Open jason-fries opened 3 years ago
The solution here is to create a compiled pypi package. I'll leave this issue open though to track that though.
This is now fixed as the install process has been cleaned up.
This is an orthogonal issue to simplifying the installation process. A user might not be able, for security or admin reasons, to install the required packages needed to build clmbr. A containerized distribution is useful in these settings and many others. Let’s leave this open as a todo.
Due to the specific system dependencies of CLMBR, can we build a singular container for training models? This will make it much easier to deploy internally as well as share with collaborators.