As an experienced LLM user, I don't use generative LLMs often

Re:As an experienced LLM user, I don't use generative LLMs o... by PantaloonFlames on Tuesday May 06, 2025 @10:09AM (#43905315)

Aider, the tool, does exactly the opposite, in my experience.

It really works, for me. It iterates by itself and fixes the problem.

[ Parent | on HN ]

Re:As an experienced LLM user, I don't use generative LLMs o... by diggan on Tuesday May 06, 2025 @10:36AM (#43905639)

I'm in the middle of test-driving Aider and I'm seeing exactly the same problem, the longer a conversation goes on, the worse the quality of replies... Currently, I'm doing something like this to prevent it from loading previous context:

    rm -r .aider.chat.history.md .aider.tags.cache.v4 || true && aider --architect --model deepseek/deepseek-reasoner --editor-model deepseek/deepseek-chat --no-restore-chat-history

Which clears the history, then I basically re-run that whenever the model gets something wrong (so I can update/improve the prompt and try again).

[ Parent | on HN ]

Re:As an experienced LLM user, I don't use generative LLMs o... by mikeqq2024 on Thursday May 08, 2025 @11:55PM (#43933666)

why not use /clear and /reset command?

[ Parent | on HN ]