HuggingFaceTB

/

SmolVLM2-2.2B-Instruct

like 233

Follow

Hugging Face Smol Models Research 2.36k

Oh interesting. Is the 2.2B not capable of video?

this change is based on a discussion with @merve ,what we understood is 'if the transformers pipeline is based on image text to text we should keep that'
did we missunderstand?

Oh I see. She's right, this unblocks a pipeline snippet in the "Use this model" button. But then, should we add video-text-to-text as a tag? The following search yields results for the 500M and 256M (which still have the old pipeline tag), but not for this one:
https://huggingface.co/models?other=video-text-to-text

Opened this for testing: https://huggingface.co/HuggingFaceTB/SmolVLM2-2.2B-Instruct/discussions/13/files