Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Data Mixing Optimization for Supervised Fine-Tuning of Large Language Models

About

Optimizing data mixtures for supervised fine-tuning (SFT) of large language models (LLMs) is critical for developing general-purpose models, yet this area remains underexplored. In this paper, we frame data mixing as an optimization problem and introduce a novel method designed to minimize validation loss. Our approach parametrizes the loss by modeling effective data transferred and leveraging scaling laws for fine-tuning. By experimenting with various small-scale data mixtures, we fit these parameters and derive the optimal weights. We provide both mathematical proofs and empirical results demonstrating that our algorithm achieves excellent overall and individual performance across all domains. Through controlled experiments, we show that models trained with our optimized weights perform on par with those using optimal weights determined via grid search, with per-domain loss only 0.66% higher than the best domain loss from grid search on average. Additionally, we show that reweighting popular SFT datasets using our method improves both validation loss and downstream performance. Finally, we discuss how our method can generalize to guide data selection for domain-specific models and provide insights into SFT.

Yuan Li, Zhengzhong Liu, Eric Xing• 2025

Related benchmarks

TaskDatasetResultRank
Generalization EvaluationAM-Thinking Distilled-math&code v1 (Unseen)
Average Unseen Score48.98
35
Language Model EvaluationTULU-3 (dev)
Knowledge Score59.15
35
Mathematical ReasoningMATH
MATH Score54.38
22
Code GenerationMBPP
MBPP Score55
7
Mathematical ReasoningGSM8K
GSM8K Score89.61
7
Code GenerationHumanEval
HumanEval Score54.88
7
Showing 6 of 6 rows

Other info

Follow for update