When building PyTorch models for custom applications from scratch there's usually one problem: The model does not learn anything. In a complex project, it can be tricky to identify the cause: Is it the data? A bug in the model? Choosing the wrong loss function at 3 am after an 8-hour coding session?

In this talk, we will build a toolbox to find the culprits in a structured manner. We will focus on simple ways to ensure a training loop is correct, generate synthetic training data to determine whether we have a model bug or problematic real-world data, and leverage pytest to safely refactor PyTorch models.

After this talk, visitors will be well equipped to take the right steps when a model is not learning, quickly identify the underlying reasons, and prevent bugs in the future.

Clara Hoffmann

Affiliation: RC Trust - TU Dortmund

I'm a former ML Engineer in the geospatial domain and currently a Ph.D. student for trustworthy ML and Data Science at the RC Trust Ruhr. My main field of interest is Computer Vision and my guilty pleasure is assigning probability densities to all relevant variables in CV models. Application-wise I focus on Remote Sensing (Synthetic Aperture Radar) and Neuroscience (modeling trajectories of disease severity from MRI scans). I also collected a splash of experience in autonomous driving. I'm classically trained in Bayesian Statistics and interested in combining Bayesian approaches with self-supervised learning and deterministic DNNs.

visit the speaker at: Github • Homepage