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 12, 04:45 UTC
[Bugfix] Fix incorrect channel order for idefics3 in edge case (#29881)
Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn> Signed-off-by: Isotr0py <2037008807@qq.com> Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
Isotr0py committed
Dec 3, 2025 at 00:03 UTC
0ec84221718d920c3f46da879cc354f94b8fb59e
1 file changed
+1
vllm/model_executor/models/idefics3.py
+1
@@ -338,6 +338,7 @@ class Idefics3MultiModalProcessor(BaseMultiModalProcessor[Idefics3ProcessingInfo
338
prompt_ids = self._apply_hf_processor_tokens_only(prompt_ids)
339
return BatchFeature(dict(input_ids=[prompt_ids]), tensor_type="pt")
340
341
+ mm_kwargs = {"input_data_format": "channels_last", **mm_kwargs}
342
processed_outputs = super()._call_hf_processor(
343
prompt,
344
mm_data,