getaskclaw/amber
琥珀式封存历史回放评测(AMBER):封进琥珀,重做当时的题 —— 真实事件回放、物理封存、预注册评分的方法规范
GitHub repository with 22 stars and 0 forks.
Language: Python
Topics: agent-evaluation, ai-agents, benchmark, benchmarking, evaluation-methodology, historical-replay, llm-evaluation, reproducibility