models_vit.py

import torch
import torch.nn as nn
import torch.nn.functional as F
import torch.hub
from itertools import repeat
import collections.abc


def drop_path(
    x, drop_prob: float = 0.0, training: bool = False, scale_by_keep: bool = True
):
    """Drop paths (Stochastic Depth) per sample (when applied in main path of residual blocks).
    This is the same as the DropConnect impl I created for EfficientNet, etc networks, however,
    the original name is misleading as 'Drop Connect' is a different form of dropout in a separate paper...
    See discussion: https://github.com/tensorflow/tpu/issues/494#issuecomment-532968956 ... I've opted for
    changing the layer and argument names to 'drop path' rather than mix DropConnect as a layer name and use
    'survival rate' as the argument.
    """
    if drop_prob == 0.0 or not training:
        return x
    keep_prob = 1 - 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 scale_by_keep:
        random_tensor.div_(keep_prob)
    return x * random_tensor


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):
        return drop_path(x, self.drop_prob, self.training, self.scale_by_keep)


def _ntuple(n):
    def parse(x):
        if isinstance(x, collections.abc.Iterable):
            return x
        return tuple(repeat(x, n))

    return parse


to_2tuple = _ntuple(2)


class Mlp(nn.Module):
    """MLP as used in Vision Transformer, MLP-Mixer and related networks"""

    def __init__(
        self,
        in_features,
        hidden_features=None,
        out_features=None,
        act_layer=nn.GELU,
        drop=0.0,
    ):
        super().__init__()
        out_features = out_features or in_features
        hidden_features = hidden_features or in_features
        drop_probs = to_2tuple(drop)

        self.fc1 = nn.Linear(in_features, hidden_features)
        self.act = act_layer()
        self.drop1 = nn.Dropout(drop_probs[0])
        self.fc2 = nn.Linear(hidden_features, out_features)
        self.drop2 = nn.Dropout(drop_probs[1])

    def forward(self, x):
        x = self.fc1(x)
        x = self.act(x)
        x = self.drop1(x)
        x = self.fc2(x)
        x = self.drop2(x)
        return x


class Attention(nn.Module):
    def __init__(
        self,
        dim,
        num_heads=8,
        qkv_bias=False,
        qk_scale=None,
        attn_drop=0.0,
        proj_drop=0.0,
    ):
        super().__init__()
        self.num_heads = num_heads
        head_dim = dim // num_heads
        # NOTE scale factor was wrong in my original version, can set manually to be compat with prev weights
        self.scale = qk_scale or head_dim**-0.5

        self.qkv = nn.Linear(dim, dim * 3, bias=qkv_bias)
        self.attn_drop = nn.Dropout(attn_drop)
        self.proj = nn.Linear(dim, dim)
        self.proj_drop = nn.Dropout(proj_drop)

    def forward(self, x):
        B, N, C = x.shape
        qkv = (
            self.qkv(x)
            .reshape(B, N, 3, self.num_heads, C // self.num_heads)
            .permute(2, 0, 3, 1, 4)
        )
        q, k, v = (
            qkv[0],
            qkv[1],
            qkv[2],
        )  # make torchscript happy (cannot use tensor as tuple)

        attn = (q @ k.transpose(-2, -1)) * self.scale
        attn = attn.softmax(dim=-1)
        attn = self.attn_drop(attn)

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


class CrossAttention(nn.Module):
    def __init__(
        self,
        dim,
        num_heads=8,
        qkv_bias=False,
        qk_scale=None,
        attn_drop=0.0,
        proj_drop=0.0,
    ):
        super().__init__()
        self.num_heads = num_heads
        head_dim = dim // num_heads
        # NOTE scale factor was wrong in my original version, can set manually to be compat with prev weights
        self.scale = qk_scale or head_dim**-0.5

        self.wq = nn.Linear(dim, dim, bias=qkv_bias)
        self.wk = nn.Linear(dim, dim, bias=qkv_bias)
        self.wv = nn.Linear(dim, dim, bias=qkv_bias)
        self.attn_drop = nn.Dropout(attn_drop)
        self.proj = nn.Linear(dim, dim)
        self.proj_drop = nn.Dropout(proj_drop)

    def forward(self, x, y):
        B, Nx, C = x.shape
        Ny = y.shape[1]
        # BNxC -> BNxH(C/H) -> BHNx(C/H)
        q = (
            self.wq(x)
            .reshape(B, Nx, self.num_heads, C // self.num_heads)
            .permute(0, 2, 1, 3)
        )
        # BNyC -> BNyH(C/H) -> BHNy(C/H)
        k = (
            self.wk(y)
            .reshape(B, Ny, self.num_heads, C // self.num_heads)
            .permute(0, 2, 1, 3)
        )
        # BNyC -> BNyH(C/H) -> BHNy(C/H)
        v = (
            self.wv(y)
            .reshape(B, Ny, self.num_heads, C // self.num_heads)
            .permute(0, 2, 1, 3)
        )

        attn = (q @ k.transpose(-2, -1)) * self.scale  # BHNx(C/H) @ BH(C/H)Ny -> BHNxNy
        attn = attn.softmax(dim=-1)
        attn = self.attn_drop(attn)

        x = (
            (attn @ v).transpose(1, 2).reshape(B, Nx, C)
        )  # (BHNxNy @ BHNy(C/H)) -> BHNx(C/H) -> BNxH(C/H) -> BNxC
        x = self.proj(x)
        x = self.proj_drop(x)
        return x


class CrossAttentionBlock(nn.Module):
    def __init__(
        self,
        dim,
        num_heads,
        mlp_ratio=4.0,
        qkv_bias=False,
        qk_scale=None,
        drop=0.0,
        attn_drop=0.0,
        drop_path=0.0,
        act_layer=nn.GELU,
        norm_layer=nn.LayerNorm,
    ):
        super().__init__()

        self.norm0 = norm_layer(dim)
        self.selfattn = Attention(
            dim,
            num_heads=num_heads,
            qkv_bias=qkv_bias,
            qk_scale=qk_scale,
            attn_drop=attn_drop,
            proj_drop=drop,
        )
        self.drop_path0 = DropPath(drop_path) if drop_path > 0.0 else nn.Identity()

        self.norm1 = norm_layer(dim)
        self.attn = CrossAttention(
            dim,
            num_heads=num_heads,
            qkv_bias=qkv_bias,
            qk_scale=qk_scale,
            attn_drop=attn_drop,
            proj_drop=drop,
        )
        # NOTE: drop path for stochastic depth, we shall see if this is better than dropout here
        self.drop_path1 = DropPath(drop_path) if drop_path > 0.0 else nn.Identity()

        self.norm2 = norm_layer(dim)
        self.mlp = Mlp(
            in_features=dim,
            hidden_features=int(dim * mlp_ratio),
            act_layer=act_layer,
            drop=drop,
        )
        self.drop_path2 = DropPath(drop_path) if drop_path > 0.0 else nn.Identity()

    def forward(self, x, y):
        x = x + self.drop_path0(self.selfattn(self.norm0(x)))
        x = x + self.drop_path1(self.attn(self.norm1(x), y))
        x = x + self.drop_path2(self.mlp(self.norm2(x)))
        return x