Q* Hypothesis: Enhancing Reasoning, Rewards, and Synthetic Data

Re:Q* Hypothesis: Enhancing Reasoning, Rewards, and Syntheti... by tunesmith on Friday November 24, 2023 @03:45PM (#38408074)

Yeah I just saw the video from that researcher (later an OpenAI researcher?) that talked about it back in 2016... not that I understood much, but it definitely seemed that Q* was a generalization of the Q algorithm described on the previous slide. The optimum something across all somethings.

[ Parent | on HN ]

Re:Q* Hypothesis: Enhancing Reasoning, Rewards, and Syntheti... by maaaaattttt on Friday November 24, 2023 @03:56PM (#38408228)

If you have the possibility I would be quite interested in a link to the video or alternatively the name of the researcher you mention.

[ Parent | on HN ]

Re:Q* Hypothesis: Enhancing Reasoning, Rewards, and Syntheti... by andrew3726 on Friday November 24, 2023 @06:06PM (#38409359)

It's Noam Brown, he worked at Meta AI before on Cicero and No-hands Poker before that.

[ Parent | on HN ]

Re:Q* Hypothesis: Enhancing Reasoning, Rewards, and Syntheti... by resource0x on Friday November 24, 2023 @03:56PM (#38408236)

LeCun: Please ignore the deluge of complete nonsense about Q*. https://twitter.com/ylecun/status/1728126868342145481

[ Parent | on HN ]

Re:Q* Hypothesis: Enhancing Reasoning, Rewards, and Syntheti... by zaptrem on Saturday November 25, 2023 @02:32AM (#38411941)

As someone with a borderline acceptable understanding of RL this is the most accurate take so far.

[ Parent | on HN ]