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

Fix issue of using only the part of video frame [Nemotron Nano] (#26186)

Browse Source 
Signed-off-by: Eugene Khvedchenia <ekhvedchenia@nvidia.com>
This commit is contained in:
Eugene Khvedchenya
2025-10-04 03:21:00 +03:00
committed by GitHub
parent 7ef40bb983
commit 5a05f26603
1 changed files with 1 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
vllm/model_executor/models/nano_nemotron_vl.py
View File

@@ -208,7 +208,7 @@ def video_to_pixel_values(
) )
# dynamic_preprocess returns tensors already; take the single tile # dynamic_preprocess returns tensors already; take the single tile
assert len(pil_frame) >= 1 assert len(pil_frame) >= 1
frames_tensors.append(pil_frame[0]) frames_tensors.append(pil_frame[-1])
return torch.stack(frames_tensors) return torch.stack(frames_tensors)