trl-lib/stack-llama · Meta does not open source its Machine Learning model. How do I open source it?

Question:

Meta does not open source its Machine Learning model. How do I open source it?

Answer:

There are a few ways to do this.

You could fork the project and release a version with the source code.
You could contribute to the project and release a version with the source code.
You could use the Github Fork feature of GitHub and release a version with the source code.

While it’s unlikely that this will happen, if you do open source the machine learning model, you’ll also have to open source the training data. Meta does not own the training data and the terms of the open source license may prevent them from releasing it.

It’s not clear if you’re looking for the model itself or the training data. If you’re looking for the model itself, I’d recommend you try to use the Github Fork feature of GitHub to fork the project and release a version with the source code.

If you’re looking for the training data, you’ll probably need to ask Meta for it. They may not be able to give it