🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

Very promising results!

🚀 Introducing RAGEN—the world’s first reproduction of DeepSeek-R1(-Zero) methods for training agentic AI models! We’re betting big on the future of RL + LLM + Agents 🤖✨. This release is a minimally viable leap toward that vision. Code and more intro 🔗:

Super excited to be part of the project! Betting big on the future of RL + LM + Agent!

Check this nice benchmark! The first most comprehensive benchmark evaluates MLLMs!

🚀 New model results on EmbodiedBench! 🚀 🔹 Qwen2.5 VL surpasses Qwen2 VL as embodied agents! 🔹 InternVL2_5 MPO leads as the best-performing open-source model! Check out the latest results: embodiedbench.github.io Explore the evaluation code: github.com/EmbodiedBench/…

Very happy to see the search agent Search-R1 and our RAGEN codebase has been able to support it😄 We put a lot of efforts to make the RAGEN codebase easy to reuse/follow. Welcome to play with RAGEN for agent framework using simple RL recipe like DeepSeek R1

🚀 Introducing Chain-of-Experts (CoE), A Free-lunch optimization method for DeepSeek-like MoE models! within $200, we explore to train MoEs that enables 17.6-42% efficiency boost in memory! Code: github.com/ZihanWang314/c… Blog: notion.so/Chain-of-Exper…

In the last two months, RAGEN has powered Agent RL training frameworks for over 300,000 people. Now, we’re introducing VAGEN—the first open-source framework that trains *Visual* Agents using multi-turn Reinforcement Learning! 🚀(1/n)

Training agentic settings with LLM +RL is truly a fascinating topic, and now we make a solid step towards *Multimodal* scenarios! With visual inputs, we now see how agents think and interact into an environment when information sources are diverse 😆

No visual models one can survive the challenge, but... ... ... ... ... Our VAGEN can we are doing small progress but visual agent has yet even more to do

We are embarrassed to say that VAGEN is the No. 1 visual agent framework, but... it's true X Post: x.com/wzihanw/status… Blog: mll-lab.notion.site/vagen Code: github.com/RAGEN-AI/VAGEN

Thrilled to release our RAGEN paper! We had some interesting findings, and you can check them out here ⬇️

We are very excited announcing our MLL lab! We are looking for collaborators on RAGEN, VAGEN, Chain-of-experts, T*, LongVideoHaystack, foundation models for embodied agents, etc mll-lab-nu.github.io

Big news! Our CVPR 2025 Workshop on Foundation Models + Embodied Agents has extended the non-archival deadline to May 17! If you’re exploring the intersection of foundation models + embodiment, we’d love to see your work. Catch us (and our tutorial) in Music City!

Today is the day! Welcome to join #CVPR2025 workshop on Foundation Models meet Embodied Agents! 🗓️Jun 11 📍Room 214 🌐…models-meet-embodied-agents.github.io/cvpr2025/ Looking forward to learning insights from wonderful speakers Jitendra MALIK Ranjay Krishna Katerina Fragkiadaki Shuang Li Yilun Du

Excited to be leading this project! Humans naturally build spatial mental models to understand object permanence and reason about space. We find that VLMs can effectively approximate this by first forming an overall scene understanding, then reasoning over it. More to explore!

MindCube is the result of a joint effort of Northwestern Engineering Stanford AI Lab Stanford HAI Stanford Vision and Learning Lab NYU Courant Allen School. Huge thanks to project leader Qineng Wang Baiqiao Yin and our incredible team Fei-Fei Li Jiajun Wu Saining Xie Ranjay Krishna @HanLiu WilliamZhang

Please check our awesome project team here led by Prof. Manling Li ! If you have any thought or idea about spatial intelligence, feel free to contact us😆

Congratulations to my amazing advisor Manling Li !!!

Excited to do a talk at Agentic AI Summit! Session 4: Foundations of Agents 📍 Frontier Stage 📅 2:45pm PT Will talk about "RAGEN: Training Agents via Reinforcing Reasoning", including: - How to monitor RL training to converge for multi-turn LLM Agents? RAGEN