from Hacker News

FrankenVAR: Enabling generative vision modality for text-only LLMs

by sinenomine on 5/24/24, 3:24 AM with 1 comments

  • by sinenomine on 5/24/24, 3:24 AM

    The idea looks technically solid to me. Would be great if someone committed a few grand to check if it works on Llama-8B.