Frequently Asked Questions

What is Jet Machine Learning?

JetML helps people quickly leverage the latest open source machine learning tools by making it super simple to launch and provision machine learning environments that scale. Whether you’re in need of running a modest scikit-learn project on a single CPU or a massive Tensorflow project on a multicore GPU, we have you covered.

Is JetML good for teams?

Yes. JetML was designed with both small and enterprise teams in mind. JetML makes it simple to share accounts, projects, and payment methods amongst team members.

How do compute credits work?

Compute credits allow users to start cloud based computers (servers) to write code and process data. You will be charged compute credits for every minute you have an active server running (including server startup and shutdown time).

For more information on compute credits, visit our pricing page.

Can I restart a server once it's terminated?

No. Once a server is terminated, it cannot be restarted. Instead, just start a new server. Your project files are persistent, so you can start where you left off with the new server.

I'm not used to this workflow. Why did you design JetML this way?

JetML uses a unique workflow for projects.  We designed it to be both flexible and affordable for our users.

Servers can be expensive, and we don’t want you to have to pay for idol time. We also want you to be able to test ideas and scale to larger servers when needed. So keeping your storage and compute separate allows you to use the compute power you need when you need it.

Can I install my own Python packages?

Yes!

Recommended Method (requirements.txt)

When a new server starts, it’ll look for a requirements.txt file at the root directory of your project and will attempt to install your packages. You can add your requirements.txt file either through your git repo or by adding one using Jupyter on an existing server tied to the project.

This method will affect all future servers that are tied to this project.

Alternative Method (terminal)

You can run pip installs by starting a terminal on your server using Jupyter. Note that this method will only affect the server you’re using and will not affect future servers tied to the project.

Can I install my own Ubuntu package using apt-get?

Yes!

Recommended Method (start.sh)

When a new server starts, it’ll look for a start.sh file at the root directory of your project and will attempt to run it as a shell script. You can add your start.sh file either through your git repo or by adding one using Jupyter on an existing server tied to the project.

This method will affect all future servers that are tied to this project.

Alternative Method (terminal)

You can run apt-get packages by starting a terminal on your server using Jupyter. Note that this method will only affect the server you’re using and will not affect future servers tied to the project.