Is anyone training LLMs outside of Meta, OpenAI, etc... ?<p>I don't much get the point. For huge models, it's impossible to outcompete them. For smaller models, isn't mistral or LLaMa good enough?<p>What are other startups finetuning LLMs for?