A script to download assets from NVIDIA Developer website. You need to have a valid NVIDIA Developer account, have enrolled to the program required to download the asset and have agreed with the terms of the corresponding license agreement.
- Python 2.7, 3.3, 3.4, 3.5 or 3.6
- Google Chrome (with headless support)
- ChromeDriver
See Dockerfile for the detailed setup procedures.
It is recommended to run nvget using the Docker image hosted on DockerHub, as it's a bit tough work to install Google Chrome and ChromeDriver correctly.
# Pull images from DockerHub.
docker pull kmaehashi/nvget
# Set NVIDIA Developer credentials to the environment variable, and add alias.
# These steps are just for convenience; add these lines to your ``.bashrc``.
export NVGET_USER="your.email@example.net"
export NVGET_PASSWORD="p@ssw0rd"
alias nvget-docker='docker run --rm -e NVGET_USER -e NVGET_PASSWORD -u "${UID}" -v "${PWD}:/download" kmaehashi/nvget'
# Download the file.
nvget-docker "https://developer.nvidia.com/compute/machine-learning/nccl/secure/v2.1/prod/nccl_2.1.2-1+cuda8.0_x86_64"
Modify the local nvget source tree and docker build -t nvget-dev -f Dockerfile-dev . to build new image for testing.
MIT License