Carta Thomas

Finally, a big thank you to the jury members for their time, questions, and insights: 👨‍🏫 Jean Ponce (president) 🧠 Reviewers @pierrelux & @GMartius 💡 @edwardfhughes & @white_martha for their excellent questions and engagement alongside my supervisors.

And to my brilliant co-authors — @ClementRomac, @LorisGaven, @cedcolas, @GuillaumeAP, @GuillaumelLevy, and @KovacGrgur — your ideas, energy, and discussions made this adventure unforgettable. Couldn’t have done it without you all. 🙏✨

After four incredible years in the Flowers AI & CogSci Lab, I’m thrilled to share that I’ve officially become a Doctor! 🎓💫 My thesis: “Language as a Cognitive Tool for Open Agents.” It explores how language helps artificial agents explore, adapt, and learn efficiently.

Phase 1 of Physics of Language Models code release ✅our Part 3.1 + 4.1 = all you need to pretrain strong 8B base model in 42k GPU-hours ✅Canon layers = strong, scalable gains ✅Real open-source (data/train/weights) ✅Apache 2.0 license (commercial ok!) 🔗github.com/facebookresear…

The automation of innovation is within reach! Delighted that my @raais talk is now available for anyone to watch, alongside an excellent blogpost summary by the inimitable @nathanbenaich.

Human ideation beats AI ideation when measured on execution outcomes: arxiv.org/abs/2506.20803. There's a clear path to fixing this - work to be done!

🔔 Join our MAGELLAN talk on July 2! We'll explore how LLM agents can monitor their own learning progress and choose what to learn next, like curious humans 🤔 1h presentation + 1h Q&A on autotelic agents & more! 📅 July 2, 4:30 PM CEST 🎟️ forms.gle/1PC2fxJx1PZYfq…

📰 Check out the full paper here: arxiv.org/abs/2506.06725 Don't hesitate to reach out if you want to discuss WorldLLMs or RL for grounding LLMs! Or come see us at the poster session on Thursday!

I'll be at RLDM this week to present our new paper: WorldLLM 😃 In the same spirit as our previous works — e.g. GLAM, MAGELLAN... — which investigate how to ground LLMs through interactions with external environments, WorldLLM takes a deep dive into the world modeling aspect.

New Paper: Continuous Thought Machines 🧠 Neurons in brains use timing and synchronization in the way that they compute, but this is largely ignored in modern neural nets. We believe neural timing is key for the flexibility and adaptability of biological intelligence. We propose a new neural architecture, “Continuous Thought Machines” (CTMs), which is built from the ground up to use neural dynamics as a core representation for intelligence. By using neural dynamics as a first-class representational citizen, CTMs naturally perform adaptive computation. Many emergent, interesting behaviors arise as a result: CTMs solve mazes by observing a raw maze image and producing step-by-step instructions directly from its neural dynamics. When tasked with image recognition, the CTM naturally takes multiple steps to examine different parts of the image before making its decision. This step-by-step approach not only makes its behavior more interpretable but also improves accuracy: the longer it “thinks,” the more accurate its answers become. We also found that this allows the CTM to decide to spend less time thinking on simpler images, thus saving energy. When identifying a gorilla, for example, the CTM’s attention moves from eyes to nose to mouth in a pattern remarkably similar to human visual attention. I think this work underscores an important, yet often lost, synergy between neuroscience and AI. While modern AI is ostensibly brain-inspired, the two fields often operate in surprising isolation. By starting with such inspiration and iteratively following the emergent, interesting behaviors, we developed a model with unexpected capabilities, such as its surprisingly strong calibration in classification tasks, a feature that was not explicitly designed for. When we initially asked, “why do this research?”, we hoped the journey of the CTM would provide compelling answers. By embracing light biological inspiration and pursuing the novel behaviors observed, we have arrived at a model with emergent capabilities that exceeded our initial designs. We are committed to continuing this exploration, borrowing further concepts to discover what new and exciting behaviors will emerge, pushing the boundaries of what AI can achieve.

Checkout Generative Hierarchical Materials Search (GenMS) – a framework for generating crystal structures from natural language. Website: generative-materials.github.io Paper: arxiv.org/abs/2409.06762