3 ms·
It doesn't. This is particularly tough to search for and i'm not on social media. I'd be surprised if Le Cunn somehow thought these reasoning models were someho
by throwaway314155 2y ago
It doesn't. This is particularly tough to search for and i'm not on social media. I'd be surprised if Le Cunn somehow thought these reasoning models were somehow architecturally unique from a good old LLM. It's all in the training regime, right?
In any case I'll take your word for it, but that's still surprising to me.
- tucnak 2y agoJust search for "System 1 / System 2" and add word "researcher" there, filter by time <1y< and you'll find plenty. OP is right that a bunch of people made far-going predictions, & now doing mental gymnastics to justify they were right all along. Most notably, LeCun https://www.threads.net/@yannlecun/post/DD0ac1_v7Ij https://www.threads.net/@yannlecun/post/DD0ac1_v7Ij and Chollet https://www.youtube.com/watch?v=w9WE1aOPjHc https://www.youtube.com/watch?v=w9WE1aOPjHc Some argue they were kind of tricked into thinking that; see https://www.interconnects.ai/p/openais-o1-using-search-was-a-psyop https://www.interconnects.ai/p/openais-o1-using-search-was-a... and some other writing by Lambert which has turned out pretty much on-point as far as RL and verifiers are concerned.
- throwaway314155 2y agoFantastic thank you for linking so much information!