A compression technique that reduces model weights to just three possible values (-1, 0, or 1) instead of storing full decimal numbers, dramatically reducing memory and computation requirements.