News for hackers. Stuff that matters.
Re:Getting 50% (SoTA) on Arc-AGI with GPT-4o by imperfect_light on Monday June 17, 2024 @10:28PM (#40713512)
A bit of a stretch given that Chollet is a researcher in deep learning and transformers and his criticism is that memorization (training LLMs on lots and lots of problems) doesn't equate to AGI.
Re:Getting 50% (SoTA) on Arc-AGI with GPT-4o by refulgentis on Monday June 17, 2024 @11:04PM (#40713693)
> A bit of a stretch

Is that true?

C.f. what we're discussing

He's actively encouraging using LLMs to solve his benchmark, called ARC AGI.

8 hours ago, from Chollet, re: TFA

"The best solution to fight combinatorial explosion is to leverage intuition over the structure of program space, provided by a deep learning model. For instance, you can use a LLM to sample a program..."

Source: https://x.com/fchollet/status/1802801425514410275

Re:Getting 50% (SoTA) on Arc-AGI with GPT-4o by imperfect_light on Monday June 17, 2024 @11:25PM (#40713793)
The stretch was in reference to comparing Chollet to Einstein. Chollet clearly understands LLMs (and transformers and deep learning), he simply doesn't believe they are sufficient for AGI.
Re:Getting 50% (SoTA) on Arc-AGI with GPT-4o by refulgentis on Tuesday June 18, 2024 @12:39AM (#40714132)
I don't know what you mean, it's a straightforward analogy, but yes, that's right, except for the part where he's heralding this news by telling people the LLM is an underexplored solution space for a possible solution to his AGI benchmark he made to disprove LLMs are AGI.

I don't mean to offend, but to be really straightforward: he's the one saying it's possible they might be AGI now. I'm as flummoxed as you, but I think its hiding the ball to file it under "he doesn't mean what he's saying, because he doesn't believe LLMs can ever be AGI." The only steelman for that is playing at: AGI-my-benchmark, which I say is for AGI, is not the AGI I mean

Re:Getting 50% (SoTA) on Arc-AGI with GPT-4o by imperfect_light on Tuesday June 18, 2024 @01:51AM (#40714424)
You're reading a whole lot into a tweet, in his interview with Dwarkesh Patel he says, about 20 different times, that scaling LLMs (as they are currently conceived) won't lead to AGI.
Re:Getting 50% (SoTA) on Arc-AGI with GPT-4o by anoncareer0212 on Tuesday June 18, 2024 @03:17AM (#40714964)
You keep changing topics so I don't get it either, I can attest it's not a fringe view that the situation is interesting, seen it discussed several times today by unrelated people.
Re:Getting 50% (SoTA) on Arc-AGI with GPT-4o by imperfect_light on Tuesday June 18, 2024 @02:20PM (#40720617)
He's said it pretty clearly, an LLM could be part of the solution in combination with program synthesis, but an LLM alone won't achieve AGI.