I have successfully made a LoRA for a small Qwen model based on my email responses and it noticeably adopts my tone and cadence in responses. I think its realistic that I could take posts from the board and train a NEET-GPT that we could all talk to.