Content

HPC-Häppchen: Deep Learning I/O: Profiling and Optimization

Efficient data input pipelines are essential for optimizing Deep Learning model performance, especially when dealing with large datasets. This presentation explores how to profile and optimize data pipelines using TensorBoard's built-in tools. We will cover key techniques for identifying bottlenecks, improving data loading efficiency, and maximizing GPU utilization.

Access:

To access the event, click on the following link: 

Big Blue Button Link <https://webconf.hrz.uni-marburg.de/n/rooms/hsy-cmh-knb-ndw/join> 

Access Code: kw7i98