Skip to content

🧠 两小时从 0 训出你的 64M 大模型。Train your own 64M LLM from scratch in 2 hours.

[fix] update model

jingyaogonga044578d7374
diff --git a/model/model_minimind.py b/model/model_minimind.py
index e6b6096..2d640f6 100755
--- a/model/model_minimind.py
+++ b/model/model_minimind.py
@@ -193,8 +193,16 @@ class Attention(nn.Module):
repeat_kv(xv, self.n_rep).transpose(1, 2)
)
- if self.flash and seq_len > 1 and (attention_mask is None or torch.all(attention_mask == 1)):
- output = F.scaled_dot_product_attention(xq, xk, xv, dropout_p=self.dropout if self.training else 0.0, is_causal=True)
+ if self.flash and seq_len > 1:
+ if attention_mask is None or torch.all(attention_mask == 1):
+ attn_mask, is_causal = None, True
+ else:
+ causal_mask = torch.triu(torch.full((seq_len, seq_len), float("-inf"), device=xq.device), diagonal=1)
+ extended_mask = (1.0 - attention_mask.unsqueeze(1).unsqueeze(2)) * float("-inf")
+ attn_mask, is_causal = causal_mask.unsqueeze(0) + extended_mask, False
+
+ dropout_p = self.dropout if self.training else 0.0
+ output = F.scaled_dot_product_attention(xq, xk, xv, attn_mask=attn_mask, dropout_p=dropout_p, is_causal=is_causal)
else:
scores = (xq @ xk.transpose(-2, -1)) / math.sqrt(self.head_dim)
scores = scores + torch.triu(