I’ll be presenting this work today at 10.45 in #NeurIPS2023 👇 Drop by or reach out by DM if you are interested in interpretability of transformer models!

2/3 - Martina Vilas et al. reverse engineer ViTs and demonstrate how the inner representations can be projected onto the class embedding space to uncover how these networks build categorical representations for their predictions. #interpretability #explainability

🚨New round of recruiting! Join our growing team to bridge ML models with interpretable domain expertise. Learn more: neuroexplicit.org Apply by Jan 7

🚀 Exciting News! 📚 Unveiling our latest Paper: "Concept-based Explainable Artificial Intelligence: A Survey" Delve into the world of #Concept-based #XAI with our comprehensive review of concept-based approaches! 🌐 Gabriele Ciravegna Eliana Pastor @ KDD2024 🇪🇸 arxiv.org/abs/2312.12936

Every time I get *yet another* rejection of my work analyzing existing models/datasets (because it "lacks novelty"), I worry that our obsession with novelty in ML will make us repeat the same mistakes, without ever understanding why.

Our community-led ML Theory Group is looking forward to hosting Petar Veličković, Research Scientist at Google DeepMind, next week on Thursday, April 25th for a presentation on "Categorical Deep Learning. An Algebraic Theory of Architectures." Learn more: cohere.com/events/c4ai-Pe…

This Thursday, check out the presentation on "LLM Processes: Numerical Predictive Distributions Conditioned on Natural Language" with James Requeima! Learn more: cohere.com/events/cohere-… Thank you Anier Velasco Sotomayor & Martina Vilas for organizing this talk! 👏

I’ll be presenting this work next week at #ICML👇 Come by or send a DM if you are interested in discussing these topics in Vienna! 📌 Hall C 4-9 #3004 🕜 Tuesday, July 23rd, at 11:30 a.m. - 1:00 pm CEST

August 24th Martina Vilas , will be giving an exciting presentation with our community-led NLP group on "Probing the representations and capacities of Vision-Language Models," be sure to check it out! 🤩 Learn more: cohere.com/events/cohere-…

Applications for Cohere For AI scholars program close tomorrow. Something special about the program is our commitment that a research scientist or engineer will read every application.