Rohan Paul· @rohanpaul_ai · X·· 6 天前AI 评分69
AI 导读
OpenAI 发布 722 篇 AI 生成的数学证明手稿，覆盖 372 个数学结果。其内部模型在测试中处理约 4K 个研究问题，再经分组和筛选形成结果族，几乎所有结果都来自同一标准设置，每个结果平均使用相当于 3 小时 ChatGPT Pro 思考的算力。其中一篇论文将两个 n×n 矩阵乘法的已证明复杂度从 AlphaEvolve 在 8 月创下的约 n^2.371177 降至约 n^2.25，并提供 Lean 形式化证明；这是理论界，尚不能直接为真实 GPU 工作负载带来更快例程。
正文
OpenAI today released 722 AI generated math proofs.
This is one of the achievement, on the topics of Matrix multiplication
Here it lowers the proven cost of multiplying 2 n×n matrices from about n^2.371177, the record that AlphaEvolve set in August, to about n^2.25.
The result comes with a Lean formalization, but it is a theoretical bound and does not yet hand engineers a faster routine for real GPU workloads.
OpenAI published the full papers behind its open-problems claim: 722 manuscripts covering 372 math results. OpenAI built this catalogue by giving its internal model roughly 4K research problems during testing, then grouping and filtering the output into 372 significant result families. Almost all of those results came from the same standard setup, and each used on average the computing equivalent of 3 hours of ChatGPT Pro thinking.在 X 查看被引用的帖子
来源：Rohan Paul · x.com