c6ai/katib

Katib is a Kubernetes-native system for automated machine learning (AutoML) that supports hyperparameter tuning, early stopping, and neural architecture search. It's a framework-agnostic tool that can optimize machine learning models written in any language and supports popular frameworks like TensorFlow, PyTorch, and XGBoost. Katib is designed for data scientists, machine learning engineers, and researchers who want to automate the process of finding optimal model configurations using various search algorithms like Random Search, Bayesian Optimization, and HyperBand.