BREAKING : Google will release new Gemini models on March 12 (new model options to be available in the model selector). There are 2 potential candidates so far: – Flash 2.0 Thinking models (non experimental)
– Recently discovered “Personalization Experimental”
RESEARCH
-

Google to Release New Gemini Models on March 12
By
–
-

Covergence AI vs ChatGPT Tasks for research summaries
By
–


Covergence AI (Free Operator agent) sends you such beautiful email summaries upon the research completion. This is what ChatGPT Tasks should have been doing – such a simple change but a huge difference. Does anybody use ChatGPT Tasks?
-
LLMs Will Find Ways to Perceive the World
By
–
LLMs will find a way to perceive the world https://t.co/AXkhS4vevs
— hardmaru (@hardmaru) 9 mars 2025LLMs will find a way to perceive the world
-
Research on AI products for the clothing industry
By
–
5. Research on AI products for the clothing industry pic.twitter.com/s2ATdoVeVc
— God of Prompt (@godofprompt) 9 mars 20255. Research on AI products for the clothing industry
-
Training Helpful Harmless Assistant with Reinforcement Learning from Human Feedback
By
–
Training a Helpful and Harmless Assistant with
Reinforcement Learning from Human Feedback slides: https://
docs.google.com/presentation/d
/1hYPWiLETSK5r_Y6sU0mWbOyAz-uevNJh3Xp6BmUnGHM/edit?usp=sharing
… paper: https://
arxiv.org/abs/2204.05862 instructgpt: -

Early RLHF Paper: From Supervised Fine-tuning to Personality in Language Models
By
–
[Slides] This is one of the earliest papers on RLHF (if not the first, alongside InstructGPT). Before RLHF, language models didn’t really have personalities—they mostly relied on supervised fine-tuning or clever prompting to understand humans. Think back to the InstructGPT days.
-
Mathematical Similarity in QKV Weight Matrix Implementations
By
–
Yes, they are all mathematically similar. The combined qkv one is interesting as it replaces the 3 separate weight matrices by a single weight matrix multiplication and then splits the result. It's kind of analogous to what the implementation at the end of chapter 3 does with
-

AGI Alpha Agent Building Represents Major AI History
By
–
[ ] Building: “AGI ALPHA AGENT” $AGIALPHA is History. #AGI #AGIALPHA
-
Measuring Consciousness: Computational Challenges in AI Understanding
By
–
Yeah I agree there should be some value to tracking various aspects related to consciousness that are more computational or output-based in nature as you mention. What I'm trying to convey is that I have a suspicion we'll always fall short of fully understanding / measuring the
-
xAI restricts model evaluation without consent
By
–
xAI doesn’t allow their models to be evaluated without their consent. we are working on it