# Human Preference Optimization

A video caption generation model based on VILA-v1.5-8B, capable of generating detailed and human-preference-aligned captions for input videos.

A video-text generation model developed based on VILA-v1.5-13B, capable of generating fine-grained descriptive text for input videos that aligns with human preferences.

This is a powerful text-image interleaved input-output model, deeply aligned through the Align Anything algorithm, improving image generation capabilities and human preference alignment.

BPO is a training-free black-box alignment technique that improves model output quality by optimizing user input prompts.