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

[V0 Deprecation] Remove max_seq_len_to_capture (#25543)

Browse Source 
Signed-off-by: Woosuk Kwon <woosuk.kwon@berkeley.edu>
This commit is contained in:
Woosuk Kwon
2025-09-24 01:51:39 -07:00
committed by GitHub
parent 77a7fce1bb
commit 2e19a848d4
7 changed files with 2 additions and 48 deletions
Download Patch File Download Diff File Expand all files Collapse all files

14
vllm/model_executor/models/config.py
View File

@@ -245,19 +245,6 @@ class SnowflakeGteNewModelConfig(VerifyAndUpdateConfig):
}
class GraniteMoeHybridModelConfig(VerifyAndUpdateConfig):
@staticmethod
def verify_and_update_config(vllm_config: "VllmConfig") -> None:
config = vllm_config.model_config
config.max_seq_len_to_capture = config.max_model_len
logger.info(
"Setting max_seq_len_to_capture to %d "
"to ensure that CUDA graph capture "
"covers sequences of length up to max_model_len.",
config.max_model_len)
class GptOssForCausalLMConfig(VerifyAndUpdateConfig):
@staticmethod
@@ -426,7 +413,6 @@ MODELS_CONFIG_MAP: dict[str, type[VerifyAndUpdateConfig]] = {
"XLMRobertaModel": JinaRobertaModelConfig,
"JinaVLForRanking": JinaVLForSequenceClassificationConfig,
"JambaForSequenceClassification": JambaForSequenceClassificationConfig,
"GraniteMoeHybridForCausalLM": GraniteMoeHybridModelConfig,
"GptOssForCausalLM": GptOssForCausalLMConfig,
"MambaForCausalLM": MambaModelConfig,
"Mamba2ForCausalLM": MambaModelConfig,