Eventually, the MIT algorithm could assist w/enterprise documents, using the latest info it has to update them accordingly. Co-LLM could also train the small LLM on documents that must remain within the server, and then we can use both models for many different cases.

→ View original post on X — @mit_csail