My team (i.e., the AI Infra Team) GenBio AI is hiring! If you’re passionate about developing: 1. Large-scale AI systems, apply here: jobs.lever.co/genbio/33e5454… 2. Optimized CUDA Kernels for efficient foundation model computing, apply here: jobs.lever.co/genbio/824c21b… We’re also

Viewing LLMs as systems with latent "skills" and tasks/benchmarks as having "required skills" is a fruitful research perspective inspired by Item Response Theory. The resulting statistical models are interpretable and easy to fit using publicly available LLM evaluation data.

Announcing SGI 2025! Undergrads and MS students: Apply for 6 weeks of paid summer geometry processing research. No experience needed: 1 week tutorials + 5 weeks of projects. Mentors are top researchers in this emerging branch of graphics/computing/math. sgi.mit.edu

I am super excited to announce the call for papers for the New Frontiers in Associative Memories workshop at ICLR 2025. New architectures and algorithms, memory-augmented LLMs, energy-based models, Hopfield networks, associative memory and diffusion, and many other exciting

Great work by Hongli Zhan ✈️ ICML (on the job market)! It started during his summer internship at IBM to improve synthetic data generation with principles/constitutions. Allowing LLMs to first "interpret" principles within each query improves the quality, especially in domains requiring subject experts.

Our new paper with open-source data & models for LLM routing ⬇️

Proudly present MegaMath, the largest open-source math reasoning pretraining corpus—371B tokens of high-quality mathematical web, code, and synthetic data, designed to build the data foundation for next-generation math-proficient LLMs like o1 and R1. 🧵👇 #LLM #OpenSource

How can synthetic data be leveraged for accurate OOD detection? In our work, we use LLMs to generate high-quality OOD proxies, improving detection accuracy and reducing false positives rates, outperforming existing methods across various tasks. ICLR 2026 1/n

The MBZUAI IFM and the LLM360 team's first day at ICLR 2026, come to visit our new Institute of Foundation Models! Booth D04 in Hall 2! We’re looking forward to meeting researchers and engineers to introduce them to MBZUAI .

Congrats Hongli Zhan ✈️ ICML and thanks for your hard work 🎉 context-situated principles is a very promising approach for alignment and other applications 🔥

📢📢 TxT360 has been updated to v1.1: 🌟 BestofWeb: high-quality doc set from the web ❓ QA: Large Scale Synthetic Q&A dataset 📖 Wiki_extended: extended wiki articles via links 🌍 Europarl Aligned: reformatted long aligned corpus huggingface.co/datasets/LLM36… #AIResearch

🤯What we know about RL for reasoning might not hold outside math and code? We revisit established findings on RL for LLM reasoning on six domains (Math, Code, Science, Logic, Simulation, Tabular) and found that previous conclusions drawn on math and code are surprisingly

If you took this class and enjoyed doing it - we are hiring 😃

I'll be at #icml2025 ICML Conference to present SPRI next week! Come by our poster on Tuesday, July 15, 4:30pm, and let’s catch up on LLM alignment! 😃 🚀TL;DR: We introduce Situated-PRInciples (SPRI), a framework that automatically generates input-specific principles to align

Very happy to share that our work "Out-of-Distribution Detection using Synthetic Data Generation" has been accepted at COLM 2025! 🎉 Grateful to have worked with an incredible team Muneeza Azmat, Rå¥å, Mikhail Yurochkin 👏 Conference on Language Modeling #COLM2025

👇Happening this afternoon 4:30pm! Come meet Mikhail Yurochkin, Rå¥å, and I, at East Exhibition Hall #1103. 📍I’m also on the industry job market this coming year! Let’s connect and chat about opportunities in the industry :)

I had a lot of fun working on this project 😃 Training 1000+ LoRAs to do interesting experiments, digging into vLLM, improving the algorithm to work at scale. Thanks Rickard Brüel Gabrielsson for your hard work and congrats 😎

Curious about where human and LLM annotators disagree and how we can close that gap? 🔀🌁🧩 Check out Bridge 🌉, our new statistical framework for human-LLM preference gaps in evaluation. 📄 Paper: arxiv.org/abs/2508.12792 💻 Code: github.com/felipemaiapolo… 🧵1/8