Sakana AI 多智能体审稿系统检出73%核心论断错误
Sakana AI’s LLM Peer Review System Catches 73% of Core-Claim Errors
Sakana AI 用 3 个 Claude 智能体做论文审稿,核心错误检出率 73%,之前最好的系统才 15% 左右。
Sakana AI 在 TMLR 发表论文,提出 Multi-Layered Review(MLR)系统,由 3 个基于 Claude 的智能体组成。团队同时构建了包含 1,164 个错误的 Contradiction Benchmark。MLR 对核心论断错误的检出率为 73.43%,此前最佳系统仅为 14.81%。
Sakana AI’s LLM Peer Review System Catches 73% of Core-Claim Errors
Sakana AI’s TMLR paper introduces Multi-Layered Review, a 3-agent Claude-based reviewer, and a 1,164-error Contradiction Benchmark. MLR caught 73.43% of core-claim errors, versus 14.81% for the best prior system. The post Sakana AI’s LLM Peer Review System Catches 73% of Core-Claim Errors appeared first on MarkTechPost .