从0搭建Transformer

1. 位置编码模块：

import torch
import torch.nn as nn
import math

class PositonalEncoding(nn.Module):
    def __init__ (self, d_model, dropout, max_len=5000):
        super(PositionalEncoding, self).__init__()
        self.dropout = nn.Dropout(p=dropout)
        # [[1, 2, 3],
        # [4, 5, 6],
        # [7, 8, 9]]
        pe = torch.zeros(max_len, d_model)
        # [[0],
        # [1],
        # [2]]
        position = torch.arange(0, max_len, dtype = torch.float).unsqueeze(1)
        div_term = torch.exp(torch.arange(0, d_model, 2).float() * (-math.log(10000.0) / d_model))
        pe[:, 0::2] = torch.sin(position * div_term)
        pe[:, 1::2] = torch.cos(position * div_term)
        pe = pe.unsqueeze(0)
        # 位置编码固定，不更新参数
        # 保存模型时会保存缓冲区，在引入模型时缓冲区也被引入
        self.register_buffer('pe', pe)

    def forward(self, x):
        # 不计算梯度
        x = x + self.pe[:, :x.size(1)].requires_grad_(False)

2. 多头注意力模块

class MultiHeadAttention(nn.Module):
    def __init__(self, d_model, num_heads):
        super(MultiHeadAttention, self).__init__()
        assert d_model % num_heads == 0
        self.d_k = d_model // num_heads
        self.num_heads = num_heads

        self.W_q = nn.Linear(d_model, d_model)
        self.W_k = nn.Linear(d_model, d_model)
        self.W_v = nn.Linear(d_model, d_model)

        self.dropout = nn.Dropout(dropout)
        self.W_o = nn.Linear(d_model, d_model)

    def forward(self, query, key, value, mask=None):
        batch_size = query.size(0)
        Q = self.W_q(query).view(batch_size, -1, self.num_heads, self.d_k).transpose(1, 2)
        K = self.W_k(key).view(batch_size, -1, self.num_heads, self.d_k).transpose(1, 2)
        V = self.W_v(value).view(batch_size, -1, self.num_heads, self.d_k).transpose(1, 2)
        scores = torch.matmul(Q, K.transpose(-2, -1)) / math.sqrt(self.d_k)

        if mask is not None:
            scores = scores.masked_fill(mask == 0, -1e9)

        attn_weights = torch.softmax(scores, dim=-1)
        context = torch.matmul(attn_weights, V)
        context = context.transpose(1, 2).contiguous().view(batch_size, -1, self.d_k * self.num_heads)
        return self.W_o(context)

3. 编码器层

class EncoderLayer(nn.Module):
    def __init__(self, d_model, num_heads, d_ff, dropout = 0.1):
        super().__init__()
        self.atten = MultiHeadAttention(d_model, num_heads)
        self.feed_forward = nn.Sequential(
            nn.Linear(d_model, d_ff),
            nn.ReLU(),
            nn.Linear(d_ff, d_model)
        )
        self.norm1 = nn.LayerNorm(d_model)
        self.norm2 = nn.LayerNorm(d_model)
        self.dropout = nn.Dropout(dropout)
    def forward(self, x, mask=None):
        attn_output = self.attn(x, x, x, mask)
        x = self.norm1(x + self.dropout(attn_output))
        ff_output = self.feed_forward(x)
        x = self.norm2(x + self.dropout(ff_output))
        return x

4. 解码器层

class DecoderLayer(nn.Module):
    def __init__(self, d_model, num_heads, d_ff, dropout=0.1):
        super(DecoderLayer, self).__init__()
        self.self_attn = MultiHeadAttention(d_model, num_heads, dropout)
        self.cross_attn = MultiHeadAttention(d_model, num_heads, dropout)

        self.norm1 = nn.LayerNorm(d_model)
        self.norm2 = nn.LayerNorm(d_model)
        self.norm3 = nn.LayerNorm(d_model)

        self.feed_forward = nn.Sequential(
            nn.Linear(d_model, d_ff),
            nn.ReLU(),
            nn.Linear(d_ff, d_model)
        )

        self.dropout = nn.Dropout(dropout)

    def forward(self, x, enc_output, src_mask, tgt_mask):
        attn_output = self.self_attn(x, x, x, tgt_mask)
        x = self.norm1(x + self.dropout(attn_output))
        attn_output = self.cross_attn(x, enc_output, enc_output, src_mask)
        x = self.norm2(x + self.dropout(attn_output))

        ff_output = self.feed_forward(x)
        x = self.norm3(x + self.dropout(ff_output))

        return x

5. 模型整合

class Transformer(nn.module):
    def __init__(self, src_vocab_size, tgt_vocab_size, d_model=512, num_heads=8, 
    num_layers=6, d_ff=2048, dropout=0.1):
        super(Transformer, self).__init__()
        self.encoder_embed = nn.Embedding(src_vocab_size, d_model)
        self.decoder_embed = nn.Embedding(tgt_vocab_size, d_model)

        self.pos_encoder = PositionalEncoding(d_model, dropout)

        self.encoder_layers = nn.ModuleList([EncoderLayer(d_model, num_heads, d_ff, dropout) for _ in range(num_layers)])
        self.decoder_layers = nn.ModuleList([DecoderLayer(d_model, num_heads, d_ff, dropout) for _ in range(num_layers)])

        self.fc_out = nn.Linear(d_model, tgt_vocab_size)

    def encode(self, src, src_mask):
        src_embeded = self.encoder_embed(src)
        src = self.pos_encoder(src_embeded)
        for layer in self.encoder_layers:
            src = layer(src, src_mask)
        return src

    def decode(self, tgt, enc_output, src_mask, tgt_mask):
        tgt_embeded = self.decoder_embed(tgt)
        tgt = self.pos_encoder(tgt_embeded)

        for layer in self.decoder_layers:
            tgt = layer(tgt, enc_output, src_mask, tgt_mask)
        return tgt

    def forward(self, src, tgt, src_mask, tgt_mask):
        enc_output = self.encode(src, src_mask)
        dec_output = self.decode(tgt, enc_output, src_mask, tgt_mask)
        logits = self.fc_out(dec_output)

        return logits