Training adjusts a model's millions of internal numbers, called weights, by repeatedly comparing its guesses to known answers. It is a one-time, compute-heavy investment; most video teams use an already-trained model and never train from scratch.
Definition
Teaching a model by showing it labelled examples until its internal weights predict well. Done once, then reused; far costlier per run than inference.
Training adjusts a model's millions of internal numbers, called weights, by repeatedly comparing its guesses to known answers. It is a one-time, compute-heavy investment; most video teams use an already-trained model and never train from scratch.
Also known as
model training
Specialist software house for video, real-time and AI products. Founded 2005. 50 in-house engineers.
