Andrew Curran 指出，OpenAI 在 Navier–Stokes 公告中附上了解决该问题的内部模型（他称之为 GPT-Aeon）的图表，其测试时计算没有出现收益递减，性能仍在上升。他引用博客原文称，OpenAI 自 8 月 28 日起训练一个新的内部模型，该模型在包括数学在内的基准上表现前所未有，训练仍在进行、性能持续提升，而这已是两个月前的事。他引用的 @gdb 内容则称，具备大量测试时计算的前沿模型正成为科学发现的引擎。
作者引用 OpenAI 博客原文，指出内部模型在测试时计算上尚未出现收益递减，为观察前沿模型的科研能力提供了一手线索。
OpenAI posted a chart for the internal model that solved Navier–Stokes (the model I call GPT-Aeon) with the NS announcement. There were no diminishing returns on test-time compute, it was still going up. I will quote from the blog:
'Since August 28 we have been training a new internal model that has exhibited unprecedented performance in our benchmarks, including mathematics. This model’s training is ongoing and its performance continues to improve.'
That was two months ago.
frontier models, with lots of test time compute, are becoming the engine of scientific discovery在 X 查看被引用的帖子
来源：Andrew Curran · x.com