Optuna

Hyperparameter optimization (HPO) for ML models using Optuna. Use when tuning learning rate, regularization, architecture choices, or any numeric/categorical hyperparameter. Covers create_study/optimize quickstart, sampler selection (TPE, CMA-ES, grid, random, NSGA-II), pruners for early stopping (MedianPruner, HyperbandPruner), distributed search with RDBStorage, integrations with PyTorch Lightning and scikit-learn, and built-in visualization.

stanfish06 Updated 12 repo stars

File contents

stanfish06/skillquarium commit ee0c20a232

Frequently asked questions

npx get-skill add stanfish06/optuna