I’m working on developing an AI model tailored for medical inquiries, akin to ChatGPT but specialized for healthcare questions. Recently, I’ve been exploring open-source LLMs such as LLAMA, Falcon, OpenChatKit, GPT-3, and others, considering factors like model size, training datasets, architecture, multilingual support, and benchmarking results.
My question is: Has anyone here built an AI model for medical purposes? If so, which LLM did you find most effective and why? If not, what criteria would you suggest for selecting and fine-tuning an LLM for this application?
I heard of a few companies that are working on this. I believe OpenAI has the ability to fine tune models for people now.