vit-transformer模型结构及源码解读

vit简介
vit模型是transformer在图像分类领域的首次成功尝试；
但是其需要基于大量数据去预训练模型；除了训练难度，现有的 Visual Transformer 参数量和计算量多大，比如 ViT 需要 18B FLOPs 在 ImageNet 达到 78% 左右 Top1，但是 CNN 模型如 GhostNet 只需 600M FLOPs 可以达到 79% 以上 Top1。

vit网络结构

源码解读

import torch
from torch import nn

from einops import rearrange, repeat
from einops.layers.torch import Rearrange

# helpers

def pair(t):
    return t if isinstance(t, tuple) else (t, t)

# classes
#PreNorm是对层进行归一化
class PreNorm(nn.Module):
    def __init__(self, dim, fn):
        super().__init__()
        self.norm = nn.LayerNorm(dim)
        self.fn = fn
    def forward(self, x, **kwargs):
        return self.fn(self.norm(x), **kwargs)
#FeedForward就是两层线性变换
class FeedForward(nn.Module):
    def __init__(self, dim, hidden_dim, dropout = 0.):
        super().__init__()
        self.net = nn.Sequential(
            nn.Linear(dim, hidden_dim),
            nn.GELU(),
            nn.Dropout(dropout),
            nn.Linear(hidden_dim, dim),
            nn.Dropout(dropout)
        )
    def forward(self, x):
        return self.net(x)
#attention的输入和输出维度相同[1,num_patches+1,128]-->[num_patches+1,128],其目的是赋予不同patch不同的权重；
#给予不同的注意力
class Attention(nn.Module):
    def __init__(self, dim, heads = 8, dim_head = 64, dropout = 0.):
        super().__init__()
        inner_dim = dim_head *  heads
        project_out = not (heads == 1 and dim_head == dim)

        self.heads = heads
        self.scale = dim_head ** -0.5

        self.attend = nn.Softmax(dim = -1)
        self.to_qkv = nn.Linear(dim, inner_dim * 3, bias = False)

        self.to_out = nn.Sequential(
            nn.Linear(inner_dim, dim),
            nn.Dropout(dropout)
        ) if project_out else nn.Identity()

    def forward(self, x):
        qkv = self.to_qkv(x).chunk(3, dim = -1)
        #获得三个维度相同的向量q,k,v,然后q,k相乘获得权重，乘以scale,再经过softmax之后，乘到v上
        q, k, v = map(lambda t: rearrange(t, 'b n (h d) -> b h n d', h = self.heads), qkv)

        dots = torch.matmul(q, k.transpose(-1, -2)) * self.scale

        attn = self.attend(dots)

        out = torch.matmul(attn, v)
        out = rearrange(out, 'b h n d -> b n (h d)')
        return self.to_out(out)
#Transformer就是将降维后的patches叠加上不同的系数（注意力机制），再加上两层线性传输
class Transformer(nn.Module):
    def __init__(self, dim, depth, heads, dim_head, mlp_dim, dropout = 0.):
        super().__init__()
        self.layers = nn.ModuleList([])
        for _ in range(depth):
            self.layers.append(nn.ModuleList([
                PreNorm(dim, Attention(dim, heads = heads, dim_head = dim_head, dropout = dropout)),
                PreNorm(dim, FeedForward(dim, mlp_dim, dropout = dropout))
            ]))
    def forward(self, x):
        for attn, ff in self.layers:
            x = attn(x) + x
            x = ff(x) + x
        return x

class ViT(nn.Module):
    def __init__(self, *, image_size, patch_size, num_classes, dim, depth, heads, mlp_dim, pool = 'cls', channels = 3, dim_head = 64, dropout = 0., emb_dropout = 0.):
        super().__init__()
        image_height, image_width = pair(image_size)
        patch_height, patch_width = pair(patch_size)

        assert image_height % patch_height == 0 and image_width % patch_width == 0, 'Image dimensions must be divisible by the patch size.'

        num_patches = (image_height // patch_height) * (image_width // patch_width)
        patch_dim = channels * patch_height * patch_width
        assert pool in {'cls', 'mean'}, 'pool type must be either cls (cls token) or mean (mean pooling)'
        #这里是对块进行编码，将patch_height*patch_width的大小输出维度变成隐层dim
        self.to_patch_embedding = nn.Sequential(
            Rearrange('b c (h p1) (w p2) -> b (h w) (p1 p2 c)', p1 = patch_height, p2 = patch_width),
            nn.Linear(patch_dim, dim),
        )
        #这里对每一个块进行位置编码(包含cls_tokens),位置编码的长度为128，会以相加的方式叠加到原始输入上
        #关于位置编码
        #1.固定正弦编码 - 无学习参数
        #2.绝对位置编码 - 一维学习编码
        #3.轴向位置编码 - 二维学习编码
        # 大多数 NLP 模型（和 GPT）只使用 2。是的，新的视觉 SOTA 与 GPT 的架构相同，只有细微的差别。条条大路通罗马。
        self.pos_embedding = nn.Parameter(torch.randn(1, num_patches + 1, dim))
        #关于cls_token，原先在NLP中，这个主要用于生成整个句子的分类，但是看过一些解释，去掉这个，用平均值或者加权值代表整个句子的分类也是一样的
        self.cls_token = nn.Parameter(torch.randn(1, 1, dim))
        self.dropout = nn.Dropout(emb_dropout)

        self.transformer = Transformer(dim, depth, heads, dim_head, mlp_dim, dropout)

        self.pool = pool
        self.to_latent = nn.Identity()
        #输出最终的分类数
        self.mlp_head = nn.Sequential(
            nn.LayerNorm(dim),
            nn.Linear(dim, num_classes)
        )

    def forward(self, img):
        #对图像进行分块和降维编码
        x = self.to_patch_embedding(img)
        b, n, _ = x.shape
        #加上一个分类维度[1,1,128]叠加到输入上面
        cls_tokens = repeat(self.cls_token, '() n d -> b n d', b = b)
        x = torch.cat((cls_tokens, x), dim=1)
        x += self.pos_embedding[:, :(n + 1)]#加上位置编码
        x = self.dropout(x)
        #经过注意力机制和两层线性变换
        x = self.transformer(x)
        #对num_patches+1这个维度求均值，x的维度由[1,num_patches+1,128]-->[1,1,128]
        x = x.mean(dim = 1) if self.pool == 'mean' else x[:, 0]

        x = self.to_latent(x)
        #再经过一层线性变换输出维度到num_classes
        #128->num_classes
        return self.mlp_head(x)