Skip to content

Commit

Permalink
ruff
Browse files Browse the repository at this point in the history
  • Loading branch information
pcmoritz committed May 21, 2024
1 parent 434d757 commit ce3a80a
Showing 1 changed file with 2 additions and 1 deletion.
3 changes: 2 additions & 1 deletion vllm/config.py
Original file line number Diff line number Diff line change
Expand Up @@ -253,7 +253,8 @@ def get_head_size(self) -> int:
# TODO remove hard code
if hasattr(self.hf_text_config, "model_type"
) and self.hf_text_config.model_type == 'deepseek_v2':
# FlashAttention suports only head_size 32, 64, 128, 256, we need to pad head_size 192 to 256
# FlashAttention suports only head_size 32, 64, 128, 256,
# we need to pad head_size 192 to 256
return 256
if hasattr(self.hf_text_config, "head_dim"):
return self.hf_text_config.head_dim
Expand Down

0 comments on commit ce3a80a

Please sign in to comment.