optuna

Why we included this project

Tuning hyperparameters by hand is a chore most model training eventually runs into, and Optuna automates it without forcing you to restructure your code. Its define-by-run API lets you build the search space with ordinary Python conditionals and loops inside the training script, so the objective stays readable and easy to change as the model evolves. It also prunes unpromising trials early and can spread a single study across many workers with only small code changes, which pays off when each run is costly. Built-in plots and an optional web dashboard make it easy to review the optimization history and see which parameters actually matter. For teams that want a dependable, framework-agnostic tuner that works with PyTorch, TensorFlow, scikit-learn, and other libraries, it is a practical choice.