Finding Training Inefficiencies with CentML DeepView

Discover how to identify and address performance bottlenecks in deep learning training with CentML DeepView in this 40-minute conference talk from MLOps World: Machine Learning in Production. Learn about the challenges faced by ML developers in optimizing resource utilization and improving training efficiency. Explore DeepView, an open-source visual profiler designed specifically for ML developers, which offers intuitive performance visualizations and optimization suggestions. Understand how DeepView integrates with PyTorch and Visual Studio Code, and its capabilities in predicting performance to optimize deployment targets. Watch an interactive demonstration showcasing the optimization of a real model's training process, resulting in significant improvements in training throughput.

Finding training inefficiencies with CentML DeepView