our discovery featured on SPACE.com

Mysterious Explosion in a Neighboring Galaxy Astronomers have recorded a mysterious X-ray flare in the Large Magellanic Cloud. This is the closest galaxy to us. The burst was designated XRT 200515. It was detected using the Chandra Observatory and turned out to be the brightest

The biggest-ever AI model for biology, Evo 2, out now on bioRxiv 🧬🤖 biorxiv.org/content/10.110…

Fun to see Evo 2 in Andrej Karpathy's new practical guide for using LLMs :)

Student Steven Dillmann wrote a nice Chandra Observatory blog post about his discovery: an unexpected transient in the Large Magellanic Cloud, using representation learning: youtube.com/watch?v=3zEKjE… chandra.harvard.edu/blog/node/919

Andrej Karpathy's advice for ambitious college students:

Awesome to see our Evo 2 model highlighted in Jensen's keynote at GTC Arc Institute 🤝 NVIDIA

Fellow academics, please consider signing. I did. sites.google.com/view/we-must-l…

Officially graduated from the University of Cambridge with an MPhil in Data Intensive Science.👨‍🎓🎓 Truly a humbling experience, standing in the Senate House and reflecting on the legacy of scientific discovery that echoes through this place. Cambridge University St John's College, Cambridge

🚀 Introducing CAVA: The Comprehensive Assessment for Voice Assistants A new benchmark for evaluating end-to-end, speech-in-speech-out voice assistants in real-world scenarios. We go beyond single tasks or metrics to test the capabilities required for voice assistants:

40% with just 1 try per task: SWE-agent-LM-32B is the new #1 open source model on SWE-bench Verified. We built it by synthesizing a ton of agentic training data from 100+ Python repos. Today we’re open-sourcing the toolkit that made it happen: SWE-smith.

I’ll be at the YC AI Summer School in San Francisco this week. Reach out if you’re here too and want to chat! 🚀 Thank you Y Combinator, Garry Tan, Nicolas Dessaigne and others for the invite!

Great to see such a strong representation of “AI-for-Science” at the Y Combinator AI Summer School from Nobel Prize winner John Jumper, but also Sam Altman from OpenAI, Jared Kaplan from Anthropic and more talking about AI to drive scientific discovery! Garry Tan #ai4science

More than 2,000 new asteroids ☄️ See how NSF-DOE Rubin Observatory has discovered thousands of asteroids over just a few nights of data collection, and why it is expected to discover 5 million more: rubinobservatory.org/news/rubin-fir…

Our study led by CLS reveals an “ideation–execution gap” 😲 Ideas from LLMs may sound novel, but when experts spend 100+ hrs executing them, they flop: 💥 👉 human‑generated ideas outperform on novelty, excitement, effectiveness & overall quality!

check out this amazing paper by CLS @COLM2025 ⬇️

Just arrived in Vancouver to present our work with Rafael Martínez Galarza on "Learning Representations of Event Time Series with Sparse Autoencoders" at the ICML 2025 ML4Astro Workshop 🇨🇦🍁 Paper: arxiv.org/abs/2507.11620 Workshop: ml4astro.github.io/icml2025/ Reach out if you want to chat!

had a great time at the IAIFI Summer School at Harvard University this week