Skip to content

Glossary: Computer Vision Beyond the Basics

INT8

8-bit whole numbers, used in quantization instead of 32-bit floats. This makes a model about four times smaller.

Where the book teaches it

Related terms

QuantizationCalibration