Part 2 on Free GPUs

I managed to rebuild part 2 and tested using the free GPUs on Onepanel. All seems to work good so far. Just fork the fast-ai project then create a ‘pytorch’ notebook…

https://c.onepanel.io/onepanel-demo/projects/fast-ai-courses-1to2/code

3 Likes

FYI the product homepage is here

I just tried this and wanted to share the steps I took for trying it out,

Web interface

  1. Register & Login
  2. Go to project page
  3. Click fork button
  4. Click Notebooks
  5. Click Create
  6. Select a GPU machine + the software and disk space configuration you need
  7. Wait 5-10 minutes for the instance to be created
  8. Click Launch
  9. Jupyter Notebooks opens and you can run code from there

My rough understanding of this product is they want to make working on machine learning projects easier by offering faster setup and ways to collaborate. Their current tag lines are Cognitive Computing Simplified and Simplify Reproducibility and Collaboration. There is no about page at the moment, though they do have a blog.

Some of the ways they try to do this are (1) sharing large datasets across instances and (2) offering to connect users working on the same projects. I think it is an interesting idea.

There is also a command line interface for creating projects using a onepanel command. I gather this is so you can easily sync code and datasets between your onepanel instances and your local machine. I did not complete those steps, but what I tried is below.

CLI full instructions

Command line interface

  1. Install git lfs

    • For Ubuntu 16.04 (see instructions for other machines)
      • sudo apt-get install software-properties-common
      • sudo add-apt-repository ppa:git-core/ppa
      • sudo apt-get update
      • curl -s https://packagecloud.io/install/repositories/github/git-lfs/script.deb.sh | sudo bash
      • sudo apt-get install git-lfs
      • git lfs install
  2. pip3 install -U onepanel

  3. Turn on git credential helper,
    git config --global credential.helper cache

  4. openpanel login

  5. onepanel clone onepanel-demo/projects/fast-ai-courses-1to2

As I understand it, this is still under development and setup might become easier.

3 Likes

Thank you for this! I am just getting started with using OnePanel.

One small addition/edit to Step 4:
Click on Workspaces–>Create workspace