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

[Bugfix] Standardize merging multimodal embeddings (#26771)

Browse Source 
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
This commit is contained in:
Cyrus Leung
2025-10-14 17:36:21 +08:00
committed by GitHub
parent 577d498212
commit d2f816d6ff
19 changed files with 57 additions and 57 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
vllm/model_executor/models/hyperclovax_vision.py
View File

@@ -749,12 +749,12 @@ class HCXVisionForCausalLM(nn.Module, SupportsMultiModal, SupportsPP):
for modality in modalities:
if modality == "images":
image_input = modalities["images"]
vision_embeddings = self._process_image_input(image_input)
multimodal_embeddings += vision_embeddings
image_embeddings = self._process_image_input(image_input)
multimodal_embeddings += tuple(image_embeddings)
if modality == "videos":
video_input = modalities["videos"]
video_embeddings = self._process_video_input(video_input)
multimodal_embeddings += video_embeddings
multimodal_embeddings += tuple(video_embeddings)
return multimodal_embeddings