🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

Giving a talk on Stat Mech of Transfer Learning at 3pm EST today at the Deepmath conference in Philadelphia. Here's a link for the live stream:

We churn out more & more papers *per scientist*, at an increasing pace, in a rapidly changing publication landscape. Why? How? Want to make sense of this? In a new paper just published in Quantitative Science Studies we dive in & look for answers. direct.mit.edu/qss/article/do…

At NeurIPS Conference until Sunday. Come to my poster Wed, 4:30 PM (neurips.cc/virtual/2024/p…). Also reach out if you want to chat about hierarchical data structures and theory of deep learning!

I just landed in Vancouver to present NeurIPS Conference the findings of our new work! Few-shot learning and fine-tuning change the hidden layers of LLMs in a dramatically different way, even when they perform equally well on multiple-choice question-answering tasks. 🧵1/6

New paper with Leon and Erin Grant! Why do we see localized receptive fields so often, even in models without sparisity regularization? We present a theory in the minimal setting from Alessandro Ingrosso and Sebastian Goldt

Really cool work by Leon, Andrew Saxe and Erin Grant !

I have an opening for a two year postdoc position at Harvard. The application link: academicpositions.harvard.edu/postings/14420

Human-LLM Coevolution: Evidence from Academic Writing arxiv.org/abs/2502.09606 Hint 1: To delve or not to delve, that is the intricate question! Hint 2: A short and easy-to-read paper! Still the word frequency in arXiv abstracts! 👇👇👇

🤖🚀 Announcing 🚀🤖 2025 Princeton ML Theory Summer School August 12-21, 2025 (in person) Lecturers: Krzakala Florent (EPFL) Jianfeng Lu (Duke) Theodor Misiakiewicz (Yale) Yue Lu (Harvard) Yuri Polanskiy (MIT) Apply by March 31 mlschool.princeton.edu

L'Antico Caffè San Marco di Trieste ha appena spernacchiato pubblicamente Cazzullo.

📢 Apply for the #PhD 🎓 in Molecular and Statistical Biophysics at SISSA in lovely Trieste! 4 fully funded fellowships starting Oct 2025. ~8 months of classes, then choose your project. Remote entrance exam. Deadline: 28.4.2025. Please RT! 👉 msb.sissa.it/join-us

The university will not surrender its independence or relinquish its constitutional rights. Neither Harvard nor any other private university can allow itself to be taken over by the federal government. hrvd.me/ResearchFundin…

🚨 Deadline approaching! Apply by May 15 for the #StatPhys29 satellite workshop on Learning and Optimization at Bocconi, Milan. 📍 Free registration – don’t miss out! 🔗 bidsa.unibocconi.eu/learning-and-o…

Our paper on the statistical mechanics of transfer learning is now published in PRL. Franz-Parisi meets Kernel Renormalization in this nice collaboration with friends in Bologna (Federica) and Parma (P. Rodondo, Rosalba Pacelli). journals.aps.org/prl/abstract/1…

Non vedo l'ora di dare un'occhiata dietro le quinte dei "Large language models" come ChatGPT con tutti voi -- ci vediamo lunedì sera all'Hangar di Trieste!

If I had known about this master when I was coming out of my Bachelor, I would have applied in a heartbeat, so please help me spread the word among the budding physicists with an interest in complex systems !

Really happy to see this paper out, led by Nishil Patel in collaboration with Stefano Sarao Mannelli and Andrew Saxe: we apply the statistical physics toolbox to analyse a simple model of reinforcement learning, and find some cool effects, like a speed-accuracy trade-off for generalisation 🚀