arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2402.17641cs.LGcs.AIcs.CLmath.OCstat.ML
收藏

变分学习对于大型深度网络是有效的

Variational Learning is Effective for Large Deep Networks

Technical University of Munich(慕尼黑工业大学);Munich Center for Machine Learning(慕尼黑机器学习中心);Technical University of Darmstadt(达姆施塔特工业大学);Tokyo Institute of Technology(东京工业大学);RIKEN Center for AI Project(理化学研究所人工智能项目中心)

Yuesong Shen, Nico Daheim, Bai Cong, Peter Nickl, Gian Maria Marconi, Clement Bazan, Rio Yokota, Iryna Gurevych, Daniel Cremers, Mohammad Emtiyaz Khan, Thomas Möllenhoff

AI总结:

本文通过实证证明 IVON 优化器在大型深度网络训练中可匹敌或超越 Adam,并展示其在微调、模型合并、泛化误差预测和数据敏感性估计方面的优势,反驳了变分学习无效的普遍看法。

摘要:

我们提供了大量实证证据,反驳了变分学习对大型神经网络无效的普遍看法。我们表明,一种名为 Improved Variational Online Newton(IVON)的优化器在从头训练 GPT-2 和 ResNets 等大型网络时,始终能够匹敌或优于 Adam。IVON 的计算成本与 Adam 几乎相同,但其预测不确定性更好。我们展示了 IVON 的几个新用例:改进大型语言模型中的微调和模型合并、准确预测泛化误差,以及忠实估计对数据的敏感性。我们发现压倒性的证据表明变分学习是有效的。

英文摘要:

We give extensive empirical evidence against the common belief that variational learning is ineffective for large neural networks. We show that an optimizer called Improved Variational Online Newton (IVON) consistently matches or outperforms Adam for training large networks such as GPT-2 and ResNets from scratch. IVON's computational costs are nearly identical to Adam but its predictive uncertainty is better. We show several new use cases of IVON where we improve finetuning and model merging in Large Language Models, accurately predict generalization error, and faithfully estimate sensitivity to data. We find overwhelming evidence that variational learning is effective.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑