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 14, 22:35 UTC
[Bugfix] Fix multi-api server not working for text models (#21933)
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
Cyrus Leung committed
Jul 30, 2025 at 23:42 UTC
366f6b3a4d92ee0b2df8e5620a88ddf57afc3681
1 file changed
+1
-14
vllm/config.py
+1
-14
@@ -856,7 +856,7 @@ class ModelConfig:
856
self.tokenizer = s3_tokenizer.dir
857
858
def _init_multimodal_config(self) -> Optional["MultiModalConfig"]:
859
- if self.registry.is_multimodal_model(self.architectures, self):
859
+ if self._model_info.supports_multimodal:
860
return MultiModalConfig(
861
limit_per_prompt=self.limit_mm_per_prompt,
862
media_io_kwargs=self.media_io_kwargs,
@@ -865,19 +865,6 @@ class ModelConfig:
865
disable_mm_preprocessor_cache,
866
interleave_mm_strings=self.interleave_mm_strings)
867
868
- if self.limit_mm_per_prompt:
869
- raise ValueError("`limit_mm_per_prompt` is only supported for "
870
- "multimodal models.")
871
- if self.mm_processor_kwargs:
872
- raise ValueError("`mm_processor_kwargs` is only supported for "
873
- "multimodal models.")
874
- if self.disable_mm_preprocessor_cache:
875
- raise ValueError("`disable_mm_preprocessor_cache` is only "
876
- "supported for multimodal models.")
877
- if self.interleave_mm_strings:
878
- raise ValueError("`interleave_mm_strings` is only "
879
- "supported for multimodal models.")
880
-
868
return None
869
870
def _get_encoder_config(self):