PackForcing

KO | EN

gitlite — search

#github #privacy #video #typescript #c #proxy #llm #graphql #html #framework

(★ 233)

No description available.

File Explorer

ZIP Download

- 1
- c2cc6abde8b8de5bb3a82681cb840492.jpg
- c2cc6abde8b8de5bb3a82681cb840492.mp4
- overview.pdf
- overview.png
README.md

# Use via CDN

jsDelivr

jsDelivr serves any public GitHub repository as a CDN with zero setup. Pick a version and a file to get a ready-to-paste link and snippet.

Version

File

Link

Example

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

## PackForcing: Short Video Training Suffices for Long Video Sampling and Long Context Inference

[![arXiv](https://img.shields.io/badge/arXiv%20paper-2603.23497-b31b1b.svg)](https://arxiv.org/abs/2603.25730)&nbsp;

</div>

This repository contains the official implementation and benchmark code for the paper:

> [**PackForcing: Short Video Training Suffices for Long Video Sampling and Long Context Inference**](https://arxiv.org/abs/2603.23497)
>
> Xiaofeng Mao, Shaohao Rui, Kaining Ying, Bo Zheng, Chuanhao Li, Mingmin Chi, Kaipeng Zhang 
> 
> Alaya Studio, Shanda AI Research, Fudan University, Shanghai Innovation Institute

## 💡 Introduction

![pipeline](https://raw.githubusercontent.com/AlayaLab/PackForcing/main/assets/overview.png)

Autoregressive video diffusion models have demonstrated remarkable progress, yet they remain bottlenecked by intractable linear KV-cache growth, temporal repetition, and compounding errors during long-video generation.

To address these challenges, we present **PackForcing**, a unified framework that efficiently manages the generation history through a novel three-partition KV-cache strategy:

1. **Sink tokens**: Preserve early anchor frames at full resolution to maintain global semantics.
2. **Mid tokens**: Achieve a massive spatiotemporal compression (~32x token reduction) via a dual-branch network fusing progressive 3D convolutions with low-resolution VAE re-encoding. To bound memory footprint, we introduce a dynamic top-k context selection mechanism.
3. **Recent tokens**: Kept at full resolution to ensure local temporal coherence.

Empowered by this principled hierarchical context compression and a continuous Temporal RoPE Adjustment, PackForcing can generate coherent 2-minute, 832x480 videos at 16 FPS on a single H200 GPU. It achieves a bounded KV cache of just ~4GB and enables a remarkable 24x temporal extrapolation (from 5s to 120s), operating effectively either zero-shot or trained on merely 5-second clips.

## 🎥 Generation Result

[![PackForcing 120s 视频生成结果](https://raw.githubusercontent.com/AlayaLab/PackForcing/main/assets/c2cc6abde8b8de5bb3a82681cb840492.jpg)](assets/c2cc6abde8b8de5bb3a82681cb840492.mp4)

## 📝 TODO List

- [ ] Release train/inference code
- [ ] Release pre-trained model weights

## 📖 Citation

If you find this project helpful, please consider citing our work:

```bibtex
@misc{mao2026packforcing,
      title={PackForcing: Short Video Training Suffices for Long Video Sampling and Long Context Inference}, 
      author={Xiaofeng Mao and Shaohao Rui and Kaining Ying and Bo Zheng and Chuanhao Li and Mingmin Chi and Kaipeng Zhang},
      year={2026},
      eprint={2603.25730},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2603.25730}, 
}
```

PackForcing

Owner Repos

File Explorer

# Use via CDN

Was this content helpful?

Recent Feedback

PackForcing

Owner Repos

File Explorer

# Use via CDN

Was this content helpful?

Recent Feedback

Download README