At this seminar, we are going to showcase some more advanced techniques for training your networks. While the simple method of running the training loop for a given number of epochs is sometimes enough, you may wish to have more control over the process. Using monitoring tools and callbacks, it is possible to tune hyperparameters on the fly, save checkpoints and much more.

Goals:

• Learn how to use the MLFlow library.
• Write callback functions for logging the training losses and tuning hyperparameters.
• Gain a deeper understanding of the hyperparameters used in the training loop.