Hyperparameter Tuning with a Focus on Weights and Biases Sweeps

Explore hyperparameter tuning techniques for deep learning models in this 47-minute conference talk by Stacey Svetlichnaya, Deep Learning Engineer at Weights & Biases. Learn high-level approaches and useful visualizations for hyperparameter search, with concrete examples from semantic segmentation, language understanding, and other domains. Discover how to adapt powerful deep learning models to new use cases, finetune pretrained networks on new data, build intuition for complex models, and apply various architectures to unique problems. While focusing on Weights & Biases Sweeps as a comprehensive tool, gain framework-agnostic practices to accelerate your progress in hyperparameter exploration, regardless of your development setup.