xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2025-12-21 14:55:45 +08:00
Code Issues Packages Projects Releases Wiki Activity

[ux] Switch a warning to debug about a pytorch fallback (#23750)

Browse Source 
Signed-off-by: Russell Bryant <rbryant@redhat.com>
This commit is contained in:
Russell Bryant 2025-09-25 10:38:16 -04:00 committed by GitHub
parent eb32335e35
commit 532a6cfccb
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 3 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
vllm/v1/sample/ops/topk_topp_sampler.py
View File

@ -109,9 +109,9 @@ class TopKTopPSampler(nn.Module):
# CPU-GPU synchronization while `flashinfer_sample` does. # CPU-GPU synchronization while `flashinfer_sample` does.
if (k is None and p is None) or generators: if (k is None and p is None) or generators:
if generators: if generators:
logger.warning_once("FlashInfer 0.2.3+ does not support " logger.debug_once("FlashInfer 0.2.3+ does not support "
"per-request generators. Falling back to " "per-request generators. Falling back to "
"PyTorch-native implementation.") "PyTorch-native implementation.")
return self.forward_native(logits, generators, k, p) return self.forward_native(logits, generators, k, p)
assert self.logprobs_mode not in ( assert self.logprobs_mode not in (
"processed_logits", "processed_logprobs" "processed_logits", "processed_logprobs"