🤔Are your open-source LLMs really safe? 🚨It may be injected with misinformation or bias! Our new paper "𝐂𝐚𝐧 𝐄𝐝𝐢𝐭𝐢𝐧𝐠 𝐋𝐋𝐌𝐬 𝐈𝐧𝐣𝐞𝐜𝐭 𝐇𝐚𝐫𝐦?" (Project website: llm-editing.github.io ) sheds light on the emerging challenges of LLMs, especially the

👉 Check out our new paper on injecting misinformation and bias into LLMs via knowledge editing, as a new type of safety threat: editing threat. 🧐We found that: (1) Editing attach can inject both commonsense and long-tail misinformation into LLMs. (2) Editing attack can

I made a multi-agent system for multimodal retrieval and report generation 🎨 - check it out! 👇 Have talked to a lot of users recently that are interested in using agents to build the final document instead of getting back a chatbot response. There's a general feeling that this

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

🚀 Since its invention, the mouse has been our way to control computers. But what if it didn’t have to be? 🤔 Thrilled to introduce Agent S, a new state-of-the-art GUI agent framework that interacts with computers just like a human and takes on the toughest automation challenges.

I am on job market for full-time industry positions. My research focuses on text generation evaluation and LLM alignment. If you have relevant positions, I’d love to connect! Here are list of my publications and summary of my research:

🧑‍💻 Human software engineers constantly re-evaluate their approaches through experience. 🤖 However, LLM-based software agents can often get stuck in ineffective dead ends. Introducing SWE-Search: a multi-agent framework integrating search and self-refinement to enable software

Everyone talks about scaling inference compute after o1. But how exactly should we do that? We studied compute allocation for sampling -- a basic operation in most LLM meta-generators, and found that optimized allocation can save as much as 128x compute! arxiv.org/abs/2410.22480

I'm on the academic job market this year! My research centers around applying sequential decision-making techniques to build more efficient and reliable real-world AI systems. My work spans theory, methodology, and applications in RL/AI. If you're hiring, I'd love to connect!

The field of “Agents” is expanding rapidly, making it challenging to keep up with the latest developments. We’ve compiled a list of awesome papers in three subareas to help the community: 🌟GUI Agents: github.com/OSU-NLP-Group/… (the most crowded subarea), led by Boyu Gou

I spent the weekend reading some recent great math+reasoning papers: 1. AceMath (arxiv.org/abs/2412.15084) 2. rStar-Math (arxiv.org/pdf/2501.04519) 3. PRIME (arxiv.org/abs/2412.01981) Here are some of my naive thoughts! It could be wrong. All of these papers are showing possible

Large Language Diffusion Models (LLaDA) Proposes a diffusion-based approach that can match or beat leading autoregressive LLMs in many tasks. If true, this could open a new path for large-scale language modeling beyond autoregression. More on the paper: Questioning

🌟Introducing 𝗥𝗲𝗦𝗲𝗮𝗿𝗰𝗵: Learning to Reason with Search for LLMs via Reinforcement Learning. An open-source project that combines 𝗥𝗟 and 𝗥𝗔𝗚 for LLMs! 💡Like Deepseek-R1-Zero and Deep Research, we start with pretrained models and use RL to empower them with the

🚀 Introducing MassGen — our new Multi-Agent Scaling System! Inspired by Grok Heavy & Gemini Deep Think, MassGen enables: 🧠 Parallel processing 🔗 Intelligence sharing 🔁 Iterative refinement ✅ Cross-model consensus (across Google AI, OpenAI, @xAI agents and more) Check it