I take your point that the linear/exponential terminology is a bit dubious. But the simple way to make sense of it is just going by various benchmarks. E.g. the

I take your point that the linear/exponential terminology is a bit dubious. But the simple way to make sense of it is just going by various benchmarks. E.g. the power-law relationship between the model accuracy and the model size: https://eliaszwang.com/paper-reviews/scaling-laws-neural-lm/ https://eliaszwang.com/paper-reviews/scaling-laws-neural-lm/