Efficiently Estimating Optimal Hyperparameter Scaling Laws through Power-Law Entropy Search

2026-09-01

AI Summary

本文提出了一种名为Power-Law Entropy Search的方法，旨在通过自适应实验高效估计大规模语言模型训练中最佳超参数的缩放规律，大幅减少了计算资源的需求。

Abstract

Optimal hyperparameter scaling laws describe how the best hyperparameters for large language model (LLM) training change with model and data scale, enabling practitioners to predict optimal configurations at production scales without expensive large-scale tuning. However, estimating these scaling laws conventionally requires exhaustive grid searches over thousands of training runs, consuming enormous computational resources. We introduce Power-Law Entropy Search (PLES), a computational cost-aware acquisition function built on multi-fidelity Bayesian optimization that efficiently estimates optimal hyperparameter scaling laws through adaptive experimentation. A key innovation in PLES is that it searches for candidates that reduce the overall uncertainty of a scaling law estimate, instead of optimizing a single objective function. At each iteration, PLES selects the candidate configuration that maximally reduces the uncertainty of the scaling law estimates per unit computational cost, naturally favoring informative small-scale experiments. We evaluate PLES on synthetic benchmarks, surrogate models fitted to real LLM training data, and actual LLM pre-training runs. Across all settings, PLES converges to accurate optimal hyperparameter scaling laws using less than one-tenth of the computational budget required by conventional grid search and other baselines.

Problem

• hyperparameter scaling laws
• large language model
• computational resources
• grid search

Innovation

• Power-Law Entropy Search
• multi-fidelity Bayesian optimization
• computational cost-aware
• adaptive experimentation
• uncertainty reduction