AI 导读
两款模型在最高努力下于 GDPval-AA v2.1 中均出现退步,原因是交付内容更短,更常遗漏必需元素。https://t.co/J7cHxHve2c
正文
Both models regress in GDPval-AA v2.1 at max effort, driven by shorter deliverables that more often omit required elements. https://t.co/J7cHxHve2c
来源:@ArtificialAnlys · x.com