戒酒的李白

Calculates the scaling dot product attention

import torch
import torch.nn as nn
import torch.nn.functional as F
class MultiHeadAttentionLayer(nn.Module):
def __init__(self, embed_size, num_heads):
... ... @@ -19,16 +20,21 @@ class MultiHeadAttentionLayer(nn.Module):
N = query.shape[0] # batch_size
# Linear transformations for Q, K, V
Q = self.q_linear(query) # shape: (N, seq_len, embed_size)
K = self.k_linear(keys) # shape: (N, seq_len, embed_size)
V = self.v_linear(values) # shape: (N, seq_len, embed_size)
Q = self.q_linear(query)
K = self.k_linear(keys)
V = self.v_linear(values)
# Reshape Q, K, V into multiple heads
# Reshape into multiple heads
Q = Q.reshape(N, -1, self.num_heads, self.head_dim)
K = K.reshape(N, -1, self.num_heads, self.head_dim)
V = V.reshape(N, -1, self.num_heads, self.head_dim)
return Q, K, V
# Compute scaled dot-product attention scores
attention_scores = torch.einsum("nqhd,nkhd->nhqk", [Q, K])
attention_scores = attention_scores / (self.head_dim ** 0.5)
attention = torch.softmax(attention_scores, dim=-1) # Normalize
return attention
if __name__ == "__main__":
... ... @@ -36,10 +42,9 @@ if __name__ == "__main__":
num_heads = 8
mha_layer = MultiHeadAttentionLayer(embed_size, num_heads)
# Dummy data
values = torch.randn(2, 10, embed_size)
keys = torch.randn(2, 10, embed_size)
query = torch.randn(2, 10, embed_size)
Q, K, V = mha_layer(values, keys, query)
print(f"Q shape: {Q.shape}, K shape: {K.shape}, V shape: {V.shape}")
attention = mha_layer(values, keys, query)
print(f"Attention shape: {attention.shape}")
... ...