ViT-CIFAR10/Classification-noisy/scripts/model.py

import torch
import torch.nn as nn
import torch.nn.functional as F

class PatchEmbed(nn.Module):
    """ 将图像分成patch并进行embedding """
    def __init__(self, img_size=32, patch_size=4, in_chans=3, embed_dim=96):
        super().__init__()
        self.img_size = img_size
        self.patch_size = patch_size
        self.n_patches = (img_size // patch_size) ** 2
        
        self.proj = nn.Conv2d(
            in_chans, embed_dim,
            kernel_size=patch_size, stride=patch_size
        )

    def forward(self, x):
        x = self.proj(x)               # (B, E, H/P, W/P)
        x = x.flatten(2)               # (B, E, N)
        x = x.transpose(1, 2)          # (B, N, E)
        return x

class Attention(nn.Module):
    """ 多头自注意力机制 """
    def __init__(self, dim, n_heads=8, qkv_bias=True, attn_p=0., proj_p=0.):
        super().__init__()
        self.n_heads = n_heads
        self.dim = dim
        self.head_dim = dim // n_heads
        self.scale = self.head_dim ** -0.5

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

    def forward(self, x):
        n_samples, n_tokens, dim = x.shape

        if dim != self.dim:
            raise ValueError

        qkv = self.qkv(x)                                           # (n_samples, n_patches + 1, 3 * dim)
        qkv = qkv.reshape(
            n_samples, n_tokens, 3, self.n_heads, self.head_dim
        )                                                           # (n_samples, n_patches + 1, 3, n_heads, head_dim)
        qkv = qkv.permute(2, 0, 3, 1, 4)                          # (3, n_samples, n_heads, n_patches + 1, head_dim)
        q, k, v = qkv[0], qkv[1], qkv[2]                          # each with shape (n_samples, n_heads, n_patches + 1, head_dim)

        k_t = k.transpose(-2, -1)                                  # (n_samples, n_heads, head_dim, n_patches + 1)
        dp = (q @ k_t) * self.scale                               # (n_samples, n_heads, n_patches + 1, n_patches + 1)
        attn = dp.softmax(dim=-1)                                 # (n_samples, n_heads, n_patches + 1, n_patches + 1)
        attn = self.attn_drop(attn)

        weighted_avg = attn @ v                                    # (n_samples, n_heads, n_patches + 1, head_dim)
        weighted_avg = weighted_avg.transpose(1, 2)                # (n_samples, n_patches + 1, n_heads, head_dim)
        weighted_avg = weighted_avg.flatten(2)                     # (n_samples, n_patches + 1, dim)

        x = self.proj(weighted_avg)                               # (n_samples, n_patches + 1, dim)
        x = self.proj_drop(x)                                     # (n_samples, n_patches + 1, dim)

        return x

class MLP(nn.Module):
    """ 多层感知机 """
    def __init__(self, in_features, hidden_features, out_features, p=0.):
        super().__init__()
        self.fc1 = nn.Linear(in_features, hidden_features)
        self.act = nn.GELU()
        self.fc2 = nn.Linear(hidden_features, out_features)
        self.drop = nn.Dropout(p)

    def forward(self, x):
        x = self.fc1(x)               # (n_samples, n_patches + 1, hidden_features)
        x = self.act(x)               # (n_samples, n_patches + 1, hidden_features)
        x = self.drop(x)              # (n_samples, n_patches + 1, hidden_features)
        x = self.fc2(x)               # (n_samples, n_patches + 1, out_features)
        x = self.drop(x)              # (n_samples, n_patches + 1, out_features)

        return x

class Block(nn.Module):
    """ Transformer编码器块 """
    def __init__(self, dim, n_heads, mlp_ratio=4.0, qkv_bias=True,
                 p=0., attn_p=0.):
        super().__init__()
        self.norm1 = nn.LayerNorm(dim, eps=1e-6)
        self.attn = Attention(
            dim,
            n_heads=n_heads,
            qkv_bias=qkv_bias,
            attn_p=attn_p,
            proj_p=p
        )
        self.norm2 = nn.LayerNorm(dim, eps=1e-6)
        hidden_features = int(dim * mlp_ratio)
        self.mlp = MLP(
            in_features=dim,
            hidden_features=hidden_features,
            out_features=dim,
        )

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

class ViT(nn.Module):
    """ Vision Transformer """
    def __init__(
            self,
            img_size=32,
            patch_size=4,
            in_chans=3,
            num_classes=10,
            embed_dim=96,
            depth=12,
            n_heads=8,
            mlp_ratio=4.,
            qkv_bias=True,
            p=0.,
            attn_p=0.,
    ):
        super().__init__()

        self.patch_embed = PatchEmbed(
            img_size=img_size,
            patch_size=patch_size,
            in_chans=in_chans,
            embed_dim=embed_dim,
        )
        self.cls_token = nn.Parameter(torch.zeros(1, 1, embed_dim))
        self.pos_embed = nn.Parameter(
            torch.zeros(1, 1 + self.patch_embed.n_patches, embed_dim)
        )
        self.pos_drop = nn.Dropout(p=p)

        self.blocks = nn.ModuleList([
            Block(
                dim=embed_dim,
                n_heads=n_heads,
                mlp_ratio=mlp_ratio,
                qkv_bias=qkv_bias,
                p=p,
                attn_p=attn_p,
            )
            for _ in range(depth)
        ])

        self.norm = nn.LayerNorm(embed_dim, eps=1e-6)
        self.head = nn.Linear(embed_dim, num_classes)

    def forward(self, x):
        n_samples = x.shape[0]
        x = self.patch_embed(x)

        cls_token = self.cls_token.expand(n_samples, -1, -1)
        x = torch.cat((cls_token, x), dim=1)
        x = x + self.pos_embed
        x = self.pos_drop(x)

        for block in self.blocks:
            x = block(x)

        x = self.norm(x)

        cls_token_final = x[:, 0]
        x = self.head(cls_token_final)

        return x