Batch size and epochs in TensorFlow - Cheat Sheet & Quick Revision

Recall & Review

beginner

What is batch size in machine learning?

Batch size is the number of training examples the model looks at before updating its internal settings (weights). It controls how many samples are processed together in one go.

beginner

Define epoch in the context of training a model.

An epoch is one full pass through the entire training dataset. After one epoch, the model has seen every training example once.

intermediate

How does increasing batch size affect training speed and memory?

Increasing batch size usually speeds up training because more data is processed at once, but it also uses more memory. Very large batches may reduce the model's ability to learn well.

intermediate

What happens if you increase the number of epochs too much?

If you train for too many epochs, the model might memorize the training data and perform poorly on new data. This is called overfitting.

beginner

In TensorFlow, how do you specify batch size and epochs when training a model?

You specify batch size and epochs in the model's fit() method like this: model.fit(x_train, y_train, batch_size=32, epochs=10). This trains the model with batches of 32 samples for 10 full passes over the data.

What does one epoch represent in training?

A One batch of data processed

B One update of model weights

C One pass through the entire training dataset

D One prediction made by the model

An epoch means the model has seen all training samples once.

If batch size is 64, what does this mean?

A The model trains on 64 epochs

B The model updates weights after every 64 samples

C The model sees 64 batches per epoch

D The model uses 64 layers

Batch size defines how many samples are processed before updating the model.

What is a risk of training with too many epochs?

A Overfitting the data

B Less memory usage

C Faster training

D Underfitting the data

Too many epochs can cause the model to memorize training data and not generalize well.

Which of these is true about increasing batch size?

A It decreases the number of epochs needed

B It reduces memory usage

C It always improves model accuracy

D It can speed up training but uses more memory

Larger batch sizes process more data at once, speeding training but using more memory.

How do you set batch size and epochs in TensorFlow's model.fit()?

A model.fit(x_train, y_train, batch_size=32, epochs=10)

B model.train(batch_size=32, epochs=10)

C model.fit(batch=32, epoch=10)

D model.train(x_train, y_train, batch=32, epoch=10)

The correct syntax is model.fit(x_train, y_train, batch_size=32, epochs=10).

Explain in your own words what batch size and epochs mean in training a machine learning model.

Think about how many samples the model sees at once and how many times it sees the whole dataset.

Describe the effects of choosing very large batch sizes and very high numbers of epochs on model training.

Consider both computer resources and model performance.