I guess everybody is trying to generate an opamp with GPT4..😆 While it is still immature, I'm seeing extreme potential since the design flow seems to be "entree" level. • While the vanilla gpt struggles generation of even diff. pairs, how far can it go with fine-tuning? (Hey, we got the data) • Can we integrate the simulation and gpt-api for fast prototyping so the LLMs will get sense of the PDK?