Maximize Your Data's Potential: Enhancing LLM Accuracy with Two-Phase Pretraining
This addresses the challenge of optimizing data mixtures for practitioners in AI/ML, offering incremental guidance for scaling pretraining effectively.
The paper tackles the problem of strategic data selection for pretraining large language models by formalizing a two-phase pretraining approach, finding it improves average accuracies by 3.4% and 17% compared to random ordering and natural token distribution.
Pretraining large language models effectively requires strategic data selection, blending and ordering. However, key details about data mixtures especially their scalability to longer token horizons and larger model sizes remain underexplored due to limited disclosure by model developers. To address this, we formalize the concept of two-phase pretraining and conduct an extensive systematic study on how to select and mix data to maximize model accuracies for the two phases. Our findings illustrate that a two-phase approach for pretraining outperforms random data ordering and natural distribution of tokens by 3.4% and 17% on average accuracies. We provide in-depth guidance on crafting optimal blends based on quality of the data source and the number of epochs to be seen. We propose to design blends using downsampled data at a smaller scale of 1T tokens and then demonstrate effective scaling of our approach to larger token horizon of 15T tokens and larger model size of 25B model size. These insights provide a series of steps practitioners can follow to design and scale their data blends.