A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 4, 22:30 UTC
[Bug] Fix dcp error message (#40053)
Signed-off-by: yewentao256 <zhyanwentao@126.com>
Wentao Ye committed
Apr 20, 2026 at 10:52 UTC
b42e878ec0182aafbaed5cc28708e99287ef6856
1 file changed
+1
-1
vllm/v1/worker/cp_utils.py
+1
-1
@@ -33,7 +33,7 @@ def check_attention_cp_compatibility(vllm_config: VllmConfig) -> None:
33
"implementations to return the softmax LSE during decode, "
34
f"but {layer_impl.__class__.__name__} does not. "
35
"Try a different backend by setting "
36
- "VLLM_ATTENTION_BACKEND or disable DCP."
36
+ "--attention-backend or disable DCP."
37
)
38
39
if pcp_size > 1: