# Fréchet Forcing: Training Long-Horizon Streaming World Models with Short Video Clips

- Source: HuggingFace Daily Papers
- Published: 2026-10-19T00:00:00.000Z
- RoboHot: https://robohot.news/en/items/vmhu2r13weql2lp3rg3abiiff
- Original: https://huggingface.co/papers/2610.25666

## Summary

Fréchet Forcing proposes a fine-tuning method that keeps streaming video world models stable during long-horizon generation beyond the training length and lets them continue to follow action commands. The method applies a flow-matching loss where self-generated history meets real data, constrains context fidelity with a Fréchet distance centered on the real clip itself, and enforces visual realism with a Fréchet distance on real frames in a frozen perceptual space. All three constraints come from short video clips, with no teacher model, online judge, or long videos needed.