Environment Configuration
The operating environment must be the same as the system environment where the .run package is packed and the Python version. Currently, the tested environment is Linux x86_64 and Python 3.7.5.
The dependency libraries are as follows:
- NumPy (version: > 1.13.1)
- faiss-gpu (version: 1.7.1)
- torch (cuda version: 10.2)
- argparse
- NVIDIA Apex
Run the pip install command to install NumPy, faiss-gpu, torch, and argparse. The following is an example:
pip install numpy faiss-gpu torch argparse
For details about how to install NVIDIA Apex, visit https://github.com/NVIDIA/apex.
- Download the NVIDIA Apex source code.
git clone https://github.com/NVIDIA/apex
- Go to the Apex directory.
cd apex
- Install Apex.
pip install -v --disable-pip-version-check --no-cache-dir --global-option="--cpp_ext" --global-option="--cuda_ext" ./
Parent topic: IVFast Training APIs