xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2025-12-10 06:45:01 +08:00
Code Issues Packages Projects Releases Wiki Activity

[BugFix] Fix logits repetition penalty cuda check (#22592)

Browse Source
This commit is contained in:
Eugene Cheah 2025-08-10 22:52:31 -07:00 committed by GitHub
parent afa5b7ca0b
commit f919d4cb8f
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
vllm/_custom_ops.py
View File

@ -311,7 +311,7 @@ def apply_repetition_penalties(logits: torch.Tensor, prompt_mask: torch.Tensor,
output_mask: A boolean tensor indicating which tokens appear in the output. output_mask: A boolean tensor indicating which tokens appear in the output.
repetition_penalties: The repetition penalties of shape (num_seqs, ). repetition_penalties: The repetition penalties of shape (num_seqs, ).
""" """
if current_platform.is_cuda() and logits.is_contiguous(): if logits.is_cuda and logits.is_contiguous():
apply_repetition_penalties_cuda(logits, prompt_mask, output_mask, apply_repetition_penalties_cuda(logits, prompt_mask, output_mask,
repetition_penalties) repetition_penalties)
else: else: