Welcome to FastVideo

:::{figure} ../../assets/logo.jpg :align: center :alt: FastVideo :class: no-scaled-link :width: 60% :::

:::{raw} html

FastVideo is a lightweight framework for accelerating large video diffusion models.

:::

FastVideo is a lightweight framework for accelerating large video diffusion models developed by the Hao AI Lab.

FastVideo currently offers: (with more to come)

[NEW!] Sliding Tile Attention.
FastHunyuan and FastMochi: consistency distilled video diffusion models for 8x inference speedup.
First open distillation recipes for video DiT, based on PCM.
Support distilling/finetuning/inferencing state-of-the-art open video DiTs: 1. Mochi 2. Hunyuan.
Scalable training with FSDP, sequence parallelism, and selective activation checkpointing, with near linear scaling to 64 GPUs.
Memory efficient finetuning with LoRA, precomputed latent, and precomputed text embeddings.

Dev in progress and highly experimental.

Documentation

% How to start using vLLM?

:::{toctree} :caption: Getting Started :maxdepth: 1

getting_started/installation getting_started/examples/examples_index :::

% What is STA Kernel?

:::{toctree} :caption: Sliding Tile Attention :maxdepth: 1

sliding_tile_attention/installation sliding_tile_attention/usage sliding_tile_attention/test sliding_tile_attention/demo :::

:::{toctree} :caption: Inference :maxdepth: 1

inference/stepvideo inference/hunyuanvideo inference/fasthunyuan inference/fastmochi :::

:::{toctree} :caption: Developer Guide :maxdepth: 1

developer_guide/overview :::