Please call your representatives now on behalf of our international students! I wrote a script here to help you do this: docs.google.com/document/d/1dW…

dall-e 2 illustrations of my friends' twitter bios

#arXiv #machinelearning [csLG] How Useful are Gradients for OOD Detection Really?. (arXiv:2205.10439v1 [cs.LG]) bit.ly/3GdCnxR #mw One critical challenge in deploying highly performant machine learning models in real-life applications is out of distribution (OOD) dete…

The author perception experiment at NeurIPS blog.ml.cmu.edu/2022/11/22/neu… is quite interesting. An amazing disparity between expectation and outcome at a significant scale.

What is the best way to explore in RL? Historically: (1) get info about dynamics -> explores too much! (2) add noise to best policy -> explores too little! We present TIP (just right!) at #NeurIPS22 that plans actions for exploration by getting info on the *optimal trajectory*

I’ve been tinkering away on this lil side project for a while, but thought it’d be fun to build in public, so I present to you—lynkmi—a site to share and curate links!

Today, we're releasing LAB-Bench, a set of >2000 evaluations for language models and agents on scientific research tasks in biology. Public models underperform PhD/postdoc-level humans on nearly all tasks. Claude 3.5 Sonnet is the clear frontrunner atm, but long way to go. 1/

🖤 Thanks for the paper award, ICML2024 MFM-EAI

Introducing PaperQA2, the first AI agent that conducts entire scientific literature reviews on its own. PaperQA2 is also the first agent to beat PhD and Postdoc-level biology researchers on multiple literature research tasks, as measured both by accuracy on objective benchmarks

We just dropped Wikipedia-style articles for all ~20,000 human protein-coding genes. These were made using PaperQA2, and are more accurate on average than actual human-written articles on Wikipedia. Enjoy, and let us know if you have requests for other articles we could write...

new overview paper on "Scientific Progress in Artificial Intelligence: History, Status and Futures," written with my co-author Eric Horvitz . cs.cmu.edu/~tom/pubs/AI_O…

🎇 I’m on the academic job market! I’m a PhD candidate at Machine Learning Dept. at Carnegie Mellon. My research tackles challenges that arise from the sequential nature of human-AI interaction. Toward this goal, my work involves: 🤖 reinforcement learning, 🧠 foundation models, and 👩‍💻 human-centered AI.

✨ Introducing: SHOW NOTES ✨ Ever been at a great party where a few people mentioned some really cool links—but you're just not all in one group chat? Those links probably got lost—but no more! With Show Notes you can share a semi-private link via QR code for anyone to add to!

Andrej Karpathy Here's one for agents doing scientific tasks github.com/Future-House/a…

Watch our team explain how you can use the FutureHouse Platform to come up with new hypotheses and make new discoveries.

Another life update!! 🎉 I’m joining JHU Computer Science as an Assistant Professor starting Fall 2026! Apply to work with me on reinforcement learning, foundation models, & human-centered AI. Let’s build better AI agents 🤖🙆‍♀️🦀 Before that, I’ll join NYU Courant as an Assistant

It was a dream come true to teach the course I wish existed at the start of my PhD. We built up the algorithmic foundations of modern-day RL, imitation learning, and RLHF, going deeper than the usual "grab bag of tricks". All 25 lectures + 150 pages of notes are now public! 🧵