Docker container for running JUPYTER Notebook within an ANACONDA nvironment, exploiting NVIDIA-DOCKER for GPUs The ANACONDA environment is also provided with:
- tensorflow-gpu (latest version, 1.6.0 when buildt)
- keras
- tensorboard jupyter extension
- NVIDIA driver (cuda-drivers, not whole cuda!) for your GPU
- nvidia-docker installed
- DOCKER
docker build -t nvidia-anaconda-docker <PATH_TO_DOCKERFILE_ROOT_PATH>
nvidia-docker run -v <HOST_NOTEBOOK_DIR>:/tmp -p 8888:8888 nvidia-anaconda-docker
From Dockerhub image
nvidia-docker run -v <HOST_NOTEBOOK_DIR>:/tmp -p 8888:8888 luke035/nvidia-anaconda-docker
Buildt image requires approximately 5GB of disk space.