From 10b34e36b9be7f68e236ec9c78425b6304be4cd1 Mon Sep 17 00:00:00 2001 From: Chauncey Date: Tue, 25 Mar 2025 11:48:08 +0800 Subject: [PATCH] [Bugfix] Fixed the issue of not being able to input video and image simultaneously (#15387) Signed-off-by: chaunceyjiang --- vllm/entrypoints/chat_utils.py | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) diff --git a/vllm/entrypoints/chat_utils.py b/vllm/entrypoints/chat_utils.py index 988fa01446076..d3613384590de 100644 --- a/vllm/entrypoints/chat_utils.py +++ b/vllm/entrypoints/chat_utils.py @@ -556,11 +556,11 @@ class MultiModalItemTracker(BaseMultiModalItemTracker[object]): raise ValueError(\ "Only one message can have {'type': 'image_embeds'}") mm_inputs["image"] = image_embeds_lst[0] - elif "image" in items_by_modality: + if "image" in items_by_modality: mm_inputs["image"] = items_by_modality["image"] # A list of images - elif "audio" in items_by_modality: + if "audio" in items_by_modality: mm_inputs["audio"] = items_by_modality["audio"] # A list of audios - elif "video" in items_by_modality: + if "video" in items_by_modality: mm_inputs["video"] = items_by_modality["video"] # A list of videos return mm_inputs @@ -589,11 +589,11 @@ class AsyncMultiModalItemTracker(BaseMultiModalItemTracker[Awaitable[object]]): raise ValueError( "Only one message can have {'type': 'image_embeds'}") mm_inputs["image"] = image_embeds_lst[0] - elif "image" in items_by_modality: + if "image" in items_by_modality: mm_inputs["image"] = items_by_modality["image"] # A list of images - elif "audio" in items_by_modality: + if "audio" in items_by_modality: mm_inputs["audio"] = items_by_modality["audio"] # A list of audios - elif "video" in items_by_modality: + if "video" in items_by_modality: mm_inputs["video"] = items_by_modality["video"] # A list of videos return mm_inputs