Group-Level Data Selection for Efficient Pretraining

The efficiency and quality of language model pretraining are largely determined by the way pretraining data are selected. In this paper, we introduce Group-MATES, an efficient group-level data selection approach to optimize the speed-quality frontier of language model pretraining. Specifically, Group-MATES parameterizes costly group-level selection with a relational data influence model. To train this model, we sample training trajectories of the language model and collect oracle data influences alongside. The relational data influence model approximates the oracle data influence by weighting individual influence with relationships among training data. To enable efficient selection with our relational data influence model, we within partition each the cluster dataset independently into small clusters . Experiments using relationship on DCLM weights 400M-4x, and 1B-1x, select data and 3B-1x show that Group-MATES achieves 3.5%-9.4%