Floating-Point Operation (FLOP)
A floating-point operation is a single arithmetic calculation on decimal numbers; the total number of such operations used to train a model is a standard measure of its training compute.
What is Floating-Point Operation (FLOP)?
The EU AI Act presumes a general-purpose AI model has systemic risk if training used more than 10^25 FLOPs. FLOP/s (operations per second) measures hardware speed, which is different from the total FLOP count.
How it works
Training compute is roughly estimated as six times the number of parameters multiplied by the number of training tokens.
← All Artificial Intelligence & Superintelligence definitions · ← Back to the A–Z definitions