biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[bug] Fix "Current vLLM config is not set." warnings when FlashInfer attention is used (#30241)

Browse Source 
Signed-off-by: Po-Han Huang <pohanh@nvidia.com>
This commit is contained in:
Po-Han Huang (NVIDIA)
2025-12-11 03:18:51 +08:00
committed by GitHub
parent 9f042ba26b
commit eea41804a4
2 changed files with 6 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
vllm/utils/flashinfer.py
View File

@@ -269,6 +269,8 @@ def supports_trtllm_attention() -> bool:
def force_use_trtllm_attention() -> bool | None:
"""
This function should only be called during initialization stage when vllm config
is set.
Return `None` if --attention-config.use_trtllm_attention is not set,
return `True` if TRTLLM attention is forced to be used,
return `False` if TRTLLM attention is forced to be not used.
@@ -296,11 +298,12 @@ def use_trtllm_attention(
kv_cache_dtype: str,
q_dtype: torch.dtype,
is_prefill: bool,
# None means auto-detection, True means force on, False means force off
force_use_trtllm: bool | None = None,
has_sinks: bool = False,
has_spec: bool = False,
) -> bool:
"""Return `True` if TRTLLM attention is used."""
force_use_trtllm = force_use_trtllm_attention()
# CLI argument is set to 0 - respect it
if force_use_trtllm is not None and not force_use_trtllm: