@jun7ingx Perhaps, but I'd like to feed it to my agent as context. If it knows what a correct implementation looks like, perhaps it can reproduce it cleaner. All of my work will be open-sourced, under this repo: https://t.co/qtMAz7cjWq
@sqs@amitrochates_x@sqs The funny thing is models are getting more and more token efficient. It's increasing becoming a priority for labs (esp OAI) to produce the same intelligence for fewer token