Is there way to use lars to test chunking strategy and embedding models? How do I side load custom embedding models and use other chunking strategies?

[mustavikhan05]

I need to build a physics chatbot based on a textbook. My only focus is to test multiple chunking strategies and embedding models.

I was thinking of modifying the LARS code base to keep everything and change the embedding and chunking part