

        # Construct the stages
        prev_width = stem_width
        curr_stride = 2
        per_stage_args, common_args = self._get_stage_args(
            cfg,
            output_stride=output_stride,
            drop_path_rate=drop_path_rate,
        )
        assert len(per_stage_args) == 4
        block_fn = PreBottleneck if cfg.preact else Bottleneck
        for i, stage_args in enumerate(per_stage_args):
            stage_name = "s{}".format(i + 1)
            self.add_module(
                stage_name,
                RegStage(
                    in_chs=prev_width,
                    block_fn=block_fn,
                    **stage_args,
                    **common_args,
                )
            )
            prev_width = stage_args['out_chs']
            curr_stride *= stage_args['stride']
            self.feature_info += [dict(num_chs=prev_width, reduction=curr_stride, module=stage_name)]

        # Construct the head
        if cfg.num_features:
            self.final_conv = ConvNormAct(prev_width, cfg.num_features, kernel_size=1, **na_args)
            self.num_features = cfg.num_features
        else:
            final_act = cfg.linear_out or cfg.preact
            self.final_conv = get_act_layer(cfg.act_layer)() if final_act else nn.Identity()
            self.num_features = prev_width
        self.head = ClassifierHead(
            in_features=self.num_features,
            num_classes=num_classes,
            pool_type=global_pool,
            drop_rate=drop_rate,
        )

        named_apply(partial(_init_weights, zero_init_last=zero_init_last), self)

    def _get_stage_args(self, cfg: RegNetCfg, default_stride=2, output_stride=32, drop_path_rate=0.):
        # Generate RegNet ws per block
        widths, num_stages, stage_gs = generate_regnet(cfg.wa, cfg.w0, cfg.wm, cfg.depth, cfg.group_size)

        # Convert to per stage format
        stage_widths, stage_depths = np.unique(widths, return_counts=True)
        stage_br = [cfg.bottle_ratio for _ in range(num_stages)]
        stage_strides = []
        stage_dilations = []
        net_stride = 2
        dilation = 1
        for _ in range(num_stages):
            if net_stride >= output_stride:
                dilation *= default_stride
                stride = 1
            else:
                stride = default_stride
                net_stride *= stride
            stage_strides.append(stride)
            stage_dilations.append(dilation)
        stage_dpr = np.split(np.linspace(0, drop_path_rate, sum(stage_depths)), np.cumsum(stage_depths[:-1]))

        # Adjust the compatibility of ws and gws
        stage_widths, stage_gs = adjust_widths_groups_comp(
            stage_widths, stage_br, stage_gs, min_ratio=cfg.group_min_ratio)
        arg_names = ['out_chs', 'stride', 'dilation', 'depth', 'bottle_ratio', 'group_size', 'drop_path_rates']
        per_stage_args = [
            dict(zip(arg_names, params)) for params in
            zip(stage_widths, stage_strides, stage_dilations, stage_depths, stage_br, stage_gs, stage_dpr)
        ]
        common_args = dict(
            downsample=cfg.downsample,
            se_ratio=cfg.se_ratio,
            linear_out=cfg.linear_out,
            act_layer=cfg.act_layer,
            norm_layer=cfg.norm_layer,
        )
        return per_stage_args, common_args

    @torch.jit.ignore
    def group_matcher(self, coarse=False):
        return dict(
            stem=r'^stem',
            blocks=r'^s(\d+)' if coarse else r'^s(\d+)\.b(\d+)',
        )

    @torch.jit.ignore
    def set_grad_checkpointing(self, enable=True):
        for s in list(self.children())[1:-1]:
            s.grad_checkpointing = enable

    @torch.jit.ignore
    def get_classifier(self):
        return self.head.fc

    def reset_classifier(self, num_classes, global_pool='avg'):
        self.head.reset(num_classes, pool_type=global_pool)

    def forward_features(self, x):
        x = self.stem(x)
        x = self.s1(x)
        x = self.s2(x)
        x = self.s3(x)
        x = self.s4(x)
        x = self.final_conv(x)
        return x

    def forward_head(self, x, pre_logits: bool = False):
        return self.head(x, pre_logits=pre_logits)

    def forward(self, x):
        x = self.forward_features(x)
        x = self.forward_head(x)
        return x


def _init_weights(module, name='', zero_init_last=False):
    if isinstance(module, nn.Conv2d):
        fan_out = module.kernel_size[0] * module.kernel_size[1] * module.out_channels
        fan_out //= module.groups
        module.weight.data.normal_(0, math.sqrt(2.0 / fan_out))
        if module.bias is not None:
            module.bias.data.zero_()
    elif isinstance(module, nn.Linear):
        nn.init.normal_(module.weight, mean=0.0, std=0.01)
        if module.bias is not None:
            nn.init.zeros_(module.bias)
    elif zero_init_last and hasattr(module, 'zero_init_last'):
        module.zero_init_last()


def _filter_fn(state_dict):
    state_dict = state_dict.get('model', state_dict)
    replaces = [
        ('f.a.0', 'conv1.conv'),
        ('f.a.1', 'conv1.bn'),
        ('f.b.0', 'conv2.conv'),
        ('f.b.1', 'conv2.bn'),
        ('f.final_bn', 'conv3.bn'),
        ('f.se.excitation.0', 'se.fc1'),
        ('f.se.excitation.2', 'se.fc2'),
        ('f.se', 'se'),
        ('f.c.0', 'conv3.conv'),
        ('f.c.1', 'conv3.bn'),
        ('f.c', 'conv3.conv'),
        ('proj.0', 'downsample.conv'),
        ('proj.1', 'downsample.bn'),
        ('proj', 'downsample.conv'),
    ]
    if 'classy_state_dict' in state_dict:
        # classy-vision & vissl (SEER) weights
        import re
        state_dict = state_dict['classy_state_dict']['base_model']['model']
        out = {}
        for k, v in state_dict['trunk'].items():
            k = k.replace('_feature_blocks.conv1.stem.0', 'stem.conv')
            k = k.replace('_feature_blocks.conv1.stem.1', 'stem.bn')
            k = re.sub(
                r'^_feature_blocks.res\d.block(\d)-(\d+)',
                lambda x: f's{int(x.group(1))}.b{int(x.group(2)) + 1}', k)
            k = re.sub(r's(\d)\.b(\d+)\.bn', r's\1.b\2.downsample.bn', k)
            for s, r in replaces:
                k = k.replace(s, r)
            out[k] = v
        for k, v in state_dict['heads'].items():
            if 'projection_head' in k or 'prototypes' in k:
                continue
            k = k.replace('0.clf.0', 'head.fc')
            out[k] = v
        return out
    if 'stem.0.weight' in state_dict:
        # torchvision weights
        import re
        out = {}
        for k, v in state_dict.items():
            k = k.replace('stem.0', 'stem.conv')
            k = k.replace('stem.1', 'stem.bn')
            k = re.sub(
                r'trunk_output.block(\d)\.block(\d+)\-(\d+)',
                lambda x: f's{int(x.group(1))}.b{int(x.group(3)) + 1}', k)
            for s, r in replaces:
                k = k.replace(s, r)
            k = k.replace('fc.', 'head.fc.')
            out[k] = v
        return out
    return state_dict


# Model FLOPS = three trailing digits * 10^8
model_cfgs = dict(
    # RegNet-X
    regnetx_002=RegNetCfg(w0=24, wa=36.44, wm=2.49, group_size=8, depth=13),
    regnetx_004=RegNetCfg(w0=24, wa=24.48, wm=2.54, group_size=16, depth=22),
    regnetx_004_tv=RegNetCfg(w0=24, wa=24.48, wm=2.54, group_size=16, depth=22, group_min_ratio=0.9),
    regnetx_006=RegNetCfg(w0=48, wa=36.97, wm=2.24, group_size=24, depth=16),
    regnetx_008=RegNetCfg(w0=56, wa=35.73, wm=2.28, group_size=16, depth=16),
    regnetx_016=RegNetCfg(w0=80, wa=34.01, wm=2.25, group_size=24, depth=18),
    regnetx_032=RegNetCfg(w0=88, wa=26.31, wm=2.25, group_size=48, depth=25),
    regnetx_040=RegNetCfg(w0=96, wa=38.65, wm=2.43, group_size=40, depth=23),
    regnetx_064=RegNetCfg(w0=184, wa=60.83, wm=2.07, group_size=56, depth=17),
    regnetx_080=RegNetCfg(w0=80, wa=49.56, wm=2.88, group_size=120, depth=23),
    regnetx_120=RegNetCfg(w0=168, wa=73.36, wm=2.37, group_size=112, depth=19),
    regnetx_160=RegNetCfg(w0=216, wa=55.59, wm=2.1, group_size=128, depth=22),
    regnetx_320=RegNetCfg(w0=320, wa=69.86, wm=2.0, group_size=168, depth=23),

    # RegNet-Y
    regnety_002=RegNetCfg(w0=24, wa=36.44, wm=2.49, group_size=8, depth=13, se_ratio=0.25),
    regnety_004=RegNetCfg(w0=48, wa=27.89, wm=2.09, group_size=8, depth=16, se_ratio=0.25),
    regnety_006=RegNetCfg(w0=48, wa=32.54, wm=2.32, group_size=16, depth=15, se_ratio=0.25),
    regnety_008=RegNetCfg(w0=56, wa=38.84, wm=2.4, group_size=16, depth=14, se_ratio=0.25),
    regnety_008_tv=RegNetCfg(w0=56, wa=38.84, wm=2.4, group_size=16, depth=14, se_ratio=0.25, group_min_ratio=0.9),
    regnety_016=RegNetCfg(w0=48, wa=20.71, wm=2.65, group_size=24, depth=27, se_ratio=0.25),
    regnety_032=RegNetCfg(w0=80, wa=42.63, wm=2.66, group_size=24, depth=21, se_ratio=0.25),
    regnety_040=RegNetCfg(w0=96, wa=31.41, wm=2.24, group_size=64, depth=22, se_ratio=0.25),
    regnety_064=RegNetCfg(w0=112, wa=33.22, wm=2.27, group_size=72, depth=25, se_ratio=0.25),
    regnety_080=RegNetCfg(w0=192, wa=76.82, wm=2.19, group_size=56, depth=17, se_ratio=0.25),
    regnety_080_tv=RegNetCfg(w0=192, wa=76.82, wm=2.19, group_size=56, depth=17, se_ratio=0.25, group_min_ratio=0.9),
    regnety_120=RegNetCfg(w0=168, wa=73.36, wm=2.37, group_size=112, depth=19, se_ratio=0.25),
    regnety_160=RegNetCfg(w0=200, wa=106.23, wm=2.48, group_size=112, depth=18, se_ratio=0.25),
    regnety_320=RegNetCfg(w0=232, wa=115.89, wm=2.53, group_size=232, depth=20, se_ratio=0.25),
    regnety_640=RegNetCfg(w0=352, wa=147.48, wm=2.4, group_size=328, depth=20, se_ratio=0.25),
    regnety_1280=RegNetCfg(w0=456, wa=160.83, wm=2.52, group_size=264, depth=27, se_ratio=0.25),
    regnety_2560=RegNetCfg(w0=640, wa=230.83, wm=2.53, group_size=373, depth=27, se_ratio=0.25),
    #regnety_2560=RegNetCfg(w0=640, wa=124.47, wm=2.04, group_size=848, depth=27, se_ratio=0.25),

    # Experimental
    regnety_040_sgn=RegNetCfg(
        w0=96, wa=31.41, wm=2.24, group_size=64, depth=22, se_ratio=0.25,
        act_layer='silu', norm_layer=partial(GroupNormAct, group_size=16)),

    # regnetv = 'preact regnet y'
    regnetv_040=RegNetCfg(
        depth=22, w0=96, wa=31.41, wm=2.24, group_size=64, se_ratio=0.25, preact=True, act_layer='silu'),
    regnetv_064=RegNetCfg(
        depth=25, w0=112, wa=33.22, wm=2.27, group_size=72, se_ratio=0.25, preact=True, act_layer='silu',
        downsample='avg'),

    # RegNet-Z (unverified)
    regnetz_005=RegNetCfg(
        depth=21, w0=16, wa=10.7, wm=2.51, group_size=4, bottle_ratio=4.0, se_ratio=0.25,
        downsample=None, linear_out=True, num_features=1024, act_layer='silu',
    ),
    regnetz_040=RegNetCfg(
        depth=28, w0=48, wa=14.5, wm=2.226, group_size=8, bottle_ratio=4.0, se_ratio=0.25,
        downsample=None, linear_out=True, num_features=0, act_layer='silu',
    ),
    regnetz_040_h=RegNetCfg(
        depth=28, w0=48, wa=14.5, wm=2.226, group_size=8, bottle_ratio=4.0, se_ratio=0.25,
        downsample=None, linear_out=True, num_features=1536, act_layer='silu',
    ),
)


def _create_regnet(variant, pretrained, **kwargs):
    return build_model_with_cfg(
        RegNet, variant, pretrained,
        model_cfg=model_cfgs[variant],
        pretrained_filter_fn=_filter_fn,
        **kwargs)


def _cfg(url='', **kwargs):
    return {
        'url': url, 'num_classes': 1000, 'input_size': (3, 224, 224), 'pool_size': (7, 7),
        'test_input_size': (3, 288, 288), 'crop_pct': 0.95, 'test_crop_pct': 1.0,
        'interpolation': 'bicubic', 'mean': IMAGENET_DEFAULT_MEAN, 'std': IMAGENET_DEFAULT_STD,
        'first_conv': 'stem.conv', 'classifier': 'head.fc',
        **kwargs
    }


def _cfgpyc(url='', **kwargs):
    return {
        'url': url, 'num_classes': 1000, 'input_size': (3, 224, 224), 'pool_size': (7, 7),
        'crop_pct': 0.875, 'interpolation': 'bicubic',
        'mean': IMAGENET_DEFAULT_MEAN, 'std': IMAGENET_DEFAULT_STD,
        'first_conv': 'stem.conv', 'classifier': 'head.fc',
        'license': 'mit', 'origin_url': 'https://github.com/facebookresearch/pycls', **kwargs
    }


def _cfgtv2(url='', **kwargs):
    return {
        'url': url, 'num_classes': 1000, 'input_size': (3, 224, 224), 'pool_size': (7, 7),
        'crop_pct': 0.965, 'interpolation': 'bicubic',
        'mean': IMAGENET_DEFAULT_MEAN, 'std': IMAGENET_DEFAULT_STD,
        'first_conv': 'stem.conv', 'classifier': 'head.fc',
        'license': 'bsd-3-clause', 'origin_url': 'https://github.com/pytorch/vision', **kwargs
    }


default_cfgs = generate_default_cfgs({
    # timm trained models
    'regnety_032.ra_in1k': _cfg(
        hf_hub_id='timm/',
        url='https://github.com/huggingface/pytorch-image-models/releases/download/v0.1-weights/regnety_032_ra-7f2439f9.pth'),
    'regnety_040.ra3_in1k': _cfg(
        hf_hub_id='timm/',
        url='https://github.com/huggingface/pytorch-image-models/releases/download/v0.1-tpu-weights/regnety_040_ra3-670e1166.pth'),
    'regnety_064.ra3_in1k': _cfg(
        hf_hub_id='timm/',
        url='https://github.com/huggingface/pytorch-image-models/releases/download/v0.1-tpu-weights/regnety_064_ra3-aa26dc7d.pth'),
    'regnety_080.ra3_in1k': _cfg(
        hf_hub_id='timm/',
        url='https://github.com/huggingface/pytorch-image-models/releases/download/v0.1-tpu-weights/regnety_080_ra3-1fdc4344.pth'),
    'regnety_120.sw_in12k_ft_in1k': _cfg(hf_hub_id='timm/'),
    'regnety_160.sw_in12k_ft_in1k': _cfg(hf_hub_id='timm/'),
    'regnety_160.lion_in12k_ft_in1k': _cfg(hf_hub_id='timm/'),

    # timm in12k pretrain
    'regnety_120.sw_in12k': _cfg(
        hf_hub_id='timm/',
        num_classes=11821),
    'regnety_160.sw_in12k': _cfg(
        hf_hub_id='timm/',
        num_classes=11821),

    # timm custom arch (v and z guess) + trained models
    'regnety_040_sgn.untrained': _cfg(url=''),
    'regnetv_040.ra3_in1k': _cfg(
        hf_hub_id='timm/',
        url='https://github.com/huggingface/pytorch-image-models/releases/download/v0.1-tpu-weights/regnetv_040_ra3-c248f51f.pth',
        first_conv='stem'),
    'regnetv_064.ra3_in1k': _cfg(
        hf_hub_id='timm/',
        url='https://github.com/huggingface/pytorch-image-models/releases/download/v0.1-tpu-weights/regnetv_064_ra3-530616c2.pth',
        first_conv='stem'),

    'regnetz_005.untrained': _cfg(url=''),
    'regnetz_040.ra3_in1k': _cfg(
        hf_hub_id='timm/',
        url='https://github.com/huggingface/pytorch-image-models/releases/download/v0.1-tpu-weights/regnetz_040_ra3-9007edf5.pth',
        input_size=(3, 256, 256), pool_size=(8, 8), crop_pct=1.0, test_input_size=(3, 320, 320)),
    'regnetz_040_h.ra3_in1k': _cfg(
        hf_hub_id='timm/',
        url='https://github.com/huggingface/pytorch-image-models/releases/download/v0.1-tpu-weights/regnetz_040h_ra3-f594343b.pth',
        input_size=(3, 256, 256), pool_size=(8, 8), crop_pct=1.0, test_input_size=(3, 320, 320)),

    # used in DeiT for distillation (from Facebook DeiT GitHub repository)
    'regnety_160.deit_in1k': _cfg(
        hf_hub_id='timm/', url='https://dl.fbaipublicfiles.com/deit/regnety_160-a5fe301d.pth'),

    'regnetx_004_tv.tv2_in1k': _cfgtv2(
        hf_hub_id='timm/',
        url='https://download.pytorch.org/models/regnet_x_400mf-62229a5f.pth'),
    'regnetx_008.tv2_in1k': _cfgtv2(
        hf_hub_id='timm/',
        url='https://download.pytorch.org/models/regnet_x_800mf-94a99ebd.pth'),
    'regnetx_016.tv2_in1k': _cfgtv2(
        hf_hub_id='timm/',
        url='https://download.pytorch.org/models/regnet_x_1_6gf-a12f2b72.pth'),
    'regnetx_032.tv2_in1k': _cfgtv2(
        hf_hub_id='timm/',
        url='https://download.pytorch.org/models/regnet_x_3_2gf-7071aa85.pth'),
    'regnetx_080.tv2_in1k': _cfgtv2(
        hf_hub_id='timm/',
        url='https://download.pytorch.org/models/regnet_x_8gf-2b70d774.pth'),
    'regnetx_160.tv2_in1k': _cfgtv2(
        hf_hub_id='timm/',
        url='https://download.pytorch.org/models/regnet_x_16gf-ba3796d7.pth'),
    'regnetx_320.tv2_in1k': _cfgtv2(
        hf_hub_id='timm/',
        url='https://download.pytorch.org/models/regnet_x_32gf-6eb8fdc6.pth'),

    'regnety_004.tv2_in1k': _cfgtv2(
        hf_hub_id='timm/',
        url='https://download.pytorch.org/models/regnet_y_400mf-e6988f5f.pth'),
    'regnety_008_tv.tv2_in1k': _cfgtv2(
        hf_hub_id='timm/',
        url='https://download.pytorch.org/models/regnet_y_800mf-58fc7688.pth'),
    'regnety_016.tv2_in1k': _cfgtv2(
        hf_hub_id='timm/',
        url='https://download.pytorch.org/models/regnet_y_1_6gf-0d7bc02a.pth'),
    'regnety_032.tv2_in1k': _cfgtv2(
        hf_hub_id='timm/',
        url='https://download.pytorch.org/models/regnet_y_3_2gf-9180c971.pth'),
    'regnety_080_tv.tv2_in1k': _cfgtv2(
        hf_hub_id='timm/',
        url='https://download.pytorch.org/models/regnet_y_8gf-dc2b1b54.pth'),
    'regnety_160.tv2_in1k': _cfgtv2(
        hf_hub_id='timm/',
        url='https://download.pytorch.org/models/regnet_y_16gf-3e4a00f9.pth'),
    'regnety_320.tv2_in1k': _cfgtv2(
        hf_hub_id='timm/',
        url='https://download.pytorch.org/models/regnet_y_32gf-8db6d4b5.pth'),

    'regnety_160.swag_ft_in1k': _cfgtv2(
        hf_hub_id='timm/',
        url='https://download.pytorch.org/models/regnet_y_16gf_swag-43afe44d.pth', license='cc-by-nc-4.0',
        input_size=(3, 384, 384), pool_size=(12, 12), crop_pct=1.0),
    'regnety_320.swag_ft_in1k': _cfgtv2(
        hf_hub_id='timm/',
        url='https://download.pytorch.org/models/regnet_y_32gf_swag-04fdfa75.pth', license='cc-by-nc-4.0',
        input_size=(3, 384, 384), pool_size=(12, 12), crop_pct=1.0),
    'regnety_1280.swag_ft_in1k': _cfgtv2(
        hf_hub_id='timm/',
        url='https://download.pytorch.org/models/regnet_y_128gf_swag-c8ce3e52.pth', license='cc-by-nc-4.0',
        input_size=(3, 384, 384), pool_size=(12, 12), crop_pct=1.0),

    'regnety_160.swag_lc_in1k': _cfgtv2(
        hf_hub_id='timm/',
        url='https://download.pytorch.org/models/regnet_y_16gf_lc_swag-f3ec0043.pth', license='cc-by-nc-4.0'),
    'regnety_320.swag_lc_in1k': _cfgtv2(
        hf_hub_id='timm/',
        url='https://download.pytorch.org/models/regnet_y_32gf_lc_swag-e1583746.pth', license='cc-by-nc-4.0'),
    'regnety_1280.swag_lc_in1k': _cfgtv2(
        hf_hub_id='timm/',
        url='https://download.pytorch.org/models/regnet_y_128gf_lc_swag-cbe8ce12.pth', license='cc-by-nc-4.0'),

    'regnety_320.seer_ft_in1k': _cfgtv2(
        hf_hub_id='timm/',
        license='other', origin_url='https://github.com/facebookresearch/vissl',
        url='https://dl.fbaipublicfiles.com/vissl/model_zoo/seer_finetuned/seer_regnet32_finetuned_in1k_model_final_checkpoint_phase78.torch',
        input_size=(3, 384, 384), pool_size=(12, 12), crop_pct=1.0),
    'regnety_640.seer_ft_in1k': _cfgtv2(
        hf_hub_id='timm/',
        license='other', origin_url='https://github.com/facebookresearch/vissl',
        url='https://dl.fbaipublicfiles.com/vissl/model_zoo/seer_finetuned/seer_regnet64_finetuned_in1k_model_final_checkpoint_phase78.torch',
        input_size=(3, 384, 384), pool_size=(12, 12), crop_pct=1.0),
    'regnety_1280.seer_ft_in1k': _cfgtv2(
        hf_hub_id='timm/',
        license='other', origin_url='https://github.com/facebookresearch/vissl',
        url='https://dl.fbaipublicfiles.com/vissl/model_zoo/seer_finetuned/seer_regnet128_finetuned_in1k_model_final_checkpoint_phase78.torch',
        input_size=(3, 384, 384), pool_size=(12, 12), crop_pct=1.0),
    'regnety_2560.seer_ft_in1k': _cfgtv2(
        hf_hub_id='timm/',
        license='other', origin_url='https://github.com/facebookresearch/vissl',
        url='https://dl.fbaipublicfiles.com/vissl/model_zoo/seer_finetuned/seer_regnet256_finetuned_in1k_model_final_checkpoint_phase38.torch',
        input_size=(3, 384, 384), pool_size=(12, 12), crop_pct=1.0),

    'regnety_320.seer': _cfgtv2(
        hf_hub_id='timm/',
        url='https://dl.fbaipublicfiles.com/vissl/model_zoo/seer_regnet32d/seer_regnet32gf_model_iteration244000.torch',
        num_classes=0, license='other', origin_url='https://github.com/facebookresearch/vissl'),
    'regnety_640.seer': _cfgtv2(
        hf_hub_id='timm/',
        url='https://dl.fbaipublicfiles.com/vissl/model_zoo/seer_regnet64/seer_regnet64gf_model_final_checkpoint_phase0.torch',
        num_classes=0, license='other', origin_url='https://github.com/facebookresearch/vissl'),
    'regnety_1280.seer': _cfgtv2(
        hf_hub_id='timm/',
        url='https://dl.fbaipublicfiles.com/vissl/model_zoo/swav_ig1b_regnet128Gf_cnstant_bs32_node16_sinkhorn10_proto16k_syncBN64_warmup8k/model_final_checkpoint_phase0.torch',
        num_classes=0, license='other', origin_url='https://github.com/facebookresearch/vissl'),
    # FIXME invalid weight <-> model match, mistake on their end
    #'regnety_2560.seer': _cfgtv2(
    #    url='https://dl.fbaipublicfiles.com/vissl/model_zoo/swav_ig1b_cosine_rg256gf_noBNhead_wd1e5_fairstore_bs16_node64_sinkhorn10_proto16k_apex_syncBN64_warmup8k/model_final_checkpoint_phase0.torch',
    #    num_classes=0, license='other', origin_url='https://github.com/facebookresearch/vissl'),

    'regnetx_002.pycls_in1k': _cfgpyc(hf_hub_id='timm/'),
    'regnetx_004.pycls_in1k': _cfgpyc(hf_hub_id='timm/'),
    'regnetx_006.pycls_in1k': _cfgpyc(hf_hub_id='timm/'),
    'regnetx_008.pycls_in1k': _cfgpyc(hf_hub_id='timm/'),
    'regnetx_016.pycls_in1k': _cfgpyc(hf_hub_id='timm/'),
    'regnetx_032.pycls_in1k': _cfgpyc(hf_hub_id='timm/'),
    'regnetx_040.pycls_in1k': _cfgpyc(hf_hub_id='timm/'),
    'regnetx_064.pycls_in1k': _cfgpyc(hf_hub_id='timm/'),
    'regnetx_080.pycls_in1k': _cfgpyc(hf_hub_id='timm/'),
    'regnetx_120.pycls_in1k': _cfgpyc(hf_hub_id='timm/'),
    'regnetx_160.pycls_in1k': _cfgpyc(hf_hub_id='timm/'),
    'regnetx_320.pycls_in1k': _cfgpyc(hf_hub_id='timm/'),

    'regnety_002.pycls_in1k': _cfgpyc(hf_hub_id='timm/'),
    'regnety_004.pycls_in1k': _cfgpyc(hf_hub_id='timm/'),
    'regnety_006.pycls_in1k': _cfgpyc(hf_hub_id='timm/'),
    'regnety_008.pycls_in1k': _cfgpyc(hf_hub_id='timm/'),
    'regnety_016.pycls_in1k': _cfgpyc(hf_hub_id='timm/'),
    'regnety_032.pycls_in1k': _cfgpyc(hf_hub_id='timm/'),
    'regnety_040.pycls_in1k': _cfgpyc(hf_hub_id='timm/'),
    'regnety_064.pycls_in1k': _cfgpyc(hf_hub_id='timm/'),
    'regnety_080.pycls_in1k': _cfgpyc(hf_hub_id='timm/'),
    'regnety_120.pycls_in1k': _cfgpyc(hf_hub_id='timm/'),
    'regnety_160.pycls_in1k': _cfgpyc(hf_hub_id='timm/'),
    'regnety_320.pycls_in1k': _cfgpyc(hf_hub_id='timm/'),
})


@register_model
def regnetx_002(pretrained=False, **kwargs) -> RegNet:
    """RegNetX-200MF"""
    return _create_regnet('regnetx_002', pretrained, **kwargs)


@register_model
def regnetx_004(pretrained=False, **kwargs) -> RegNet:
    """RegNetX-400MF"""
    return _create_regnet('regnetx_004', pretrained, **kwargs)


@register_model
def regnetx_004_tv(pretrained=False, **kwargs) -> RegNet:
    """RegNetX-400MF w/ torchvision group rounding"""
    return _create_regnet('regnetx_004_tv', pretrained, **kwargs)


@register_model
def regnetx_006(pretrained=False, **kwargs) -> RegNet:
    """RegNetX-600MF"""
    return _create_regnet('regnetx_006', pretrained, **kwargs)


@register_model
def regnetx_008(pretrained=False, **kwargs) -> RegNet:
    """RegNetX-800MF"""
    return _create_regnet('regnetx_008', pretrained, **kwargs)


@register_model
def regnetx_016(pretrained=False, **kwargs) -> RegNet:
    """RegNetX-1.6GF"""
    return _create_regnet('regnetx_016', pretrained, **kwargs)


@register_model
def regnetx_032(pretrained=False, **kwargs) -> RegNet:
    """RegNetX-3.2GF"""
    return _create_regnet('regnetx_032', pretrained, **kwargs)


@register_model
def regnetx_040(pretrained=False, **kwargs) -> RegNet:
    """RegNetX-4.0GF"""
    return _create_regnet('regnetx_040', pretrained, **kwargs)


@register_model
def regnetx_064(pretrained=False, **kwargs) -> RegNet:
    """RegNetX-6.4GF"""
    return _create_regnet('regnetx_064', pretrained, **kwargs)


@register_model
def regnetx_080(pretrained=False, **kwargs) -> RegNet:
    """RegNetX-8.0GF"""
    return _create_regnet('regnetx_080', pretrained, **kwargs)


@register_model
def regnetx_120(pretrained=False, **kwargs) -> RegNet:
    """RegNetX-12GF"""
    return _create_regnet('regnetx_120', pretrained, **kwargs)


@register_model
def regnetx_160(pretrained=False, **kwargs) -> RegNet:
    """RegNetX-16GF"""
    return _create_regnet('regnetx_160', pretrained, **kwargs)


@register_model
def regnetx_320(pretrained=False, **kwargs) -> RegNet:
    """RegNetX-32GF"""
    return _create_regnet('regnetx_320', pretrained, **kwargs)


@register_model
def regnety_002(pretrained=False, **kwargs) -> RegNet:
    """RegNetY-200MF"""
    return _create_regnet('regnety_002', pretrained, **kwargs)


@register_model
def regnety_004(pretrained=False, **kwargs) -> RegNet:
    """RegNetY-400MF"""
    return _create_regnet('regnety_004', pretrained, **kwargs)


@register_model
def regnety_006(pretrained=False, **kwargs) -> RegNet:
    """RegNetY-600MF"""
    return _create_regnet('regnety_006', pretrained, **kwargs)


@register_model
def regnety_008(pretrained=False, **kwargs) -> RegNet:
    """RegNetY-800MF"""
    return _create_regnet('regnety_008', pretrained, **kwargs)


@register_model
def regnety_008_tv(pretrained=False, **kwargs) -> RegNet:
    """RegNetY-800MF w/ torchvision group rounding"""
    return _create_regnet('regnety_008_tv', pretrained, **kwargs)


@register_model
def regnety_016(pretrained=False, **kwargs) -> RegNet:
    """RegNetY-1.6GF"""
    return _create_regnet('regnety_016', pretrained, **kwargs)


@register_model
def regnety_032(pretrained=False, **kwargs) -> RegNet:
    """RegNetY-3.2GF"""
    return _create_regnet('regnety_032', pretrained, **kwargs)


@register_model
def regnety_040(pretrained=False, **kwargs) -> RegNet:
    """RegNetY-4.0GF"""
    return _create_regnet('regnety_040', pretrained, **kwargs)


@register_model
def regnety_064(pretrained=False, **kwargs) -> RegNet:
    """RegNetY-6.4GF"""
    return _create_regnet('regnety_064', pretrained, **kwargs)


@register_model
def regnety_080(pretrained=False, **kwargs) -> RegNet:
    """RegNetY-8.0GF"""
    return _create_regnet('regnety_080', pretrained, **kwargs)


@register_model
def regnety_080_tv(pretrained=False, **kwargs) -> RegNet:
    """RegNetY-8.0GF w/ torchvision group rounding"""
    return _create_regnet('regnety_080_tv', pretrained, **kwargs)


@register_model
def regnety_120(pretrained=False, **kwargs) -> RegNet:
    """RegNetY-12GF"""
    return _create_regnet('regnety_120', pretrained, **kwargs)


@register_model
def regnety_160(pretrained=False, **kwargs) -> RegNet:
    """RegNetY-16GF"""
    return _create_regnet('regnety_160', pretrained, **kwargs)


@register_model
def regnety_320(pretrained=False, **kwargs) -> RegNet:
    """RegNetY-32GF"""
    return _create_regnet('regnety_320', pretrained, **kwargs)


@register_model
def regnety_640(pretrained=False, **kwargs) -> RegNet:
    """RegNetY-64GF"""
    return _create_regnet('regnety_640', pretrained, **kwargs)


@register_model
def regnety_1280(pretrained=False, **kwargs) -> RegNet:
    """RegNetY-128GF"""
    return _create_regnet('regnety_1280', pretrained, **kwargs)


@register_model
def regnety_2560(pretrained=False, **kwargs) -> RegNet:
    """RegNetY-256GF"""
    return _create_regnet('regnety_2560', pretrained, **kwargs)


@register_model
def regnety_040_sgn(pretrained=False, **kwargs) -> RegNet:
    """RegNetY-4.0GF w/ GroupNorm """
    return _create_regnet('regnety_040_sgn', pretrained, **kwargs)


@register_model
def regnetv_040(pretrained=False, **kwargs) -> RegNet:
    """RegNetV-4.0GF (pre-activation)"""
    return _create_regnet('regnetv_040', pretrained, **kwargs)


@register_model
def regnetv_064(pretrained=False, **kwargs) -> RegNet:
    """RegNetV-6.4GF (pre-activation)"""
    return _create_regnet('regnetv_064', pretrained, **kwargs)


@register_model
def regnetz_005(pretrained=False, **kwargs) -> RegNet:
    """RegNetZ-500MF
    NOTE: config found in https://github.com/facebookresearch/ClassyVision/blob/main/classy_vision/models/regnet.py
    but it's not clear it is equivalent to paper model as not detailed in the paper.
    """
    return _create_regnet('regnetz_005', pretrained, zero_init_last=False, **kwargs)


@register_model
def regnetz_040(pretrained=False, **kwargs) -> RegNet:
    """RegNetZ-4.0GF
    NOTE: config found in https://github.com/facebookresearch/ClassyVision/blob/main/classy_vision/models/regnet.py
    but it's not clear it is equivalent to paper model as not detailed in the paper.
    """
    return _create_regnet('regnetz_040', pretrained, zero_init_last=False, **kwargs)


@register_model
def regnetz_040_h(pretrained=False, **kwargs) -> RegNet:
    """RegNetZ-4.0GF
    NOTE: config found in https://github.com/facebookresearch/ClassyVision/blob/main/classy_vision/models/regnet.py
    but it's not clear it is equivalent to paper model as not detailed in the paper.
    """
    return _create_regnet('regnetz_040_h', pretrained, zero_init_last=False, **kwargs)


register_model_deprecations(__name__, {
    'regnetz_040h': 'regnetz_040_h',
})