I haven't tried it yet, but I do intend to. I think the code for llm inference is quite straightforward. The complexity lies in collecting the training corpus a

I haven't tried it yet, but I do intend to. I think the code for llm inference is quite straightforward. The complexity lies in collecting the training corpus and doing good rlhf. That's just my intuition.