upvote

points

by YetAnotherNick21 hours ago |
comments
upvoteby ertgbnm21 hours ago|
next
[-]
Knowledge benchmarks can't really be improved upon via distillation or RL. It requires those facts be added to the training corpus and for the model to memorize them better. Neither distillation or RL really do that and thus we shouldn't expect improvements on SimpleQA unless some other interventions are being made.

Model intelligence and knowledge aren't necessarily directly related. If we can pack greater intelligence and agency at the cost of it forgetting factoids, that would actually be a good thing. We don't need LLMs to memorize facts, we need them to learn how to interact with the world such that they can find the facts that are necessary and surface them to the user.

If we could distill all of the knowledge out of an LLM and just be left with a very agentic model that only knows facts in it's context, I think some very interesting stuff would happen.

reply
upvoteby slashdave21 hours ago|
parent|
next
[-]
RL is more than facts. Synthetic feedback is an obvious approach. Does the model suggest code that compiles and performs well?
reply
upvoteby YetAnotherNick18 hours ago|
parent|
prev|
[-]
Lot of the things aren't facts that could be stated. No one can just see the dictionary or translation of words and start talking in that language.

There isn't a clear definition of what is knowledge and what is intelligence. Is being able to write in C knowledge? Is knowing undefined behaviour in that knowledge?

reply
upvoteby onlyrealcuzzo21 hours ago|
prev|
[-]
> Well for one, we know for certain there is Mythos which is meaningfully better.

Do we?

Have you used it?

What is "meaningfully" better? It's not 3-4 orders of magnitude better. That is definitely happening for smaller models.

reply
upvoteby YetAnotherNick18 hours ago|
parent|
[-]
What do you mean by 3-4 orders of magnitude better? Was Einstein 3-4 order of magnitude better than us?

Meaningful in the sense it could find security vulnerabilities in browser and kernel that >99% of the engineers couldn't find.

reply
upvoteby onlyrealcuzzo13 hours ago|
parent|
[-]
> What do you mean by 3-4 orders of magnitude better? Was Einstein 3-4 order of magnitude better than us?

I'm talking about output quality compared to parameter size.

Mythos is not 4 orders of magnitude larger than Opus - it's quite possible no LLM model ever reaches that size (likely even), and it's output is only barely better...

reply
upvoteby YetAnotherNick8 hours ago|
parent|
[-]
Where did 4 order of magnitude even come from? If I were to guess it is just 5x larger based on the pricing, so not even 1 order of magnitude.

> Mythos is not 4 orders of magnitude larger than Opus

Again can you define this. How would 4 order of magnitude better look like?

reply