Power-Law Entropy Search for Hyperparameter Scaling
Efficiently Estimating Optimal Hyperparameter Scaling Laws through Power-Law Entropy Search
PLES reduces computational costs for LLM hyperparameter tuning by 90%, making large-scale training more accessible to researchers and companies in resource-constrained environments.
Power-Law Entropy Search (PLES) efficiently estimates optimal hyperparameter scaling laws for large language model training. The method uses multi-fidelity Bayesian optimization to reduce uncertainty per computational cost. PLES converges to accurate scaling laws using less than one-tenth of the computational budget required by conventional grid search. The approach was evaluated on synthetic benchmarks, surrogate models, and actual LLM pre-training runs.
Efficiently Estimating Optimal Hyperparameter Scaling Laws through Power-Law Entropy Search
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.
