SMAC3

SMAC3 is a high-performance Python tool for Bayesian hyperparameter optimization and algorithm configuration.

SMAC3 (Sequential Model-based Algorithm Configuration) is a robust Python framework for hyperparameter optimization and general algorithm tuning. Built by the AutoML group at the University of Freiburg, it uses a Random Forest surrogate model to navigate complex search spaces. This architecture handles diverse parameter types (categorical, continuous, and conditional) more effectively than standard Gaussian Processes. You can deploy SMAC3 to optimize deep learning learning rates or fine-tune SAT solver heuristics: it balances exploration and exploitation to find peak performance across high-dimensional landscapes.

https://github.com/automl/SMAC3