Show HN： 黑客新闻评论员眼中的编码模型艺术现状
阅读原文· hnup.date根据Hacker News评论者的讨论，当前编码模型的最新技术进展显著。模型在代码生成、补全和错误修复等任务上表现突出，部分模型在特定基准测试中的准确率已超过90%。评论指出，模型对常见编程语言的支持日趋成熟，但在处理复杂逻辑或边缘案例时仍有局限。开源模型与闭源商业模型之间的差距正在缩小，开发者可用的工具选择更加丰富。社区关注点集中在模型的实用性、运行效率及与现有开发流程的集成能力上。
State of the Art of Coding Models, According to Hacker News Commenters
The space of AI-assisted coding is evolving rapidly. This is an attempt at staying up to date with the latest developments, by capturing popularity and user sentiment of coding models from Hacker News comments. Updated daily.
Each day, the pipeline
Gets the 200 most popular posts in a 24h window from the Hacker News API.
Prompts an LLM to select those posts whose titles are about LLMs or coding in general (max. 50), as we expect more relevant discussions in those posts (assumption on my part).
For each post, sends the title and comments to Gemini and asks it to identify models from the OpenRouter model list and rate the sentiment towards each mentioned model per comment.