Bin Lin

Linking vocabulary difficulty and code naturalness to improve software comprehension.

MoE-LLaVA combines images and text using an efficient model structure.

MagicTime transforms written descriptions into dynamic time-lapse videos with improved realism.

Introducing a framework that improves image aesthetics evaluation through visual and language integration.

A new dataset and model enhance video captioning quality for machines.

A new method enhances video generation quality and efficiency.

Easily generate high-quality videos with just a few words using Open-Sora Plan.

Learn how NPP improves AI image generation efficiency and quality.