Twitxy Saved

Inria Flowers team

@FlowersINRIA

FLOWing Epigenetic Robots and Systems, Inria research team on developmental and social robotics.
Bordeaux, France flowers.inria.fr Joined January 2012
53 Following 1.7K Followers 624 Posts

Latest posts 5

Inria Flowers team @FlowersINRIA
Very important work and paper, congratulations @liweijianglw and co-authors !
Liwei Jiang @liweijianglw Oct 29, 2025
⚠️Different models. Same thoughts.⚠️

Today’s AI models converge into an π€π«π­π’πŸπ’πœπ’πšπ₯ π‡π’π―πžπ¦π’π§π 🐝, a striking case of mode collapse that persists even across heterogeneous ensembles.

Our #neurips2025 𝐃&𝐁 𝐎𝐫𝐚π₯ 𝐩𝐚𝐩𝐞𝐫 (✨𝐭𝐨𝐩 𝟎.πŸ‘πŸ“%✨) dives deep into this phenomenon, introducing 𝐈𝐧𝐟𝐒𝐧𝐒𝐭𝐲-π‚π‘πšπ­, a real-world dataset of 26K real-world open-ended user queries spanning 17 open-ended categories + 31K dense human annotations (πŸπŸ“ 𝐒𝐧𝐝𝐞𝐩𝐞𝐧𝐝𝐞𝐧𝐭 𝐚𝐧𝐧𝐨𝐭𝐚𝐭𝐨𝐫𝐬 𝐩𝐞𝐫 𝐞𝐱𝐚𝐦𝐩π₯𝐞) to push AI’s creative and discovery potential forward.

Now you can build your favorite models to be truly original, diverse, and impactful in the open-ended real world.

πŸ“Paper: arxiv.org/abs/2510.22954
πŸ“Data: huggingface.co/collections/li…

We also systematically reveal Artificial Hivemind across:

πŸ’₯ π†πžπ§πžπ«πšπ­π’π―πž πšπ›π’π₯𝐒𝐭𝐒𝐞𝐬: not only do individual LLMs repeat themselves, but different models produce strikingly similar content, even when asked fully open-ended questions.

πŸ’₯ πƒπ’π¬πœπ«π’π¦π’π§πšπ­π’π―πž πšπ›π’π₯𝐒𝐭𝐒𝐞𝐬: LLMs, LM judges, and reward models are systematically miscalibrated when rating alternative responses to open-ended queries.

(1/N)
View quoted post on X
Inria Flowers team @FlowersINRIA
Hundreds of new generative AI models, some close, some open, are released every months.

Most of them derive in some way (weights, data, architecture,etc) from previous models, but this information is rarely disclosed.

Can one infer their ancestors? Yes, with PhyloLM πŸ‘‡
Nicolas Yax @nicolas__yax Apr 23, 2025
πŸ”₯Our paper PhyloLM got accepted at ICLR 2025 !πŸ”₯
In this work we show how easy it can be to infer relationship between LLMs by constructing trees and to predict their performances and behavior at a very low cost with @StePalminteri and @pyoudeyer ! Here is a brief recap ⬇️
View quoted post on X
Inria Flowers team @FlowersINRIA
An overview of the work of the team @FlowersINRIA on grounding language with curiosity-driven agents (including LLM agents) !
🌍How to ground generative AI in the external world?

πŸ€–How to turn LLMs into curious agents that spontaneously explore, set their own goals, and become open-ended learners?

🀝How to align such agents on the social environment?

πŸ“ΊIn this talk, I outline a research program I have been developing with many colleagues over the last decade, and present experiments addressing these questions.

Thanks a lot to Stevan Harnad who invited me give this talk in his interdisciplinary seminar series (and stimulating questionsπŸ’‘at the end of the talk!) πŸ™

Video link below
View quoted post on X