File size: 3,276 Bytes
7cac834 cc3b0c1 c5de362 7cac834 bbb2b22 7cac834 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 |
---
license: mit
tags:
- Text-to-Video
pipeline_tag: text-to-video
---
<div align="center">
<!-- TITLE -->
# **Video Diffusion Alignment via Reward Gradient**
[](http://vader-vid.github.io)
</div>
For more information on how to use it, please check [GitHub](https://github.com/mihirp1998/VADER).
This is a checkpoint trained using a reward function combining HPSv2.1 and Aesthetic function with a LoRA rank of 8 on prompts in `assets/chatgpt_custom_instruments.txt` for our paper [Video Diffusion Alignment via Reward Gradient](https://vader-vid.github.io/) by
Mihir Prabhudesai*, Russell Mendonca*, Zheyang Qin*, Katerina Fragkiadaki, Deepak Pathak.
<!-- DESCRIPTION -->
## Abstract
We have made significant progress towards building foundational video diffusion models. As these models are trained using large-scale unsupervised data, it has become crucial to adapt these models to specific downstream tasks, such as video-text alignment or ethical video generation. Adapting these models via supervised fine-tuning requires collecting target datasets of videos, which is challenging and tedious. In this work, we instead utilize pre-trained reward models that are learned via preferences on top of powerful discriminative models. These models contain dense gradient information with respect to generated RGB pixels, which is critical to be able to learn efficiently in complex search spaces, such as videos. We show that our approach can enable alignment of video diffusion for aesthetic generations, similarity between text context and video, as well long horizon video generations that are 3X longer than the training sequence length. We show our approach can learn much more efficiently in terms of reward queries and compute than previous gradient-free approaches for video generation.
## Demo
| | | |
| ----------------------------------------------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------- |
| <img src="assets/videos/8.gif" width=""> | <img src="assets/videos/5.gif" width=""> | <img src="assets/videos/7.gif" width=""> |
| <img src="assets/videos/10.gif" width=""> | <img src="assets/videos/3.gif" width=""> | <img src="assets/videos/4.gif" width=""> |
| <img src="assets/videos/9.gif" width=""> | <img src="assets/videos/1.gif" width=""> | <img src="assets/videos/11.gif" width=""> |
## Citation
If you find this work useful in your research, please cite:
```bibtex
@misc{prabhudesai2024videodiffusionalignmentreward,
title={Video Diffusion Alignment via Reward Gradients},
author={Mihir Prabhudesai and Russell Mendonca and Zheyang Qin and Katerina Fragkiadaki and Deepak Pathak},
year={2024},
eprint={2407.08737},
archivePrefix={arXiv},
primaryClass={cs.CV},
url={https://arxiv.org/abs/2407.08737},
}
``` |