just curious if anyone knows or has a link to studies regarding comments in codebases and agentic dev. I recently went through and chopped 50%+ of the comments the model had left on initial build and I'm curious if this improves parsing (obviously it will improve token usage) and reasoning -- hopefully not the inverse? curious if anyone has looked into this.
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。