Senior SWE-Bench: open-source benchmark that assesses agents as senior engineers
senior-swe-bench.snorkel.ai原文 ↗
Snorkel 发布 Senior SWE-Bench,定位为评估 agents 是否具备 senior engineer 级能力的开源 benchmark。digest 描述强调“assesses agents as senior engineers”,说明它试图从简单 issue 修复扩展到更高层的软件工程判断。即使页面正文没有被抓到,标题本身也反映出 SWE benchmark 正在从 patch correctness 向工程资深度建模移动。
–浏览
评论 · Comments