60-day mapFive-Minute AI LessonsPart 1 · What AI Is

Is a bigger model always better? — Big models, small models

Big isn't what's good — the right size for the job is.

1 · Think first

A big model is one with lots of parameters, which you met on DAY 05 — it knows a lot and holds up well on hard problems it's never seen.

But big models are slow and expensive, and they won't even fit on your phone. They only run on servers somewhere across the internet.

So sometimes what a big model learned gets poured into a small model. The job decides the right size.

2 · Experiment · Pick one

Here are six situations. Pick the big model or the small one for each. Every pick is marked right away, with the reason.

As you pick, you'll spot the rule — quick, frequent jobs go small; jobs that need long thinking go big.

Figure

They're not ranked by skill — they just live in different places.

A small model is fast because it's close by. A big model holds up on hard stuff because it knows a lot.

3 · Quiz

Three questions. Pick an answer and it's marked right away, with the reason why.

Big isn't what's good —
the right size for the job is.

The paper that formalized moving what a big model learned into a small model (distillation) — Distilling the Knowledge in a Neural Network (Hinton, Vinyals, Dean, 2015)
There's also research saying that balancing model size with the amount of training matters more than just making models bigger — Training Compute-Optimal Large Language Models (2022)