arXiv Artificial Intelligence· Ali Ansari, Haoran Sun, Andy Zeyi Liu, Mark Jabbour, Yongshan Ding, Steven Girvin, Yu He, Sohrab Ismail-Beigi, Aleksander Kubica, Owen D. Miller, Corey O'Hern, Vidvuds Ozolins, David Poland, A. Douglas Stone, Frank C. van den Bosch, Logan Wright, Navid Akbari, Santanu Antu, Kangle Cai, Andrew Calabrese-Day, Mateo C\'ardenes Wuttig, Meng Cheng, Barry T. Chiang, Ali Ghorashi, Shouzhen Gu, Haoyang Huang, Zhibo Kang, Lukas Kienesberger, Hantian Liu, Charles Lomba, Zhongling Lu, Wenchao Ma, Rohin E. McIntosh, Evan McKinney, Ivan Rojkov, Xulei Sun, Yarone Meir Tokayer, Naveen Balaji Umasankar, Mira Varma, Leda Wang, Qimin Wang, Tyler Wang, Haoyu Wei, Jinming Yang, Jinchen Zhao, Sherlock Tingrui Zhao, Qinyuan Zheng, Jay S. Zou, Lucas Baker, Arman Cohan, John Sous·· 3 小时前AI 评分60
前沿模型物理能力究竟如何？专家重新评分揭示评测缺陷与领先基准接近饱和
How Good Are Frontier Models at Physics? Expert Re-Grading Reveals Broken Evaluations and Near-Saturation of Leading Benchmarks
AI 导读
研究团队用专家复核六个常用物理基准，发现多数被判错的案例源于评分错误、参考答案有误或题目表述不清，而非模型物理推理出错。修正后 GPT-5.6-Sol 在 HLE-Physics 的 mean@4 从 47.3% 升至 78.7%，在 CMT-Benchmark 从 61.0% 升至 87.2%，在保留的 54 道 CritPt 题上 pass@4 达 94.4%。
来源：arXiv Artificial Intelligence · arxiv.org