模型30
GPT-6.2 两版本减少格式使用
原标题:Both models also use much less formatting than their previous releases, with far fewer headings and em dashes per response.
两个模型相比此前版本也大幅减少了格式使用,每次回复中的标题和破折号都少得多。 ➤ 每次回复标题数:GPT-6.2 Astra 1.99 降至 0.11,GPT-6.2 Sol 1.85 降至 0.25 ➤ 每次回复破折号数:GPT-6.2 Astra 3.19 降至 0.70,GPT-6.2 Sol 3.01 降至 0.51
中文译文尚未完成,先为你展示原文。
Both models also use much less formatting than their previous releases, with far fewer headings and em dashes per response.
➤ Headings per response: GPT-6.2 Astra 1.99 to 0.11, GPT-6.2 Sol 1.85 to 0.25
➤ Em dashes per response: GPT-6.2 Astra 3.19 to 0.70, GPT-6.2 Sol 3.01 to 0.51
同一件事的其他报道2 篇
Artificial Analysis 评测 GPT-6.2 Astra 与 Sol 性格变化
Artificial Analysis · 原文
Artificial Analysis · 原文
