TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

Q* – Clues to the Puzzle? [video]

15 点作者 MaximilianEmel超过 1 年前

3 条评论

luke-stanley超过 1 年前
His core claims seems spot on to me, and his approach is nice and methodical. Not sure about the interpretation of it. I expect there are several different &quot;slow but better&quot; experiments in agency that OpenAI have been researching, among several different lines, including the ones he pointed to! Including Everything of Thoughts - XoT,Graph of Thoughts - GoT,STaR and so on. Using meta-cognition with rewards for a guided space search is a sensible area of research with different approaches worth looking into. (There is a good tweet storm by Carlos E. Perez on XoT and some related methods here: <a href="https:&#x2F;&#x2F;twitter.com&#x2F;IntuitMachine&#x2F;status&#x2F;1724454898610167973" rel="nofollow noreferrer">https:&#x2F;&#x2F;twitter.com&#x2F;IntuitMachine&#x2F;status&#x2F;1724454898610167973</a> )
thekevan超过 1 年前
Although the style and subject area is very different, this guy&#x27;s videos are on a Matt Wolfe level for quality and usefulness.
AlbertoGP超过 1 年前
I like how he follows the trail of papers, tweets, and other announcements.