PyTorch default docker configuration /w pandas, scikit-learn, scikit-image, jupyter, and matplotlib
10K+
Built on top of my unofficial PyTorch image (https://hub.docker.com/r/rwightman/pytorch/) based on official PyTorch Dockerfile. Python 3.5 with additional Conda packages installed that are useful for ML along with a jupyter notebook setup. See https://github.com/rwightman/pytorch-docker/tree/master/extra for specifics
GPU supported when using nvidia-docker
Default usage launches a Jupyter Notebook: nvidia-docker run -it --rm -p 8888:8888 rwightman/pytorch-extra
To run a bash shell: docker run -it --rm rwightman/pytorch-extra /bin/bash
Updates: 2017-07-28 - Added torchvision to v0.1.12 tag as it was missing from official PyTorch dockerfile on that branch 2017-07-31 - Fixed incorrect conda package install command missing pandas 2017-08-01 - Fix Python 3 IO output mode being ASCII and not UTF-8 2017-08-08 - Add PyTorch release v0.2.0
Content type
Image
Digest
Size
2.1 GB
Last updated
about 9 years ago
docker pull rwightman/pytorch-extra