Callbacks
Explore how to implement Keras callbacks such as ModelCheckpoint to save model weights during training and EarlyStopping to halt training when improvements plateau. This lesson helps you efficiently manage model training by checkpointing and reducing unnecessary epochs, ensuring better performance and resource use.
We'll cover the following...
Checkpointing is the process of saving the computational state of an application to recover it in case of a system failure. Checkpointing takes a snapshot of the system state that we can use as a starting point for a new run of our application. During the DL model training phase, we can checkpoint model weights. These weights can be further used for making predictions or training a new model. Let’s discuss the use of the ModelCheckpoint callback class of Keras to save model weights.
The ModelCheckpoint callback class
The ModelCheckpoint ...