Skip to content

Commit d522f1e

Browse files
remove warning msg
1 parent 3a71c54 commit d522f1e

File tree

1 file changed

+0
-8
lines changed

1 file changed

+0
-8
lines changed

vllm/config.py

Lines changed: 0 additions & 8 deletions
Original file line numberDiff line numberDiff line change
@@ -989,14 +989,6 @@ def _verify_cuda_graph(self) -> None:
989989
"to eager mode.", self.hf_config.model_type)
990990
self.enforce_eager = True
991991

992-
RECOMMENDED_MODEL_SUPPORTS_CUDA_GRAPH = ['phi3samba']
993-
if (self.hf_config.model_type in RECOMMENDED_MODEL_SUPPORTS_CUDA_GRAPH
994-
and not self.enforce_eager and self.max_seq_len_to_capture < self.max_model_len):
995-
logger.warning(
996-
"%s model performs best with the CUDA graph explicitly enabled. Set `--max-seq-len-to-capture <#>` "
997-
"when starting vLLM.", self.hf_config.model_type)
998-
999-
1000992
def _verify_bnb_config(self) -> None:
1001993
"""
1002994
The current version of bitsandbytes (0.46.1) with 8-bit models does not

0 commit comments

Comments
 (0)