https://github.com/bitnom/colabctl

Google Colaboratory background/task executioner & controller.
https://github.com/bitnom/colabctl

colab colaboratory google-cloud google-colab google-colaboratory machine-learning python python3

Last synced: 5 days ago
JSON representation

Google Colaboratory background/task executioner & controller.

• Host: GitHub
• URL: https://github.com/bitnom/colabctl
• Owner: bitnom
• License: gpl-3.0
• Created: 2019-04-23T02:51:47.000Z (over 5 years ago)
• Default Branch: master
• Last Pushed: 2020-12-17T07:29:14.000Z (almost 4 years ago)
• Last Synced: 2024-02-14T21:26:23.782Z (9 months ago)
• Topics: colab, colaboratory, google-cloud, google-colab, google-colaboratory, machine-learning, python, python3
• Language: Python
• Homepage:
• Size: 44.9 KB
• Stars: 245
• Watchers: 15
• Forks: 77
• Open Issues: 8
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# colabctl v0.2.1

Task executioner & controller for Google Colaboratory notebooks. Google Colaboratory is a game-changing innovation.

It allows anyone to access powerful hardware for machine learning, for free. Unfortunately, it's not possible to run

colab notebooks programmatically or otherwise in the background unless you leave your browser running. Colabctl is a

sort of shim that solves this problem by using headless Selenium.

## BROKEN NOTICE

After realizing a few issues being submitted, I reran the code and saw that there are a few new issues due to Google

changing their code. I'm now working to correct them. It shouldn't take too long, depending on what exactly they've

done.

## TODOs

- [ ] Resolve issues [#1](https://github.com/TensorTom/colabctl/issues/1) [#2](https://github.com/TensorTom/colabctl/issues/2) [#3](https://github.com/TensorTom/colabctl/issues/3) [#4](https://github.com/TensorTom/colabctl/issues/4)

- [ ] Async execution of tasks.

- [ ] Proper module structure for user command install via pip.

- [ ] Better args parsing.

- [ ] Revise & refactor the Seleium code.

- [ ] Proper logging.

Pull requests welcomed.

## Usage

First, add your Colaboratory URLs to `notebooks.csv`.

In colab, go to `tools -> keyboard shortcuts` and set "Clear all outputs" to `CTRL + SHIFT + Q`

In colab, go to `tools -> keyboard shortcuts` and set "Reset all runtimes" to `CTRL + SHIFT + k`

At the end of your notebook, add: `print("forkin"+"me")`

Now you're ready:

`python colabctl.py  `

Example:

`python colabctl.py forkinme 1800`

## Important!!!

Notice that the example delay is 1,800 seconds (30 minutes). Please **DO NOT poll Google Colaboratory frequently**.

We're very lucky to have these notebooks and even luckier that this ctl is even possible. Please don't push them to

block us from doing this.

Thank you