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.
Borrowing it
Nothing to install: this file belongs to hao-ai-lab/FastVideo. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/hao-ai-lab/FastVideo/main/.agents/skills/add-model-04-port-vae/SKILL.mdgit clone --depth 1 https://github.com/hao-ai-lab/FastVideoWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/hao-ai-lab/fastvideo/add-model-04-port-vae)<a href="https://agentmods.dev/skills/hao-ai-lab/fastvideo/add-model-04-port-vae"><img src="https://agentmods.dev/badge/skills/hao-ai-lab/fastvideo/add-model-04-port-vae/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/hao-ai-lab/fastvideo/add-model-04-port-vae"><img src="https://agentmods.dev/badge/skills/hao-ai-lab/fastvideo/add-model-04-port-vae.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00033 | $0.00891 |
| Opus 5 | $0.00016 | $0.00445 |
| Sonnet 5 | $0.00007 | $0.00178 |
| Haiku 4.5 | $0.00003 | $0.00089 |
Grade A, and why
add-model-04-port-vae scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 11d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 101 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Add Model Port VAE
Goal
Prototype or parity-debug one VAE or autoencoder in FastVideo-native code. This skill covers video, image, and audio VAEs.
Inputs
Follow ../add-model/shared/component_skill_common.md and require the complete
packet from ../add-model/contracts/component_context.md.
VAE-specific packet fields:
component: VAE or autoencoder name.parity_test:tests/local_tests/vaes/test_<family>_<component>_parity.py.weights: converted VAE dir, HF subfolder, or local official path.target_files:fastvideo/models/vaes/<arch_or_family>.pyandfastvideo/configs/models/vaes/<arch_or_family>.py.
Modes
Use the common prototype and parity-debug modes from
../add-model/shared/component_skill_common.md.
VAE-specific prototype concerns include latent normalization, stochastic posterior behavior, tiling incompatibility, temporal/spatial/audio layout, and decode output containers.
Reuse Proof
Apply the shared reuse proof. VAE-specific comparison must include latent layout, temporal/spatial/audio compression, scaling factor, mean/std normalization, posterior behavior, encode/decode output objects, tiling flags, and cropping.
Existing FastVideo Patterns
- Shared tiling wrapper:
fastvideo/models/vaes/common.py::ParallelTiledVAE. - Config bases:
VAEConfigandVAEArchConfiginfastvideo/configs/models/vaes/base.py. - Use the matching VAE config bucket. Wrong bucket inheritance can typecheck but fail during pipeline wiring.
- Config export: add the config to
fastvideo/configs/models/vaes/__init__.py. - Registry discovery: set
EntryClass = <ClassName>in the model file. - Loader path: VAE loaders resolve
_class_namethroughModelRegistryand load converted component weights from the VAE subdir. - Reference examples:
oobleck.py,autoencoder_kl.py,wanvae.py,ltx2vae.py, andgamecraftvae.py. - Layer guidance:
fastvideo/layers/AGENTS.md.
Implementation Rules
- Name reusable VAE architectures by architecture (
oobleck.py,autoencoder_kl.py); name family-specific VAEs by family. - Match official encode/decode contracts exactly: input layout, latent layout, temporal/spatial/audio compression, scaling factor, mean/std normalization, posterior sampling behavior, decode output object, and frame/sample cropping.
- Compare deterministic outputs in parity: decode outputs, encode mean/mode, or round-trip tensors. Do not compare stochastic samples unless the RNG path is explicitly controlled.
- Use FastVideo tiling only when it preserves official numerics for the tested shape; disable it in config for audio or unsupported dimensions.
- Put architecture constants on
VAEArchConfig; putload_encoder,load_decoder, tiling, dtype, and pretrained path fields onVAEConfig. - Follow the production import boundary in
../add-model/shared/common_rules.md.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 11d ago First seen · 101 lines · 33 tokens per session scan A 357539f43d96
add-model-04-port-vae is a skill published in the GitHub repository hao-ai-lab/FastVideo (4,368 stars, last pushed 2d ago), licensed Apache-2.0. It adds 33 tokens to every session and 891 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
diffusers-ascend-pipeline
A guide for running image and video generation pipelines on Huawei Ascend NPUs with the Diffusers library. Diffusers is a software library for using generative models, and the guide covers model pipelines, memory settings, LoRA adapters, and multi-card inference.
sglang-diffusion-performance
Use when choosing the fastest SGLang Diffusion flags for a model, GPU, and VRAM budget.
sglang-diffusion-add-model
Use when adding a new diffusion model or Diffusers pipeline to SGLang.
sglang-diffusion-modelopt-quant
Use when quantizing a diffusion DiT with NVIDIA ModelOpt and making the resulting FP8 or NVFP4 checkpoint loadable, verifiable, and benchmarkable in SGLang Diffusion.
grpo-rl-training
Expert guidance for GRPO/RL fine-tuning with TRL for reasoning and task-specific model training.
gptq
Post-training 4-bit quantization for LLMs with minimal accuracy loss. Use for deploying large models (70B, 405B) on consumer GPUs, when you need 4× memory reduction with <2% perplexity degradation, or for faster inference (3-4× speedup) vs FP16. Integrates with transformers and PEFT for QLoRA fine-tuning.