DsDm:利用 Datamodels 进行模型感知的数据集选择
DsDm: Model-Aware Dataset Selection with Datamodels
AI总结:
本文提出 DsDm,将数据集选择形式化为面向目标任务与学习算法的优化问题,利用 Datamodels 显式建模训练数据效用,所选数据使语言模型在多类基准上获得相对基线 2 倍的计算收益。
摘要:
在为大规模模型训练选择数据时,标准做法是筛选符合人类数据质量观念的样本。这种筛选会产生定性上干净的数据点,直观上应能改善模型行为。但在实践中往往可能出现相反情况:我们发现,依据与“高质量”数据源的相似性进行选择,相比随机选择数据,可能不会提升性能,甚至会损害性能。为开发更好的数据选择方法,我们首先将数据集选择构建为一个可以直接求解的优化问题:给定目标任务、学习算法和候选数据,选择使模型性能最大化的子集。因此,该框架避免了手工设定的数据质量观念,转而显式建模学习过程如何使用训练数据点来预测目标任务。我们由此得到的方法显著提升了语言模型(LM)在预先指定任务和此前未见任务上的性能。具体而言,在选择代表标准 LM 问题的目标任务并在多样化的留出基准上评估时,我们所选数据集相较基线方法带来了 2 倍的计算乘数。
英文摘要:
When selecting data for training large-scale models, standard practice is to filter for examples that match human notions of data quality. Such filtering yields qualitatively clean datapoints that intuitively should improve model behavior. However, in practice the opposite can often happen: we find that selecting according to similarity with "high quality" data sources may not increase (and can even hurt) performance compared to randomly selecting data. To develop better methods for selecting data, we start by framing dataset selection as an optimization problem that we can directly solve for: given target tasks, a learning algorithm, and candidate data, select the subset that maximizes model performance. This framework thus avoids handpicked notions of data quality, and instead models explicitly how the learning process uses train datapoints to predict on the target tasks. Our resulting method greatly improves language model (LM) performance on both pre-specified tasks and previously unseen tasks. Specifically, choosing target tasks representative of standard LM problems and evaluating on diverse held-out benchmarks, our selected datasets provide a 2x compute multiplier over baseline methods.
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。