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