vit.py

# Copyright (c) Facebook, Inc. and its affiliates.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""
Mostly copy-paste from timm library.
https://github.com/rwightman/pytorch-image-models/blob/master/timm/models/vision_transformer.py
"""
import math
from functools import partial
from typing import Callable, Final, Optional, Sequence

import torch
from torch import Tensor, nn
from torch.nn import functional as F

from .common import ensure_tuple, get_act_layer, use_fused_attn


def vit_weights_init(module: nn.Module) -> None:
    if isinstance(module, nn.Linear):
        nn.init.trunc_normal_(module.weight, std=0.02)
        if module.bias is not None:
            nn.init.zeros_(module.bias)
    elif isinstance(module, nn.LayerNorm):
        nn.init.ones_(module.weight)
        nn.init.zeros_(module.bias)


class DropPath(nn.Module):
    """Drop paths (Stochastic Depth) per sample (when applied in main path of residual blocks)."""

    def __init__(self, drop_prob: float = 0.0, scale_by_keep: bool = True):
        super(DropPath, self).__init__()
        self.drop_prob = drop_prob
        self.scale_by_keep = scale_by_keep

    def forward(self, x: Tensor) -> Tensor:
        if self.drop_prob == 0 or not self.training:
            return x
        keep_prob = 1 - self.drop_prob
        shape = (x.shape[0],) + (1,) * (x.ndim - 1)  # work with diff dim tensors, not just 2D ConvNets
        random_tensor = x.new_empty(shape).bernoulli_(keep_prob)
        if keep_prob > 0.0 and self.scale_by_keep:
            random_tensor.div_(keep_prob)
        return x * random_tensor

    def extra_repr(self):
        return f"drop_prob={self.drop_prob:0.3f}"


class Mlp(nn.Module):
    def __init__(
        self,
        in_features: int,
        hidden_features: Optional[int] = None,
        out_features: Optional[int] = None,
        act_layer: Callable[[], nn.Module] = nn.GELU,
        drop: float = 0.0,
    ):
        super().__init__()
        out_features = out_features or in_features
        hidden_features = hidden_features or in_features
        self.fc1 = nn.Linear(in_features, hidden_features)
        self.act = act_layer()
        self.fc2 = nn.Linear(hidden_features, out_features)
        self.drop = nn.Dropout(drop) if drop > 0.0 else nn.Identity()

    def forward(self, x: Tensor) -> Tensor:
        x = self.fc1(x)
        x = self.act(x)
        x = self.drop(x)
        x = self.fc2(x)
        x = self.drop(x)
        return x


class Attention(nn.Module):
    fused_attn: Final[bool]

    def __init__(
        self,
        dim: int,
        num_heads: int = 8,
        qkv_bias: bool = False,
        qk_scale: Optional[float] = None,
        attn_drop: float = 0.0,
        proj_drop: float = 0.0,
    ):
        super().__init__()
        self.num_heads = num_heads
        self.head_dim = dim // num_heads
        self.scale = qk_scale or self.head_dim**-0.5
        self.fused_attn = use_fused_attn()

        self.qkv = nn.Linear(dim, dim * 3, bias=qkv_bias)
        self.attn_drop = nn.Dropout(attn_drop) if attn_drop > 0.0 else nn.Identity()
        self.proj = nn.Linear(dim, dim)
        self.proj_drop = nn.Dropout(proj_drop) if proj_drop > 0.0 else nn.Identity()

    def forward(self, x: Tensor) -> Tensor:
        B, N, C = x.shape
        qkv: Tensor = self.qkv(x).reshape(B, N, 3, self.num_heads, C // self.num_heads).permute(2, 0, 3, 1, 4)
        q, k, v = qkv.unbind(0)

        if self.fused_attn:
            dropout_p = getattr(self.attn_drop, "p", 0.0) if self.training else 0.0
            x = F.scaled_dot_product_attention(q, k, v, dropout_p=dropout_p)
        else:
            q = q * self.scale
            attn = q @ k.transpose(-2, -1)
            attn = attn.softmax(dim=-1)
            attn = self.attn_drop(attn)
            x = attn @ v

        x = x.transpose(1, 2).reshape(B, N, C)
        x = self.proj(x)
        x = self.proj_drop(x)
        return x


class Block(nn.Module):
    def __init__(
        self,
        dim: int,
        num_heads: int,
        mlp_ratio: float = 4.0,
        qkv_bias: bool = False,
        drop: float = 0.0,
        attn_drop: float = 0.0,
        drop_path: float = 0.0,
        act_layer: Callable[[], nn.Module] = nn.GELU,
        norm_layer: Callable[[], nn.Module] = nn.LayerNorm,
    ):
        super().__init__()
        self.norm1 = norm_layer(dim)
        self.attn = Attention(
            dim,
            num_heads=num_heads,
            qkv_bias=qkv_bias,
            attn_drop=attn_drop,
            proj_drop=drop,
        )
        self.drop_path = DropPath(drop_path) if drop_path > 0.0 else nn.Identity()
        self.norm2 = norm_layer(dim)
        mlp_hidden_dim = int(dim * mlp_ratio)
        self.mlp = Mlp(
            in_features=dim,
            hidden_features=mlp_hidden_dim,
            act_layer=act_layer,
            drop=drop,
        )

    def forward(self, x: Tensor) -> Tensor:
        x = x + self.drop_path(self.attn(self.norm1(x)))
        x = x + self.drop_path(self.mlp(self.norm2(x)))
        return x


class PatchEmbed(nn.Module):
    """Image to Patch Embedding"""

    def __init__(
        self,
        img_size: int | tuple[int, int] = 224,
        patch_size: int | tuple[int, int] = 16,
        in_chans: int = 3,
        embed_dim: int = 768,
        bias: bool = True,
        dynamic_pad: bool = False,
    ):
        super().__init__()
        self.img_size = ensure_tuple(img_size, 2)
        self.patch_size = ensure_tuple(patch_size, 2)
        self.num_patches = (self.img_size[0] // self.patch_size[0]) * (self.img_size[1] // self.patch_size[1])

        self.dynamic_pad = dynamic_pad

        self.proj = nn.Conv2d(in_chans, embed_dim, kernel_size=patch_size, stride=patch_size, bias=bias)

    def forward(self, x: Tensor) -> Tensor:
        _, _, H, W = x.shape
        if self.dynamic_pad:
            pad_h = (self.patch_size[0] - H % self.patch_size[0]) % self.patch_size[0]
            pad_w = (self.patch_size[1] - W % self.patch_size[1]) % self.patch_size[1]
            x = F.pad(x, (0, pad_w, 0, pad_h))
        x = self.proj(x)
        x = x.flatten(2).transpose(1, 2)  # NCHW -> NLC
        return x


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

    def __init__(
        self,
        img_size: int | tuple[int, int] = 224,
        patch_size: int | tuple[int, int] = 16,
        in_chans: int = 3,
        num_classes: int = 0,
        embed_dim: int = 768,
        depth: int = 12,
        num_heads: int = 12,
        mlp_ratio: float = 4.0,
        qkv_bias: bool = False,
        pre_norm: bool = False,
        drop_rate: float = 0.0,
        attn_drop_rate: float = 0.0,
        drop_path_rate: float = 0.0,
        norm_layer: Callable[[], nn.Module] = nn.LayerNorm,
        act_layer: Callable[[], nn.Module] = nn.GELU,
        skip_init: bool = False,
        dynamic_pad: bool = False,
        **kwargs,
    ):
        super().__init__()
        self.img_size = img_size
        self.patch_size = patch_size
        self.num_classes = num_classes
        self.num_features = self.embed_dim = embed_dim
        self.depth = depth

        self.patch_embed = PatchEmbed(
            img_size=img_size,
            patch_size=patch_size,
            in_chans=in_chans,
            embed_dim=embed_dim,
            bias=not pre_norm,  # disable bias if pre-norm is used (e.g. CLIP)
            dynamic_pad=dynamic_pad,
        )
        num_patches = self.patch_embed.num_patches
        embed_len = num_patches + 1  # num_patches + 1 for the [CLS] token

        self.cls_token = nn.Parameter(torch.zeros(1, 1, embed_dim))
        self.pos_embed = nn.Parameter(torch.zeros(1, embed_len, embed_dim))
        self.pos_drop = nn.Dropout(p=drop_rate) if drop_rate > 0.0 else nn.Identity()
        self.norm_pre = norm_layer(embed_dim) if pre_norm else nn.Identity()

        dpr = [x.item() for x in torch.linspace(0, drop_path_rate, self.depth)]  # stochastic depth decay rule
        self.blocks: list[Block] = nn.ModuleList(
            [
                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],
                    act_layer=act_layer,
                    norm_layer=norm_layer,
                )
                for i in range(self.depth)
            ]
        )
        self.norm = norm_layer(embed_dim)

        # Classifier head
        self.head = nn.Linear(embed_dim, num_classes) if num_classes > 0 else nn.Identity()

        if not skip_init:
            self.reset_parameters()

    def reset_parameters(self):
        nn.init.trunc_normal_(self.cls_token, std=0.02)
        nn.init.trunc_normal_(self.pos_embed, std=0.02)
        self.apply(vit_weights_init)

    def interpolate_pos_encoding(self, x: Tensor, w: Tensor, h: Tensor) -> Tensor:
        npatch = x.shape[1] - 1
        N = self.pos_embed.shape[1] - 1
        if npatch == N and w == h:
            return self.pos_embed
        class_pos_embed = self.pos_embed[:, 0]
        patch_pos_embed = self.pos_embed[:, 1:]
        dim = x.shape[-1]
        w0 = w // self.patch_embed.patch_size[0]
        h0 = h // self.patch_embed.patch_size[0]
        # we add a small number to avoid floating point error in the interpolation
        # see discussion at https://github.com/facebookresearch/dino/issues/8
        w0, h0 = w0 + 0.1, h0 + 0.1
        patch_pos_embed = nn.functional.interpolate(
            patch_pos_embed.reshape(1, int(math.sqrt(N)), int(math.sqrt(N)), dim).permute(0, 3, 1, 2),
            scale_factor=(w0 / math.sqrt(N), h0 / math.sqrt(N)),
            mode="bicubic",
        )
        if int(w0) != patch_pos_embed.shape[-2] or int(h0) != patch_pos_embed.shape[-1]:
            raise ValueError("Error in positional encoding interpolation.")
        patch_pos_embed = patch_pos_embed.permute(0, 2, 3, 1).view(1, -1, dim)
        return torch.cat((class_pos_embed.unsqueeze(0), patch_pos_embed), dim=1)

    def prepare_tokens(self, x: Tensor) -> Tensor:
        B, _, W, H = x.shape
        x = self.patch_embed(x)  # patch linear embedding

        # add the [CLS] token to the embed patch tokens
        cls_tokens = self.cls_token.expand(B, -1, -1)
        x = torch.cat((cls_tokens, x), dim=1)

        # add positional encoding to each token
        x = x + self.interpolate_pos_encoding(x, W, H)

        return self.pos_drop(x)

    def forward(self, x: Tensor, norm: bool = True) -> Tensor:
        x = self.forward_features(x, norm=norm)
        x = self.forward_head(x)
        return x

    def forward_features(self, x: Tensor, norm: bool = True) -> Tensor:
        x = self.prepare_tokens(x)
        x = self.norm_pre(x)
        for blk in self.blocks:
            x = blk(x)
        if norm:
            x = self.norm(x)
        return x[:, 0]

    def forward_head(self, x: Tensor) -> Tensor:
        x = self.head(x)
        return x

    def get_intermediate_layers(
        self,
        x: Tensor,
        n: int | Sequence[int] = 1,
        norm: bool = True,
    ) -> list[Tensor]:
        # we return the output tokens from the `n` last blocks
        outputs = []
        layer_indices = set(range(self.depth - n, self.depth) if isinstance(n, int) else n)
        x = self.prepare_tokens(x)
        x = self.norm_pre(x)

        for idx, blk in enumerate(self.blocks):
            x = blk(x)
            if idx in layer_indices:
                outputs.append(x)
        if norm:
            outputs = [self.norm(x) for x in outputs]
        return outputs


def vit_base_dreamsim(
    patch_size: int = 16,
    layer_norm_eps: float = 1e-6,
    num_classes: int = 512,
    act_layer: str | Callable[[], nn.Module] = "gelu",
    **kwargs,
):
    if isinstance(act_layer, str):
        act_layer = get_act_layer(act_layer)

    model = VisionTransformer(
        patch_size=patch_size,
        num_classes=num_classes,
        embed_dim=768,
        depth=12,
        num_heads=12,
        mlp_ratio=4,
        qkv_bias=True,
        norm_layer=partial(nn.LayerNorm, eps=layer_norm_eps),
        act_layer=act_layer,
        **kwargs,
    )
    return model