LogoThread Easy
  • Explorer
  • Composer un thread
LogoThread Easy

Votre partenaire tout-en-un pour les threads Twitter

© 2025 Thread Easy All Rights Reserved.

Explorer

Newest first — browse tweet threads

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

确实,随着模型能力一路狂飙,我自己在日常用的时候已经好久不做角色设定了。

这种问法曾经其打开了一组可调的认知滤镜,这也是为什么角色设定曾经那么重要,它让模型在某个子空间里模拟一个稳定的推理路径。

只是今天模型强到一个 prompt 就能自然切换视角,我们反而开始忘记这层机制的存在。

这条提醒挺好的:
世界上没有“你”,是我们在选择让 AI 作为“谁”来思考。

确实,随着模型能力一路狂飙,我自己在日常用的时候已经好久不做角色设定了。 这种问法曾经其打开了一组可调的认知滤镜,这也是为什么角色设定曾经那么重要,它让模型在某个子空间里模拟一个稳定的推理路径。 只是今天模型强到一个 prompt 就能自然切换视角,我们反而开始忘记这层机制的存在。 这条提醒挺好的: 世界上没有“你”,是我们在选择让 AI 作为“谁”来思考。

行道途中。非求速成,惟求通达。 2023 年扎进AI ,打通Know-How,不少赚钱项目,踩过坑,也见过光。 围城里待得够久了,出来聊聊世界,聊聊技术、聊聊赚钱。

avatar for 凡人小北
凡人小北
Wed Dec 10 07:19:45
RT @jclin808: When I walked around the NeurIPS 2025 poster and oral sessions, it was clear that LLMs, RL, and Agents are still the dominant…

RT @jclin808: When I walked around the NeurIPS 2025 poster and oral sessions, it was clear that LLMs, RL, and Agents are still the dominant…

ai agents @hud_evals | owned @AIHubCentral (1 million users,acq.) ex climate protester🦦 dont do the deferred life plan

avatar for Minh Nhat Nguyen
Minh Nhat Nguyen
Wed Dec 10 07:16:12
RT @TheSIAMNews: Ken Ono, a 2026 SIAM Annual Meeting invited speaker, was featured in @WSJ:

"The Math Legend Who Just Left Academia—for an…

RT @TheSIAMNews: Ken Ono, a 2026 SIAM Annual Meeting invited speaker, was featured in @WSJ: "The Math Legend Who Just Left Academia—for an…

@axiommathai : careers@axiommath.ai

avatar for Carina Hong
Carina Hong
Wed Dec 10 07:00:40
RT @PandaTalk8: 做X增长,我的核心理念非常的简单:「多发」。 

第一,是你开始逼自己每天高质量评论50条, 原创至少10条。 

第二,你会就开始琢磨,我应该创作什么, 评论什么, 倒逼你去思考, 深度的思考, 倒逼你在X上停留更长的时间。…

RT @PandaTalk8: 做X增长,我的核心理念非常的简单:「多发」。 第一,是你开始逼自己每天高质量评论50条, 原创至少10条。 第二,你会就开始琢磨,我应该创作什么, 评论什么, 倒逼你去思考, 深度的思考, 倒逼你在X上停留更长的时间。…

独立开发者 | 个人IP教练 | 帮助新手在X上完成早期成长| 公众号:PandaTalk8

avatar for Mr Panda
Mr Panda
Wed Dec 10 06:59:05
RT @xiangyue96: There are competing views on whether RL can genuinely improve base model's performance (e.g., pass@128). The answer is both…

RT @xiangyue96: There are competing views on whether RL can genuinely improve base model's performance (e.g., pass@128). The answer is both…

ai agents @hud_evals | owned @AIHubCentral (1 million users,acq.) ex climate protester🦦 dont do the deferred life plan

avatar for Minh Nhat Nguyen
Minh Nhat Nguyen
Wed Dec 10 06:57:46
The disconnect between benchmark results and economic returns in AI is a nice validation that economic systems attach greater value to solutions for less explored problems that aren't easily derivable from existing ones.

The disconnect between benchmark results and economic returns in AI is a nice validation that economic systems attach greater value to solutions for less explored problems that aren't easily derivable from existing ones.

GP @ a16z ... questionable heuristics in a grossly underdetermined world

avatar for martin_casado
martin_casado
Wed Dec 10 06:52:29
  • Previous
  • 1
  • More pages
  • 1119
  • 1120
  • 1121
  • More pages
  • 5634
  • Next