because i haven't done so yet, i've decided to burn any remainder of the bridge to the land of statistics. it wasn't statisticians nor statistics but it was really me. i am simply not good enough to do statistics myself. so, Maxime Peyrard and i decided to turn statistical

🇪🇺 One month after the AI Action Summit 2025 in Paris, I am thrilled to announce EuroBERT, a family of multilingual encoder exhibiting the strongest multilingual performance for task such as retrieval, classification and regression over 15 languages, mathematics and code. ⬇️ 1/6

🚀 Excited to announce EuroBERT: a new multilingual encoder model family for European & global languages! 🌍 🔹 EuroBERT is trained on a massive 5 trillion-token dataset across 15 languages and includes recent architecture advances such as GQA, RoPE & RMSNorm. 🔹

New paper: Language models have “universal” concept representation – but can they capture cultural nuance? 🌏 If someone from Japan asks an LLM what color a pumpkin is, will it correctly say green (as they are in Japan)? Or does cultural nuance require more than just language?

Excited to announce the 1st Workshop on Large Language Models for Cross-Temporal Research at COLM 2025 on Oct 10 in Montreal 🇨🇦 LLMs are hindered in their understanding of time due to temporal biases, conflicting knowledge, and tokenization that fragments dates, leading to

I'm excited to announce that I’ll be joining the Computer Science department at Johns Hopkins University as an Assistant Professor this Fall! I’ll be working on large language models, computational social science, and AI & society—and will be recruiting PhD students. Apply to work with me!

I'm so excited to join the CS department at Johns Hopkins University as an Assistant Professor! I'm looking for students interested in social computing, HCI, and AI—especially around designing better online systems in the age of LLMs. Come work with me! piccardi.me

Should we shut down the field of mechanistic interpretability? arxiv.org/html/2502.2091…

Scaling AI research agents is key to tackling some of the toughest challenges in the field. But what's required to scale effectively? It turns out that simply throwing more compute at the problem isn't enough. We break down an agent into four fundamental components that shape

Coming up at ICML: 🤯Distribution shifts are still a huge challenge in ML. There's already a ton of algorithms to address specific conditions. So what if the challenge was just selecting the right algorithm for the right conditions?🤔🧵

Great case of underspecification: many solutions exist to the ERM learning objective. Key question: what's formally a "good model" (low MDL?) & how to make this the objective. Short of that, we could learn a variety of solutions to examine/select post-hoc: arxiv.org/abs/2207.02598

🚀 Excited to share our latest work at ICML 2025 — zip2zip: Inference-Time Adaptive Vocabularies for Language Models via Token Compression! Sessions: 📅 Fri 18 Jul  - Tokenization Workshop 📅 Sat 19 Jul  - Workshop on Efficient Systems for Foundation Models (Oral 5/145)

Causal Abstraction, the theory behind DAS, tests if a network realizes a given algorithm. We show (w/ Denis Sutter, T. Hofmann, Tiago Pimentel) that the theory collapses without the linear representation hypothesis—a problem we call the non-linear representation dilemma.