vit_models.py

from timm.models.vision_transformer import _create_vision_transformer, checkpoint_filter_fn, Block, _init_vit_weights
from timm.models.vision_transformer_hybrid import _resnetv2, HybridEmbed, default_cfgs
from timm.models.layers import PatchEmbed, trunc_normal_
from timm.models.helpers import build_model_with_cfg
from functools import partial
import torch.nn as nn
import torch


class VisionTransformerMultiView(nn.Module):
    """ Vision Transformer

    A PyTorch impl of : `An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale`
        - https://arxiv.org/abs/2010.11929

    Includes distillation token & head support for `DeiT: Data-efficient Image Transformers`
        - https://arxiv.org/abs/2012.12877
    """

    def __init__(self, img_size=224, patch_size=16, in_chans=3, num_classes=1000, embed_dim=784, depth=12,
                 num_heads=12, mlp_ratio=4., qkv_bias=True, representation_size=None, distilled=False,
                 drop_rate=0., attn_drop_rate=0., drop_path_rate=0., embed_layer=PatchEmbed, norm_layer=None,
                 act_layer=None, weight_init=''):
        """
        Args:
            img_size (int, tuple): input image size
            patch_size (int, tuple): patch size
            in_chans (int): number of input channels
            num_classes (int): number of classes for classification head
            embed_dim (int): embedding dimension
            depth (int): depth of transformer
            num_heads (int): number of attention heads
            mlp_ratio (int): ratio of mlp hidden dim to embedding dim
            qkv_bias (bool): enable bias for qkv if True
            representation_size (Optional[int]): enable and set representation layer (pre-logits) to this value if set
            distilled (bool): model includes a distillation token and head as in DeiT models
            drop_rate (float): dropout rate
            attn_drop_rate (float): attention dropout rate
            drop_path_rate (float): stochastic depth rate
            embed_layer (nn.Module): patch embedding layer
            norm_layer: (nn.Module): normalization layer
            weight_init: (str): weight init scheme
        """
        super().__init__()
        self.num_classes = num_classes
        self.num_features = self.embed_dim = embed_dim  # num_features for consistency with other models
        self.num_tokens = 2 if distilled else 1
        norm_layer = norm_layer or partial(nn.LayerNorm, eps=1e-6)
        act_layer = act_layer or nn.GELU

        self.patch_embed = embed_layer(
            img_size=img_size, patch_size=patch_size, in_chans=in_chans, embed_dim=embed_dim)
        num_patches = self.patch_embed.num_patches

        self.cls_token = nn.Parameter(torch.zeros(1, 1, embed_dim))
        self.dist_token = nn.Parameter(torch.zeros(1, 1, embed_dim)) if distilled else None
        self.pos_embed = nn.Parameter(torch.zeros(1, num_patches + self.num_tokens, embed_dim))
        self.pos_embed_second = nn.Parameter(torch.zeros(1, num_patches + self.num_tokens, embed_dim))
        self.pos_drop = nn.Dropout(p=drop_rate)

        dpr = [x.item() for x in torch.linspace(0, drop_path_rate, depth)]  # stochastic depth decay rule
        self.blocks = nn.Sequential(*[
            Block(
                dim=embed_dim, num_heads=num_heads, mlp_ratio=mlp_ratio, qkv_bias=qkv_bias, drop=drop_rate,
                attn_drop=attn_drop_rate, drop_path=dpr[i], norm_layer=norm_layer, act_layer=act_layer)
            for i in range(depth)])
        self.norm = norm_layer(embed_dim)

        # Representation layer
        if representation_size and not distilled:
            self.num_features = representation_size
            self.pre_logits = nn.Sequential(OrderedDict([
                ('fc', nn.Linear(embed_dim, representation_size)),
                ('act', nn.Tanh())
            ]))
        else:
            self.pre_logits = nn.Identity()

        # Classifier head(s)
        self.pose_embed = nn.Linear(16, embed_dim)
        self.head = nn.Linear(self.num_features, num_classes) if num_classes > 0 else nn.Identity()
        self.head_dist = None
        if distilled:
            self.head_dist = nn.Linear(self.embed_dim, self.num_classes) if num_classes > 0 else nn.Identity()

        self.init_weights(weight_init)

    def init_weights(self, mode=''):
        assert mode in ('jax', 'jax_nlhb', 'nlhb', '')
        head_bias = -math.log(self.num_classes) if 'nlhb' in mode else 0.
        trunc_normal_(self.pos_embed, std=.02)
        trunc_normal_(self.pos_embed_second, std=.02)
        if self.dist_token is not None:
            trunc_normal_(self.dist_token, std=.02)
        if mode.startswith('jax'):
            # leave cls token as zeros to match jax impl
            named_apply(partial(_init_vit_weights, head_bias=head_bias, jax_impl=True), self)
        else:
            trunc_normal_(self.cls_token, std=.02)
            self.apply(_init_vit_weights)

    def _init_weights(self, m):
        # this fn left here for compat with downstream users
        _init_vit_weights(m)

    @torch.jit.ignore()
    def load_pretrained(self, checkpoint_path, prefix=''):
        _load_weights(self, checkpoint_path, prefix)

    @torch.jit.ignore
    def no_weight_decay(self):
        return {'pos_embed', 'cls_token', 'dist_token'}

    def get_classifier(self):
        if self.dist_token is None:
            return self.head
        else:
            return self.head, self.head_dist

    def reset_classifier(self, num_classes, global_pool=''):
        self.num_classes = num_classes
        self.head = nn.Linear(self.embed_dim, num_classes) if num_classes > 0 else nn.Identity()
        if self.num_tokens == 2:
            self.head_dist = nn.Linear(self.embed_dim, self.num_classes) if num_classes > 0 else nn.Identity()

    def forward_features(self, x):
        x = self.patch_embed(x)
        cls_token = self.cls_token.expand(x.shape[0], -1, -1)  # stole cls_tokens impl from Phil Wang, thanks
        if self.dist_token is None:
            x = torch.cat((cls_token, x), dim=1)
        else:
            x = torch.cat((cls_token, self.dist_token.expand(x.shape[0], -1, -1), x), dim=1)
        x = self.pos_drop(x + self.pos_embed)
        import pdb
        pdb.set_trace()
        print(x.size())
        x = self.blocks(x)
        x = self.norm(x)
        if self.dist_token is None:
            return self.pre_logits(x[:, 0])
        else:
            return x[:, 0], x[:, 1]

    def forward(self, x):
        import pdb
        pdb.set_trace()
        x = self.forward_features(x)
        if self.head_dist is not None:
            x, x_dist = self.head(x[0]), self.head_dist(x[1])  # x must be a tuple
            if self.training and not torch.jit.is_scripting():
                # during inference, return the average of both classifier predictions
                return x, x_dist
            else:
                return (x + x_dist) / 2
        else:
            x = self.head(x)
        return x

def _create_vision_transformer(variant, pretrained=False, default_cfg=None, **kwargs):
    default_cfg = default_cfg or default_cfgs[variant]
    if kwargs.get('features_only', None):
        raise RuntimeError('features_only not implemented for Vision Transformer models.')

    # NOTE this extra code to support handling of repr size for in21k pretrained models
    default_num_classes = default_cfg['num_classes']
    num_classes = kwargs.get('num_classes', default_num_classes)
    repr_size = kwargs.pop('representation_size', None)
    if repr_size is not None and num_classes != default_num_classes:
        # Remove representation layer if fine-tuning. This may not always be the desired action,
        # but I feel better than doing nothing by default for fine-tuning. Perhaps a better interface?
        _logger.warning("Removing representation layer for fine-tuning.")
        repr_size = None

    model = build_model_with_cfg(
        VisionTransformerMultiView, variant, pretrained,
        default_cfg=default_cfg,
        representation_size=repr_size,
        pretrained_filter_fn=checkpoint_filter_fn,
        pretrained_custom_load='npz' in default_cfg['url'],
        **kwargs)
    return model


def vit_base_r50_s16_384(pretrained=False, **kwargs):
    """ R50+ViT-B/16 hybrid from original paper (https://arxiv.org/abs/2010.11929).
    ImageNet-1k weights fine-tuned from in21k @ 384x384, source https://github.com/google-research/vision_transformer.
    """
    backbone = _resnetv2((3, 4, 9), **kwargs)
    model_kwargs = dict(embed_dim=768, depth=12, num_heads=12, **kwargs)
    model = _create_vision_transformer_hybrid(
        'vit_base_r50_s16_384', backbone=backbone, pretrained=pretrained, **model_kwargs)
    return model


def vit_base_resnet50_384(pretrained=False, **kwargs):
    # DEPRECATED this is forwarding to model def above for backwards compatibility
    return vit_base_r50_s16_384(pretrained=pretrained, **kwargs)


def _create_vision_transformer_hybrid(variant, backbone, pretrained=False, **kwargs):
    embed_layer = partial(HybridEmbed, backbone=backbone)
    kwargs.setdefault('patch_size', 1)  # default patch size for hybrid models if not set
    return _create_vision_transformer(
        variant, pretrained=pretrained, embed_layer=embed_layer, default_cfg=default_cfgs[variant], **kwargs)