hao-ai-labOfficial·24 Agent Skills Included

FastVideo

Train, distill, and serve fast AI video generation models

Accelerates video generation model training, distillation, and real-time inference on GPUs. Ports new video models with parity tests, checkpoint converters, and pipeline wiring handled step by step. Removes slow manual debugging with layer-level tracing, SSIM quality checks, and CI automation. Ships ready workflows for finetuning, sparse distillation, and realtime video streaming apps.

4,262|431

↗

npx skills add hao-ai-lab/FastVideo --all -g -y

Available:

Directs the AI agent to read the nearest directory-level AGENTS.md, follow task-specific skills, and respect the two separate training stacks before editing any code.

All Skills in This Repository (24)

Pure Emerald Level Indicators

📦 In Repo

hao-ai-lab

add-model-06-port-generic

Prototype and parity-debug FastVideo scheduler, upsampler, or vocoder components.

seed-ssim-references

Generate, verify, and upload SSIM reference artefacts to Hugging Face datasets.

search-related-work

Search indexed markdown files for related papers, repositories, and blog posts.

index-related-work

Extract key insights from papers, repositories, or articles into a structured markdown index.

add-model-08-trace

Detect layer-level numerical divergence between FastVideo and reference models.

log-experiment

Update a Markdown journal with experiment parameters, outcomes, and lessons.

add-model-05-port-encoder

Prototype and validate custom encoders for the FastVideo framework.

add-model-09-pipeline

Automate FastVideo inference pipeline setup and verification for deployment.

add-model-02-parity

Generate parity test scaffolds comparing official models with FastVideo implementations.

add-model-04-port-vae

Prototype and parity-debug FastVideo VAE modules for video, image, and audio processing.

add-model-01-prep

Prepare FastVideo model assets by cloning repositories and configuring environments.

add-model-07-conversion

Convert official FastVideo weights into standardized load-ready components with validation.

Official

Advanced

Frequently Asked Questions

FAQPage Schema

How to install FastVideo?▼

Run `npx skills add hao-ai-lab/FastVideo --all -g -y` in your terminal to install all skills in this suite globally.

What does FastVideo do?▼

FastVideo is a framework for training, distilling, and serving video generation models with high-speed inference. It supports finetuning, sparse attention, and realtime video streaming on NVIDIA GPUs and Apple Silicon.

How do the agent skills help port a new video model?▼

The add-model skills walk your AI agent through downloading weights, converting checkpoints, writing parity tests, and wiring the pipeline step by step. Each phase has clear pass criteria so the port matches the official model numerically.

Does FastVideo work with Claude Code and other AI agents?▼

Yes. The repository ships AGENTS.md files and SKILL.md workflows that guide agents like Claude Code, Cursor, and OpenClaw through installs, model ports, and CI tasks.

Can I generate video in real time with FastVideo?▼

Yes. The included Dreamverse app streams and edits video in real time, and the deploy skill lets your agent launch it on a chosen local GPU with one command.