Modified the docker image built for the 2022 course

git repo: GitHub - fzngagan/fastai-docker: Fast.AI course complete docker container for Paperspace and Gradient
docker hub: https://hub.docker.com/r/fzngagan/fastai-mod

Updates:

  • base image updated to paperspace/gradient-base:pt211-tf215-cudatk120-py311-20240202
  • fastai version 2.8.2
  • replaced the deprecated mambaforge and added Miniforge3
  • conda installation includes py-xgboost=*=cuda*
  • xgboost in the main python package available to the notebooks includes gpu support. This is probably a result of the image update. From the xgboost docs: Installation Guide — xgboost 3.0.2 documentation

You can use it on paperspace by:

  • selecting the Start from scratch option when creating the notebook.(I can’t use this updated image with the Fast.AI option. It results in an error. Can’t figure out the reason)
  • In the View Advanced Options, put the container name as fzngagan/fastai-mod:1.0.0
  • Add https://github.com/fastai/fastbook.git as the workspace url if you want fastbook notebooks.

I would be happy to make a PR to the original repo if @jeremy or any other people in charge feel right.

3 Likes

Hello,

This info will helps me a lot .

1 Like

Thank you for helping people get the information they need. Great stuff as usual. Keep up the great work!!!

1 Like