Glossary
FLOPs
Floating-point operations, a count of the basic arithmetic steps a computer performs, used to measure AI training effort.
A floating-point operation is a single arithmetic calculation, such as multiplying two decimal numbers. The total number of these operations (FLOPs) used to train a model is a common measure of how much computing went into it.
Some laws and policies use training compute in FLOPs as a threshold, presuming that models trained with more than a set amount pose greater risks and carry extra duties.
Two similar-looking terms are easy to confuse. FLOPs counts total operations, while FLOP/s (often written FLOPS) is a rate, operations per second, used to describe how fast a chip or computer is.