modality-adaptive-reasoning-visualizations

【设计与创意】Enable small vision-language models to reason over diverse data types by converting latent embeddings into visual representations, achieving specialized performance without domain-specific training.

独立 GitHub 快照 · 自动分类 · 以原作者文档为准