41年Jeopardy!知识时间胶囊：单个免费本地模型的表现
原标题：Time Capsule of Testable Human Knowledge: 41 Years of Jeopardy! in a Single Free Local Model
AI 摘要
该研究首次在完整的Jeopardy!题库（1984-2025年，529,939条线索）上评估了单个9GB开源模型（Qwen2.5-14B，4位量化），严格强制回答下准确率达67.0%，事实类类别超过85%。研究对比了IBM Watson，指出训练数据暴露是两者共有的，但本地模型在训练截止后的线索上仍保持65%准确率，而Watson为0%。作者认为知识压缩能力已从服务器机房转移到便携文件，重新定义了Watson挑战。
正文节选
Time Capsule of Testable Human Knowledge: 41 Years of Jeopardy! in a Single Free Local Model Abstract In 2011, IBM’s Watson was something like a sealed capsule of its era’s queryable knowledge. Its DeepQA system defeated the strongest human Jeopardy! champions, but the knowledge that let it do so lived in a curated billion-document corpus running on a cluster of POWER7 servers, frozen at build time and impossible to move or copy. We show that the same kind of artifact, a snapshot of what a cultu