OpenSora 安装教程

hpcaitech/Open-Sora: Open-Sora: Democratizing Efficient Video Production for All

1. 安装cmake
2. 清楚所有包管理器（apt等）安装的驱动：sudo apt purge ‘*nvidia*’
3. 安装最新计算卡驱动
   • sudo dpkg -i nvidia-driver-local-repo-ubuntu2404-570.86.15_1.0-1_amd64.deb
   • https://us.download.nvidia.com/tesla/570.86.15/NVIDIA-Linux-x86_64-570.86.15.run
4. 安装cuda-toolkit（12.8）（取消勾选其中的驱动和nvidia-fs）
   • sudo sh ./cuda_12.8.0_570.86.10_linux.run
5. 安装anaconda

# create a virtual env and activate (conda as an example)
conda create -n opensora python=3.11
conda activate opensora

# download the repo
git clone https://github.com/hpcaitech/Open-Sora
cd Open-Sora

# install torch, torchvision and xformers
pip install -r requirements/requirements-cu121.txt

# the default installation is for inference only
pip install -v . # for development mode, `pip install -v -e .`

# install the latest tensornvme to use async checkpoint saving
pip install git+https://github.com/hpcaitech/TensorNVMe.git

# install the latest colossalai to use the latest features
pip install git+https://github.com/hpcaitech/ColossalAI.git

# install flash attention
# set enable_flash_attn=False in config to disable flash attention
pip install packaging ninja
pip install flash-attn --no-build-isolation

# 更新pytorch至2.4.1 torchvision至0.19.1 colossalai至？ xformers至0.0.28
pip install torch --upgrade

# install apex
# set enable_layernorm_kernel=False in config to disable apex
pip install -v --disable-pip-version-check --no-cache-dir --no-build-isolation --config-settings "--build-option=--cpp_ext" --config-settings "--build-option=--cuda_ext" git+https://github.com/NVIDIA/apex.git

pip install gradio spaces
python gradio/app.py