Scaling Laws for Differentially Private Language Models

TL;DR

This paper establishes scaling laws for differentially private large language models, revealing how privacy constraints influence performance and guiding optimal training configurations.

Contribution

It introduces the first comprehensive scaling laws specifically modeling the dynamics of differential privacy in large language model training.

Findings

Scaling laws accurately predict DP LLM performance.

Guidelines for optimal compute-privacy-utility tradeoffs.

Insights into hyper-parameter choices under privacy constraints.

Abstract

Scaling laws have emerged as important components of large language model (LLM) training as they can predict performance gains through scale, and provide guidance on important hyper-parameter choices that would otherwise be expensive. LLMs also rely on large, high-quality training datasets, like those sourced from (sometimes sensitive) user data. Training models on this sensitive user data requires careful privacy protections like differential privacy (DP). However, the dynamics of DP training are significantly different, and consequently their scaling laws are not yet fully understood. In this work, we establish scaling laws that accurately model the intricacies of DP LLM training, providing a complete picture of the compute-privacy-utility tradeoffs and the optimal training configurations in many settings.