---
license: mit
library_name: transformers
pipeline_tag: video-text-to-text
---

This repository contains the model described in [DreamFrame: Enhancing Video Understanding via Automatically Generated QA and Style-Consistent Keyframes](https://arxiv.org/abs/2403.01422).

Code: https://github.com/Deaddawn/DreamFrame-code.

Project page: https://deaddawn.github.io/DreamFrame