Unrelated to what I said, with all due respect.

sitkack · on June 11, 2024

It isn't tho, if you look at the bulk of tokens needed to train gen1 over LLMs and what is possible with better data and smaller models.

The fact that LLMs trained on dumptrucks full of data cannot achieve what a middle schooler begrudgingly achieves using existence and snide remarks.

flaminHotSpeedo · on June 11, 2024

I'd consider it related, for two reasons:

First and foremost (and what I think the parent comment is getting at) whether you could truly say an LLM "understands" language

As a secondary quibble in the context of the parent post, though big overall, I would argue that the whole argument is moot since a human couldn't possibly learn the way an LLM does in a single lifetime