Reposted by Yohis
2/n SmolVLM is a vision variant of the recently released SmolLM2. Idefics3 is the vision encoder, but with no Llama-3.1-8b, and image compression. This results in an accurate model for less memory.
thanks @merve.bsky.social and @andimara.bsky.social and @hf.co
huggingface.co/blog/smolvlm
thanks @merve.bsky.social and @andimara.bsky.social and @hf.co
huggingface.co/blog/smolvlm
SmolVLM - small yet mighty Vision Language Model
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
huggingface.co
November 30, 2024 at 7:40 AM
2/n SmolVLM is a vision variant of the recently released SmolLM2. Idefics3 is the vision encoder, but with no Llama-3.1-8b, and image compression. This results in an accurate model for less memory.
thanks @merve.bsky.social and @andimara.bsky.social and @hf.co
huggingface.co/blog/smolvlm
thanks @merve.bsky.social and @andimara.bsky.social and @hf.co
huggingface.co/blog/smolvlm