A Framework of Small-scale Large Multimodal Models
A simple, unified multimodal models training engine. Lean, flexible, and built for hacking at scale.
A most Frontend Collection and survey of vision-language model papers, and models GitHub repository. Continuous updates.
A curated, searchable collection of research on AI and foundation models across the game lifecycle.