News for hackers. Stuff that matters.

A comment on Qwen2.5-1M: Deploy your own Qwen with context length up to 1M tokens. Parent comment

Re:Qwen2.5-1M: Deploy your own Qwen with context length up t... by anotherpaulg on Sunday January 26, 2025 @06:22PM (#42835368)
Yes, that is true. Aider expects to work with the LLM to automatically apply edits to the source files. This requires precision from the LLM, which is what breaks down when you overload them with context.
[ Parent | on HN ]
Re:Qwen2.5-1M: Deploy your own Qwen with context length up t... by noname120 on Monday January 27, 2025 @11:25AM (#42842763)
Not true. In Aider the patch produced by the LLM is sent to a second model that is just tasked with fixing the patch — it works wonders.
[ Parent | on HN ]
Re:Qwen2.5-1M: Deploy your own Qwen with context length up t... by anotherpaulg on Monday January 27, 2025 @12:18PM (#42843365)
Yes, aider can also work in architect/editor mode [0] which tends to produce the best results [1]. An architect model solves the coding problem and describes the needed changes however comes naturally to it. The editor model then takes that solution and turns it into correctly formatted instructions to edit the files.

Too much context can still confuse the LLMs in this situation, but they may be somewhat more resilient.

[0] https://aider.chat/2024/09/26/architect.html

[1] https://aider.chat/2025/01/24/r1-sonnet.html

[ Parent | on HN ]
Re:Qwen2.5-1M: Deploy your own Qwen with context length up t... by trentnelson on Monday January 27, 2025 @12:00PM (#42843159)
Based on an earlier comment, I think the person you're replying to is the author of aider.
[ Parent | on HN ]