2 ms·
OpenAI's Q* model can solve grade-school math
- saithound 3y agoEditorialized submission title. Original: "Unpacking the hype around OpenAI’s rumored new Q* model"
- free_bip 3y agoI'll save everyone a click: The only actual information in this article is "Reuters says an anonymous source says Q* is capable of doing grade-school level math"
- supriyo-biswas 3y agoDoes somebody have the links to the papers about replacing auto-regressive token prediction with planning that Yann LeCun talks about in the referenced tweet?
- verdverm 3y agoI would guess that is generally about DNN vs DRL (Deep Reinforcement Learning), but I'm not sure what the tweet you reference said. They are implemented with different algos & models ...hand wavy... Something like ChatGPT/LLaMA vs AlphaGo https://deepmind.google/technologies/alphago/ https://deepmind.google/technologies/alphago/