Hacker News Reader

Comment by ben_w | Hacker News Reader

To determine this, it would first need to be able to spell "raspberry" as letters rather than as tokens.

Given you also don't want it to memorise [for all tokens, count([for all letters]), this would probably be more like "here's two images, count all things in the big image that look like the thing in the small image", which can then be r's in a photo of a raspberry jam jar in a supermarket, or dragons in a photo of a furry convention, or whatever.

That said, they are competent enough at coding that I keep seeing them write code to do even simple tasks.

On a related note: why did I see Claude editing a file by using cat to write a python script to do a grep search and replace?

View on HN

> it would first need to be able to spell "raspberry" as letters rather than as tokens

Of any object in question they should be able to create a representation that allows correct assessment.

> Given you also don't want it to memorise

That is obviously necessary: what we want from the consultant is to check, not to remember. Answers must be correct and that implies having performed all due diligence - and being capable of doing it, before that. So, objects must be instanced internally in a way that allows effective handling. Counting letters is a good example of the ability (that must remain general).

> Given you also don't want it to memorise [for all tokens, count([for all letters])

Why not? You've memorized how words are spelled, and how sounds correspond with letters, and how concepts correspond with words. To the extent that there are shortcuts that enable compression you use these, and the model will do something similar.

Combinatorial explosion, and facts merely memorised is a huge waste of parameters that are better dedicated to effective reasoning. Not that we really know how to split facts from skills, though we are trying various approaches.

Being able to spell all the words then count letters is simpler, and more generalisable to other tasks, than memorising answers to all possible word questions.

That said, we're so bad at splitting facts from skills that trying to get them to memorise a bunch of facts might force them to learn a skill and generalise anyway.

Ah, I misunderstood what you meant. I was just trying to highlight that in order to answer these types of questions the model needs to memorize the spelling of each token. But you're right that that's all they need to memorize, and algorithms like counting are pretty simple for transformers to implement.

> Why not?

Because to "123x456" we want a reply that goes "this times that plus that...", not "Was that not nnnnnn?". If it does not perform its duty (returning solid checked answers) it is a liability.

Reply on news.ycombinator.com