Add model card and paper information
#1
by nielsr HF Staff - opened
README.md
CHANGED
|
@@ -1,3 +1,46 @@
|
|
| 1 |
---
|
| 2 |
license: mit
|
|
|
|
| 3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
license: mit
|
| 3 |
+
pipeline_tag: text-to-video
|
| 4 |
---
|
| 5 |
+
|
| 6 |
+
# UniTemp: Unlocking Video Generation in Any Temporal Order via Bidirectional Distillation
|
| 7 |
+
|
| 8 |
+
UniTemp is a few-step video diffusion model distilled from [Wan2.1](https://github.com/Wan-Video/Wan2.1) that generates videos in any temporal order (forward, backward, or in-between) with a single set of weights.
|
| 9 |
+
|
| 10 |
+
- **Paper:** [UniTemp: Unlocking Video Generation in Any Temporal Order via Bidirectional Distillation](https://arxiv.org/abs/2606.18702)
|
| 11 |
+
- **Project Page:** [https://lzhangbj.github.io/projects/unitemp/](https://lzhangbj.github.io/projects/unitemp/)
|
| 12 |
+
- **Repository:** [https://github.com/lzhangbj/UniTemp](https://github.com/lzhangbj/UniTemp)
|
| 13 |
+
|
| 14 |
+
## Introduction
|
| 15 |
+
Existing autoregressive video diffusion models are typically restricted to forward temporal generation. UniTemp bridges this gap by supporting generation in arbitrary temporal directions. It introduces **blockwise anchor latents** to address inter-block discontinuities during backward generation and utilizes a **bidirectional distillation framework** for efficient training.
|
| 16 |
+
|
| 17 |
+
## Usage
|
| 18 |
+
|
| 19 |
+
Please refer to the [official repository](https://github.com/lzhangbj/UniTemp) for environment setup and model checkpoint preparation. The model supports three inference modes:
|
| 20 |
+
|
| 21 |
+
### Forward Generation (past -> future)
|
| 22 |
+
```bash
|
| 23 |
+
bash scripts/inference_forward.sh
|
| 24 |
+
```
|
| 25 |
+
|
| 26 |
+
### Backward Generation (future -> past)
|
| 27 |
+
```bash
|
| 28 |
+
bash scripts/inference_backward.sh
|
| 29 |
+
```
|
| 30 |
+
|
| 31 |
+
### In-between Generation
|
| 32 |
+
```bash
|
| 33 |
+
# head clip + tail clip, then interpolate the transition
|
| 34 |
+
bash scripts/inference_inbetween.sh
|
| 35 |
+
```
|
| 36 |
+
|
| 37 |
+
## Citation
|
| 38 |
+
|
| 39 |
+
```bibtex
|
| 40 |
+
@article{zhang2026unitemp,
|
| 41 |
+
title={UniTemp: Unlocking Video Generation in Any Temporal Order via Bidirectional Distillation},
|
| 42 |
+
author={Zhang, Lin and Mo, Sicheng and Cai, Zefan and Lin, Jinhong and Lin, Zihao and Gu, Jiuxiang and Singh, Krishna Kumar and Li, Yuheng and Li, Yin},
|
| 43 |
+
journal={arXiv preprint arXiv:2606.18702},
|
| 44 |
+
year={2026}
|
| 45 |
+
}
|
| 46 |
+
```
|