Skip to main content
All terms
Foundations

Batch Size

The number of examples processed in one training step before the weights are updated.

Definition

Batch size is the number of examples used to estimate the needed weight adjustment in one training step before the model's weights are updated. Larger batches give a steadier estimate and use hardware efficiently but require more memory. Smaller batches add randomness that can help the model generalize but may need more steps to settle. It is a key training setting, often tuned alongside the learning rate.