DeepView
Interactive Profiling Meets Predictions.
Interactive performance profiling and debugging tool for PyTorch neural networks.
Quick guide on DeepView
Visually identify runtime bottlenecks at both model and source code level
Optimize batch size to maximize memory utilization & increase training throughout
Attain insights from GPU utilization at a layer granularity
Estimate your training cost and time using different cloud platforms like AWS, Azure and GCP
Optimize batch size to track energy consumption and environmental impacts