My training data is very large, which means it have many training steps each epoch. I want to use tf.keras.callbacks.ModelCheckpoint to save a model every 100,000 steps, and save all these models in disk, not rewrite the model files when saving a new one. But I have no idea how to do this after i read the documentation. It seems that this callback is designed to save a model by epoch.
So, do i have to implement a custom callback to save the model by steps?