SWE-Serve 揭示本地测试与实时服务之间的差距
原标题：How SWE-Serve Exposes the Gap Between Local Tests and Live Serving
AI 摘要
NVIDIA 技术博客介绍 SWE-Serve，一个与 SGLang 团队合作开发的基准，用 53 个来自 SGLang 已合并 PR 的任务评估 AI 编码代理对推理服务软件的修改。在 19 个含实时服务检查的任务中，同一批补丁排除实时检查时通过率为 69.4%，加入完整验证器后降至 45.9%，约三分之一补丁在真实服务器加载模型后失败。该基准还显示跨多个运行时域的任务通过率明显低于单一域任务。
正文节选
An AI coding agent’s patch can pass tests yet fail when the server loads a real model and handles requests. Evaluating changes to inference-serving software therefore requires checking the full serving path, including whether the system returns correct results through its public interface. Developed with input from the SGLang team, SWE-Serve evaluates this gap with 53 tasks derived from merged changes to SGLang, an open-source system for serving large language models. Across 19 tasks with live-s