Hacker News Reader

> It’s far larger than the resulting model.

Is it? How many different books are we talking about, and how much information is that, after conversion to text and lossless compression? Images, maybe, but text?

View on HN

These models are trained on way more than just books. GPT-3 was trained on about half a terabyte of filtered plaintext and the training corpuses have grown significantly by then by all accounts.

I imagine that compresses by ~90%, and current top commercial models have a couple of trillion parameters, don't they?

They aren’t trained on compressed plaintext so I’m not sure of the relevance there. But regardless it’s my understanding that’s modern models are trained with orders of magnitude more storage than their parameters require. But it’s possible I’m incorrect. This is getting to the fringe of my knowledge of concrete LLM details.

The relevance is because the LLMs are storing information, not the explicit text, so we want to know how much actual information they need to store (this being an information theoretic argument). The representation in the parameters doesn't necessarily need 1 parameter per character, if the text is highly redundant.

Reply on news.ycombinator.com