hao-ai-lab/FastVideo

A unified inference and post-training framework for accelerated video generation.

About the project

FastVideo is a framework for training and running accelerated video-generation models, including real-time inference and post-training workflows. It is for researchers and developers building or deploying diffusion-based systems that generate video.

Latest release v0.2.0 — Release 0.2.0 · 4 Jun 2026

8 files for Claude Code, Codex and OpenCode: ci-runner, dreamverse-deploy, reseed-performance-baseline, reseed-ssim-references and 4 more — 1,460 tokens loaded in every session.

Open on GitHub hao-ai-lab.github.io

Instructions

FastVideo AGENTS.md AGENTS.md A 1,455 tok
FastVideo CLAUDE.md CLAUDE.md A 5 tok

Skills

ci-runner .agents/skills/ci-runner/SKILL.md A 39 tok
dreamverse-deploy .agents/skills/dreamverse-deploy/SKILL.md A 38 tok
reseed-performance-baseline .agents/skills/reseed-performance-baseline/SKILL.md A 162 tok
reseed-ssim-references .agents/skills/reseed-ssim-references/SKILL.md A 130 tok
seed-ssim-references .agents/skills/seed-ssim-references/SKILL.md A 146 tok
env-var-conventions .agents/skills/env-var-conventions/SKILL.md A 64 tok

These files are hao-ai-lab/FastVideo's own configuration — they tell Claude Code, Codex and OpenCode how to work on this repository, so they are not mods to install elsewhere. Copy one as a starting point and replace the parts that are about this project.