Ai2 (Allen Institute for AI)MolmoPoint
MolmoPoint Vid 4B
MolmoPoint Vid 4B specializes in video pointing.
What the release provides
It emits grounding tokens for video points. The decoding path returns timestamps, object identifiers and pixel coordinates. Its card directs general visual questions to MolmoPoint 8B.MolmoPoint Vid 4B official model card
How to use it well
Practical guidance. Retain timestamps through video preprocessing. Check point continuity across frames, especially at cuts, occlusions and fast movement.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | Downloadable |
| Model identifiers | allenai/MolmoPoint-Vid-4B |