HAL manages your machine learning research environment in AWS
Project description
HAL 🤖
🚧 HAL IS STILL UNDER CONSTRUCTION - PLEASE DON'T MAKE ANY SUDDEN MOVES 🚧
HAL manages your machine learning research environment in AWS
Using HAL, you can dynamically provision your perfect machine in AWS - small instances for tinkering with code all the way up to massive GPU instances for training deep learning models. Instance creation and termination is fast, so mode switching is relatively painless, and the costs are kept low by automatically calculating spot instance bids.
When they're created, instances attach themselves to your own persistent, floating EBS volume (defined in terraform), where you can store data, notebooks, git repos, etc.
Users can access instances via ssh, or through a tunnelled jupyterlab session.
Installation
pip install hal-cli
Using HAL
See the CLI reference for more detailed documentation.
Create a new instance
hal start p2.xlarge
replacing p2.xlarge
with the instance you want.
See the full list of instance types in your region here.
Describe your running instances
hal describe
Connect to your instance via ssh
Open a new shell on your instance by running
hal connect
Alternatively, you can open localhost:8888
in a browser and interact with your instance through jupyterlab.
Move files between your local machine and the remote instance
hal put \
--local-path /path/to/file/to/send \
--remote-path /path/on/instance
hal get \
--local-path /path/to/save/file/at \
--remote-path /path/on/instance
Shut down your instance
hal stop
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.