LogoThread Easy
  • 探索
  • 線程創作
LogoThread Easy

Twitter 線程的一站式夥伴

© 2025 Thread Easy All Rights Reserved.

探索

Newest first — browse tweet threads

Keep on to blur preview images; turn off to show them clearly

sorry if this is a dumb question but are labs also RL'ing a model on native summaries / compactions?

like, instead of just doing a single inference pass with a lot of thinking tokens before the answer, we actually do multiple inference passes where the next one can have access to a summary generated by the previous one?

so instead of "think think think → answer", it goes like "think think think → summarize → think think think → summarize → think think think → answer", and then we RL on that?

I mean that *is* how humans solve problems, we don't keep all the reasoning in our heads, we have insights / aha moments that let us garbage collect the noise out and build incrementally better mental models of the problem (i.e., summaries) before actually solving it

sorry if this is a dumb question but are labs also RL'ing a model on native summaries / compactions? like, instead of just doing a single inference pass with a lot of thinking tokens before the answer, we actually do multiple inference passes where the next one can have access to a summary generated by the previous one? so instead of "think think think → answer", it goes like "think think think → summarize → think think think → summarize → think think think → answer", and then we RL on that? I mean that *is* how humans solve problems, we don't keep all the reasoning in our heads, we have insights / aha moments that let us garbage collect the noise out and build incrementally better mental models of the problem (i.e., summaries) before actually solving it

Kind / Bend / HVM / INets / λCalculus

avatar for Taelin
Taelin
Fri Dec 19 15:49:16
As I return from maternity leave, I'm excited to share the result of a fun maternity leave side quest: Nyla the NFT is a children's book about a star named Nyla, who learns that her unique characteristics are what make her an NFT. I enjoyed collaborating with the real-life Nyla @NylaCollection  on this project

As I return from maternity leave, I'm excited to share the result of a fun maternity leave side quest: Nyla the NFT is a children's book about a star named Nyla, who learns that her unique characteristics are what make her an NFT. I enjoyed collaborating with the real-life Nyla @NylaCollection on this project

Check it out here; you might spot some of your favorite NFTs as well! https://t.co/l47H0cgVMB

avatar for Maggie Hsu
Maggie Hsu
Fri Dec 19 15:49:12
RT @glangley: I'm incredibly proud that Flock Safety played a key role in working with first responders to find the suspect in the Brown an…

RT @glangley: I'm incredibly proud that Flock Safety played a key role in working with first responders to find the suspect in the Brown an…

Marketing at Andreessen Horowitz (@a16z)

avatar for Grace Ellis
Grace Ellis
Fri Dec 19 15:48:18
离谱,把 Github 做成应用商店。
> 想象一下,你发现了一个超酷的 GitHub 项目
> 你点开 Release 页面,结果被 50 个 Source Code (zip) 淹没

Github Store 来了,为 GitHub Releases 提供一个类似“应用商店”的跨平台体验,简单来说,就是给 GitHub 穿一件“应用商店”的皮。

https://t.co/mvrt8YXhkr

离谱,把 Github 做成应用商店。 > 想象一下,你发现了一个超酷的 GitHub 项目 > 你点开 Release 页面,结果被 50 个 Source Code (zip) 淹没 Github Store 来了,为 GitHub Releases 提供一个类似“应用商店”的跨平台体验,简单来说,就是给 GitHub 穿一件“应用商店”的皮。 https://t.co/mvrt8YXhkr

🧠在家居士 | 🥦素食者 | 🏃🏻马拉松爱好者 | 💰省钱小能手 | 搭🪜技术资深学者 | 👨‍💻科技宅 | 🆕更新狂 | 🆅 六边型战五渣

avatar for Geek
Geek
Fri Dec 19 15:46:33
RT @bamboo_farms: So excited! The @pollenrobotics team did such a great job with packaging!!

RT @bamboo_farms: So excited! The @pollenrobotics team did such a great job with packaging!!

Co-founder & CEO @HuggingFace 🤗, the open and collaborative platform for AI builders

avatar for clem 🤗
clem 🤗
Fri Dec 19 15:41:34
New video is out, teaching a Unitree G1 humanoid to walk using reinforcement learning (PPO).

First time I've got sim2real to work with robotics, sharing what I've learned and testing out how good the policy actually is by walking around outside on some semi challenging terrain.

New video is out, teaching a Unitree G1 humanoid to walk using reinforcement learning (PPO). First time I've got sim2real to work with robotics, sharing what I've learned and testing out how good the policy actually is by walking around outside on some semi challenging terrain.

Video: https://t.co/TOinCl47If

avatar for Harrison Kinsley
Harrison Kinsley
Fri Dec 19 15:32:21
  • Previous
  • 1
  • More pages
  • 329
  • 330
  • 331
  • More pages
  • 5634
  • Next