NVIDIA GPU¶
Instructions to install FastVideo for NVIDIA CUDA GPUs.
Requirements¶
- OS: Linux or Windows WSL
- Python: 3.10-3.12
- CUDA 12.8
- At least 1 NVIDIA GPU
Set up using Python¶
Create a new Python environment¶
uv¶
Recommended default: use uv for faster and more stable environment setup.
Please follow the documentation to install uv. After installing uv, create a new environment using:
# (Recommended) Create a new uv environment. Use `--seed` to install `pip` and `setuptools`.
uv venv --python 3.12 --seed
source .venv/bin/activate
Conda (alternative)¶
You can also create a Python environment using Conda.
1. Install Miniconda (if not already installed)¶
wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh
bash Miniconda3-latest-Linux-x86_64.sh
source ~/.bashrc
2. Create and activate a Conda environment for FastVideo¶
# Create and activate a Conda environment
conda create -n fastvideo python=3.12 -y
conda activate fastvideo
Installation¶
With uv (recommended)¶
Also optionally install FlashAttention:
With Conda environment (alternative)¶
Also optionally install FlashAttention:
Installation from Source¶
1. Clone the FastVideo repository¶
2. Install FastVideo¶
Basic installation:
Alternative with Conda environment:
Optional Dependencies¶
Flash Attention¶
Alternative with Conda environment:
Set up using Docker¶
We also have prebuilt docker images with FastVideo dependencies pre-installed: Docker Images
Development Environment Setup¶
If you're planning to contribute to FastVideo please see the following page: Contributor Guide
Hardware Requirements¶
For Basic Inference¶
- NVIDIA GPU with CUDA 12.8 support
For Lora Finetuning¶
- 40GB GPU memory each for 2 GPUs with lora
- 30GB GPU memory each for 2 GPUs with CPU offload and lora
For Full Finetuning/Distillation¶
- Multiple high-memory GPUs recommended (e.g., H100)
Troubleshooting¶
If you encounter any issues during installation, please open an issue on our GitHub repository.
You can also join our Slack community for additional support.