I see! Thanks. I remember now. Not in that form, i.e., not directly (still running too many other experiments). But a few thoughts: 1. Self-consistency does seem to be still ahead of any best-of-N approach. This is based on my experience but also a recent paper I saw: 18 Apr
RESEARCH
-
Western Ideals, Progress, and Technology’s Promise
By
–
When I moved to the 'West,' it wasn’t only for opportunity — it was for the ideals it claimed to uphold.
— Nina Schick (@NinaDSchick) 20 novembre 2025
A civilisation built on progress. A belief in reason, science, and technology as engines of human flourishing. That was the spirit of the Enlightenment. That was the promise… pic.twitter.com/lA7BjZ27EVWhen I moved to the 'West,' it wasn’t only for opportunity — it was for the ideals it claimed to uphold. A civilisation built on progress. A belief in reason, science, and technology as engines of human flourishing. That was the spirit of the Enlightenment. That was the promise
-

Chapter 4 on Inference-Time Scaling Now Available for Reading
By
–
If you are looking for something to read this upcoming weekend, chapter 4 on inference-time scaling is available now! https://
mng.bz/Dwra -

Olmo 3: Open Language Model for Reasoning and Tool Use
By
–
Announcing Olmo 3, a leading fully open LM suite built for reasoning, chat, & tool use, and an open model flow—not just the final weights, but the entire training journey. Best fully open 32B reasoning model & best 32B base model. 🧵
-
LeapSpace: AI-Assisted Research Workspace for Faster Discovery
By
–
Move from curiosity to discovery faster! LeapSpaceTM is a research-grade AI-assisted workspace that lets researchers work faster, think more deeply, and achieve more in a single secure environment.https://t.co/m8ViaqGBLw#LeapSpace #Research #Researchers #Elsevier pic.twitter.com/nwp1cGDKN0
— Catherine Adenle (@CatherineAdenle) 20 novembre 2025Move from curiosity to discovery faster! LeapSpaceTM is a research-grade AI-assisted workspace that lets researchers work faster, think more deeply, and achieve more in a single secure environment. https://
elsevier.com/academic-and-g
overnment/ai-for-researchers
…
#LeapSpace #Research #Researchers #Elsevier -

Grok detective for finding customer pain points in tweets
By
–
7. The "Pain Point" Detective Grok can read replies to find what people are complaining about now. Prompt: "Search for tweets containing the phrases 'I hate how', 'Why is it so hard to', and 'I wish there was a tool for' specifically related to [INSERT INDUSTRY, e.g.,
-
Grok 4.1: Higher EQ Test and Personal Interview
By
–
2. The "Emotional Rollercoaster" Story
— God of Prompt (@godofprompt) 20 novembre 2025
Grok 4.1 tests higher on EQ (Emotional Quotient) than previous models.
Prompt: "I have a personal story/lesson about [INSERT TOPIC, e.g., failing my first startup].
Please interview me. Ask me 5 probing questions, one by one, to get the… pic.twitter.com/RVpl2p1ByZ2. The "Emotional Rollercoaster" Story Grok 4.1 tests higher on EQ (Emotional Quotient) than previous models. Prompt: "I have a personal story/lesson about [INSERT TOPIC, e.g., failing my first startup]. Please interview me. Ask me 5 probing questions, one by one, to get the
-
Transformers Reveal AI as Discovery of a Second Brain
By
–
Bezos a raison sur les IA :
— VISION IA (@vision_ia) 20 novembre 2025
Nous n’avons pas inventé les transformers. Nous avons découvert un algorithme qui fait grandir des réseaux neuronaux tellement complexes que nous ne pouvons pas vraiment expliquer comment ils pensent.
Ça ressemble de moins en moins à du code, et de… pic.twitter.com/5I3BC6djplBezos is right about AIs: We didn’t invent transformers. We discovered an algorithm that grows neural networks so complex that we can’t really explain how they think. It’s starting to look less and less like code, and more and more like the discovery of a second type of brain,
-

Prof. García Márquez keynote on AI in Energy Production at ICSTSN2025
By
–

Just wrapped! A visionary keynote by Prof. Fausto Pedro García Márquez @faustospain on "The Outlook for AI in Energy Production" at #ICSTSN2025 Nov 20-21, 2025 IFET College of Engineering, India icstsn2025.ifet.ac.in/ #AIinEnergy #FutureOfEnergy #SustainableTech #SmartNations
→ View original post on X — @faustospain, 2025-11-20 07:02 UTC
-
Google Releases Gemini 3 Pro with Advanced Agentic Coding Capabilities
By
–
Google released Gemini 3 Pro with SOTA agentic coding and reasoning!
— Sumanth (@Sumanth_077) 20 novembre 2025
Replit has it integrated into Design Mode, and the results look solid.
These examples are pretty cool. https://t.co/1iaFNSFZSCGoogle released Gemini 3 Pro with SOTA agentic coding and reasoning! Replit has it integrated into Design Mode, and the results look solid. These examples are pretty cool.