Meta 对比 Muse Wonder Full 与 Muse Spark 1.3 的分布外表现
To quantify how the models handle unfamiliar tasks, we compared test-time compute scaling on tasks trained on during post-training against other tasks.
为量化模型处理不熟悉任务的能力，我们对比了后训练中训练过的任务与其他任务上的测试时计算扩展。 两个模型在分布外都有相似提升，但在相同分布内得分下，Muse Wonder Full 在不熟悉任务上的表现优于 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，它能更有效地把后训练收益迁移到我们未预料到的任务上。
正文
当前显示原文，中文全文译文尚未就绪。
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.