ml-toolbox

Useful docker images with cuda, Jupyter and ml/dl libraries (pytorch, tf, jax, trax, haiku)

Usage:

docker run -d \
  --name toolbox \
  --gpus=all --ipc=host \
  -v /some-large-storage:/workspace/data \
  -v /some-fast-storage:/workspace/notebooks \
  -p 8888:8888 -p 6006:6006 \
  rexhaif/ml-toolbox:cpu

• some-fast-storage stands for a directory in mount path of your ssd's, if you have one
• some-large-storage stands for a directory in a mount path of your hdd/raid/nfs

Note: you might want to change tag to use GPU, refer to Tags for more information

I encourage you to use this two directories approach, where you store your large datasets in /workspace/data and store your notebooks/code in /workspace/notebooks. However, you can easily mount both paths into the same device or don't mount them at all.

After executing following command you will be able to access your jupyter notebook at http://your-hostname-or-ip:8888/lab, default password is change-me-asap. You are encouraged to change it, i'll provide necessary scripts later.

Tags:

• cu100: Ubuntu 18.04, CUDA 10.0 + Pytorch 1.6.0 for CUDA 9.2, Jax for CUDA 10.0
• cu101: Ubuntu 18.04, CUDA 10.1 + Pytorch 1.6.0 for CUDA 10.1
• cu102: Ubuntu 18.04, CUDA 10.2 + Pytorch 1.6.0 for CUDA 10.2
• cu110, latest: Ubuntu 20.04, CUDA 11.0 + Pytorch 1.6.0 for CUDA 10.2, Jax for CUDA 11.0
• cpu: Ubuntu 20.04, CPU-only Pytorch 1.6.0, Tensorflow 2.3.0 and Jax