From the course: TensorFlow 2.0: Working with Images

Unlock the full course today

Join today to access over 24,500 courses taught by industry experts.

Monitoring the training process

Monitoring the training process

- So far, things have worked relatively smoothly. Our notebooks have worked and we haven't had to do too much work. We've been able to use transfer learning and the TensorFlow hub. Now on the job, things can often be more complicated. Let's put ourselves in the shoes of a deep learning engineer, creating a model architecture and training a model from scratch on the CFR 10 dataset. Now, when I'm working on training models, here are some of the questions that I have. How can I determine the epoch which gives me the best model performance before over-fitting occurs? And how can I stop training if the model is not improving or is over-fitting? Other questions I have is how often should I save the model during the training process? And finally, is there a way to visualize the model training process? TensorFlow's solutions to these questions is Callback functions. We'll be looking at three of the most used classes,…

Contents