keras-tuner and kernel_tuner
These are complements operating at different abstraction levels: Keras Tuner optimizes hyperparameters for neural network models, while Kernel Tuner optimizes performance parameters for low-level GPU/CPU kernels, so they could be used together in a full ML pipeline.
About keras-tuner
keras-team/keras-tuner
A Hyperparameter Tuning Library for Keras
Supports multiple search algorithms including Bayesian Optimization and Hyperband alongside Random Search, enabling both efficient exploration and extensibility for custom algorithms. Uses a define-by-run syntax where hyperparameter spaces are specified directly within model-building functions, integrating seamlessly with TensorFlow 2.0+ and Keras Sequential/Functional APIs. Scales across distributed training scenarios while tracking trial history and checkpoints for reproducible optimization workflows.
About kernel_tuner
KernelTuner/kernel_tuner
Kernel Tuner
Scores updated daily from GitHub, PyPI, and npm data. How scores work