用pytorch写transformer代码

好的，这里是一份简单的PyTorch代码示例，实现了Transformer模型：

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

class TransformerModel(nn.Module): def init(self, num_tokens, d_model, num_heads, num_layers, dim_feedforward): super().init() self.d_model = d_model self.token_embedding = nn.Embedding(num_tokens, d_model) self.pos_embedding = nn.Embedding(1000, d_model) self.layer_stack = nn.ModuleList([ TransformerLayer(d_model, num_heads, dim_feedforward) for _ in range(num_layers) ]) self.fc = nn.Linear(d_model, num_tokens)

def forward(self, x):
    seq_len = x.shape[1]
    pos = torch.arange(seq_len, dtype=torch.long)
    pos = pos.unsqueeze(0).expand(x.shape[0], seq_len)
    x = self.token_embedding(x) + self.pos_embedding(pos)
    for layer in self.layer_stack:
        x = layer(x)
    x = self.fc(x)
    return x
1
2
3
4
5
6
7
8
9
10

class TransformerLayer(nn.Module): def init(self, d_model, num_heads, dim_feedforward): super().init() self.self_attn = MultiHeadAttention(d_model, num_heads) self.feed_forward = nn.Sequential( nn.Linear(d_model, dim_feedforward), nn.ReLU(), nn.Linear(dim_feedforward, d_model) ) self.norm1 = nn.LayerNorm(d_model) self.norm2 = nn.LayerNorm(d_model)

def forward(self, x):
    attn_output = self.self_attn(x, x, x)
    x = self.norm1(x + attn_output)
    ff_output = self.feed_forward(x)
    x = self.norm2(x + ff_output)
    return x
1
2
3
4
5
6
7

class MultiHeadAttention(nn.Module): def init(self, d_model, num_heads): super().init() self.num_heads = num_heads self.d_model = d_model self.depth = d_model // num_heads self.query_lin = nn.Linear(d_model, num_heads * self.depth) self.key_lin = nn.Linear(d_model, num_heads * self.depth) self.value_lin = nn.Linear(d_model, num_