Meta 对比 Muse 模型在分布外任务的表现

AI at Meta· @AIatMeta · X·2026-10-13 01:29· 20 分钟前AI 评分31

AI 导读

为了量化模型处理陌生任务的能力，我们将后训练中训练过的任务的测试时计算扩展与其他任务进行了对比。 两个模型在分布外任务上的提升相似，但在相同的分布内得分下，Muse Wonder Full 在陌生任务上的表现优于 Muse Spark 1.3 趋势所预测的水平。我们将这一上移归因于更强的基座模型，它能更有效地将后训练带来的增益迁移到我们未曾预料到的任务上。

正文

To quantify how the models handle unfamiliar tasks, we compared test-time compute scaling on tasks trained on during post-training against other tasks.

Both models improve similarly out-of-distribution, but given the same in-distribution score, Muse Wonder Full performs better on unfamiliar tasks than the Muse Spark 1.3 trend predicts. We attribute this upward shift to a stronger base model, which carries post-training gains over more effectively to tasks we didn't anticipate.

来源：AI at Meta · x.com