Data Mixing as Mixture Experiment: Response Surface Methodology and Optimal Design for Large Language Model Pretraining

Abstract

Data mixing is a central design problem in large language model pretraining: given a fixed token budget, practitioners must decide how much data to allocate to each domain. Recent proxy-based methods address this problem by training small models on candidate mixtures, fitting a response model, and using the response to select mixtures for larger-scale training. We show that this workflow has the structure of a classical mixture experiment. Under this view, data domains are mixture components, token shares are component proportions, proxy-training runs are experimental design points, and validation loss defines a response surface over the probability simplex. We develop this formulation using sparse second-order Scheff\'{e} response-surface models and construct model-robust $\mathcal{I}$-optimal designs for proxy data-mixing experiments. Using RegMix as an empirical case study, we demonstrate how the framework can both interpret observed mixture responses and design more efficient proxy experiments. The Scheff\'{e} analysis shows that domain value is strongly relational: several domains that are weak under additive effects become favourable through pairwise interactions, especially through combinations with web-derived text. The sparse Scheff\'{e} model preserves mixture rankings across model scales and remains competitive with a flexible machine-learning predictor while providing an explicit decomposition of additive and interaction effects. In a simulation study calibrated to observed proxy-training responses, model-robust $\mathcal{I}$-optimal designs recover the relevant mixture ordering after removing about 25\% of the original proxy runs. These results suggest that LLM data mixing should be treated not only as a prediction problem, but also as an experimental-design problem in which the proxy mixtures themselves can be chosen to improve statistical efficiency.