GitHub Trending Top 100

NVIDIA/SkillEvaluator

用于评估 AI 代理技能的多层框架,包括质量关卡、语义重叠检测、合成评估数据集生成以及实时代理评估,后者用于衡量技能如何影响代理行为。

原文:Multi-tier framework for evaluating AI agent skills with quality gates, semantic overlap detection, synthetic evaluation dataset generation, and live agent evaluation that measures how skills affect agent behavior.

Python

上榜次数
1次
最好名次
#29
单日最高新增
+122
最近上榜
2026-11-16

排名走势

1 期记录
记录不足,暂无法画走势

上榜明细

日期名次当日新增总 starFork
2026-11-16 周一 #29 +122 — —