.@Neuralk_AI (startup du Microsoft GenAI Studio) lance TabBench, 1er benchmark IA tabulaire pour le commerce. NICL, son modèle entraîné sur Azure, atteint des performances à l’état de l’art sur des cas d’usage concrets en entreprise. Fiers de soutenir cette IA de terrain 🎉
We’re proud to share the first performance results of NICL (Neuralk In-Context Learning), the world’s first Tabular Foundation Model specialized for Commerce — achieving strong, state-of-the-art results on real industrial use cases.
More details in the thread 👇
- direct business impact from better predictions (e.g., better search, recs, analytics)
Detailed blog post and more info on our models dropping soon. Link in bio.
Tabular AI vs. LLMs: we went head-to-head on real-world data.
We compared @MistralAI’s new LLM classifier against our in-house @Neuralk_AI tabular models on a private product categorization dataset provided by a client (35k samples, 15 categories, highly imbalanced, noisy).
They capture both feature semantics and cross-column relationships - not just token sequences.
This translates to:
- More accurate classification results, especially across rare categories
- Clean integration into production pipelines without heavy pre-processing
Announcing the @jneurolang award for Best Paper of 2023! 👏
Information-Restricted Neural Language Models Reveal Different Brain Regions' Sensitivity to Semantics, Syntax and Context | https://t.co/hqRlK6SiAJ | by @a_pasquiou, @lakretz, @BertrandThirion & @chrplr
📢Open Position📢 The MIND team is hiring a new CEA research scientist (tenured position, shared w @Inria_Saclay).
🧠Come develop the next generation of AI-empowered MRI, in collaboration w @Philippe_Ciuciu, @demwassermann, @BertrandThirion, and myself 🧠
More info in DM!
My full course on Vector codes and the geometry of mental representations is now on-line on the Collège de France web site:
https://t.co/DgSbbGmVUg
(course in French, but all slides are in English !)
In this thread, I briefly review the topics covered.
Un cours vraiment incroyable pour toute personne francophone qui est intéressée de comprendre la recherche récente sur la façon dont le cerveau représente l’information !
On apprend comment sont représentés les visages, l’espace, et même un peu le langage.
New preprint!🚀🧠
“Information-Restricted Neural Language Models Reveal Different Brain Regions’ Sensitivity to Semantics, Syntax and Context”
Using NLMs, we've developed a new method that reveals brain maps sensitivity to different aspects of language.
https://t.co/Btmj8b2Rwr
[Summary] Information-restricted NLM showed: a left-right asymmetry, varying degrees of sensitivity to syntax and semantics, brain regions integrating long context.
Thus, training models on restricted information can be used to show which brain areas encode this information.[9/9]