Excited to share some of our recent work at @polymathicAI! With AstroCLIP we are bringing multimodal contrastive pretraining to astrophysics!

Very excited to present Multiple Physics Pretraining! MPP is a group of techniques for pretraining on a diverse set of time-dependent physics to learn models that can be fine-tuned for multiple problems. Blog: polymathic-ai.org/blog/mpp/ Paper: arxiv.org/abs/2310.02994

Very happy to start this initiative PolymathicAI with our amazing team members and scientific advisors! The mission: Building #opensource AI models trained across disciplines that deliver scientific discoveries 💪 Site: polymathic-ai.org Github: github.com/PolymathicAI/

Language models are bad a basic math. GPT-4 has right around 0% accuracy rate on 5 digit multiplication. Most open models can't even add. Why is that? There are a few reasons why numbers are hard. The main one is Tokenization. When training a tokenizer from scratch, you take

We are hiring software-focused researchers at PolymathicAI ! Our goal is to develop large foundation models for the sciences. apply.interfolio.com/136060

Excited to be at NeurIPS Conference along with the team PolymathicAI starting Wednesday! Find us 👇 Miles Cranmer Kyunghyun Cho Michael Eickenberg Francois Lanusse, Mariel Pettee Alberto Bietti @leopoldosarra Siavash Golkar Mike McCabe @liamparker Liam Ruben Ohana Bruno Régaldo-Saint Blancard

Today at NeurIPS Machine Learning for Creativity and Design ! Honored to get to share the awesome work of my dancers & musicians in "mememormee", the performance I directed & choreographed earlier this year using AI models trained on my own movements.

Artwork spotlight today Machine Learning for Creativity and Design : "mememormee", an AI-generated dance experiment (April 2023). youtu.be/PRSBaibl4e0?si…

After many years of work, so happy this result is finally public!!

Very excited to share that our team's "Multiple Physics Pre-training" paper won the Best Paper Award at AI for Science workshop at NeurIPS this year! Congrats to the team and to Mike McCabe, Bruno Régaldo-Saint Blancard Liam and Ruben Ohana for leading the effort! 🥳

Very proud of our team :) Miles Cranmer Kyunghyun Cho Michael Eickenberg Francois Lanusse, Mariel Pettee Alberto Bietti @leopoldosarra Siavash Golkar Mike McCabe @liamparker Liam Ruben Ohana Bruno Régaldo-Saint Blancard Keiya Hirashima

Very cool approach to liven up a remote talk by Mariel Pettee Siavash Golkar at #LIPS: Two speakers jointly delivering it! Also very nice results on encoding numbers for LLMs as a basis to build foundation models for physics.

The new-and-improved AstroCLIP has been published with the Monthly Notices of the Royal Astronomical Society! AstroCLIP is the first cross-modal foundation model for galaxies, and competitively performs a wide array of astronomical tasks. Web App: astroclip.streamlit.app

Berkeley Lab's Ben Nachman and Mariel Pettee LBNL Physics Division Berkeley Lab, with colleagues in ATLAS Experiment, have released a new result that uses ML to measure reaction rates in many directions simultaneously –  physicalsciences.lbl.gov/2024/07/01/vie…

Want to build powerful AI for Science? We at PolymathicAI are building foundation models for science 🔥 at Flatiron Institute Join us ( Alberto Bietti Miles Cranmer Kyunghyun Cho Michael Eickenberg Siavash Golkar Mariel Pettee, Francois Lanusse, Mike McCabe, Tom Hehir, Rudy Morel,

Back to the main session of the #IAIFISummerWorkshop with Mariel Pettee and a discussion on LLMs for physics.

Foundation Models for Science Mini-Workshop underway at CERN! h/t Raghav Kansal @grossogaia indico.cern.ch/e/fm4s

We are accepting new internship applications! So many cool & ambitious research topics on this list. (Deadline is in ~2 weeks -- Tuesday, 11/5)

What can exploding stars teach us about flowing blood? PolymathicAI releases AI training data on @HuggingFace for teaching computers to think like scientists: simonsfoundation.org/2024/12/02/new… #science #AI #machinelearning

Starting the last day at #NeurIPS2024 bright and early with the workshop on Machine Learning & the Physical Sciences #MLPS204 Always amazing to be back where it all started, with such an inspiring group of researchers. Mariel Pettee setting the stage with this year’s The Nobel Prize