Berkeley Artificial Intelligence Research

Congratulations to BAIR alumni Aviral Kumar and Jun-Yan Zhu who have been named Samsung AI Researcher of the Year Awardees! https://lnkd.in/gpiVbt6f

Can we make Transformers better and more efficient for robot learning? Excited to introduce Body Transformer (BoT) ??, an architecture that leverages robot embodiment in the attention mechanism, by treating it as a graph of sensors and actuators. Corrective localized actuation is crucial for efficient locomotion and manipulation (e.g., humans use ankles to correct for imbalance at the feet). Robot policies do not typically exploit such spatial interrelations, mostly reusing architectures developed for NLP or computer vision. In practice, we separate observations and actions into a graph of nodes representing the sensors and actuators spread across the robot body. Then, we use masked attention to make sure that, at each layer of the Body Transformer, a node can only attend to itself and its neighbors. Information propagates throughout the graph over different layers. Provided that we have a sufficient number of layers, this simply guides the learning process, without compromising the representation power of the architecture. It makes for a ‘flexible’ but strong inductive bias! BoT surpasses MLP and Transformer baselines on both imitation and reinforcement learning. It shows better generalization and strong scaling properties, as well as potential for much more efficient learning (up to 2x less FLOPs in the attention mechanism). We deployed BoT on a real robot (Unitree A1), sim-to-real, showing its feasibility for real-world deployment! A lot more details in the paper! Work done at Berkeley Artificial Intelligence Research, with absolutely great collaborators Dun-Ming H. Fangchen Liu Jongmin Lee Pieter Abbeel Website: https://lnkd.in/d8a4r2j6 Arxiv: https://lnkd.in/ddfbJjmw Code: https://lnkd.in/dqjVQx4N

Congratulations to BAIR faculty and alumni for their awards at CVPR2024 this week in Seattle. BAIR faculty Angjoo Kanazawa won the Young Researcher Award which "recognizes one or two researchers within seven years of receiving their Ph.D. who have made distinguished research contributions to computer vision" The R-CNN paper from Ross Girshick, Jeff Donahue, Trevor Darrell and Jitendra MALIK received the Longuet-Higgins Prize. "Awarded to a paper that has withstood the test of time, the 2024 Longuet-Higgins Prize recognizes the CVPR paper from 2014 with the most impact."

Looking to hire top AI talent? We've compiled a list of the brilliant Berkeley AI Research Ph.D. Graduates of 2024 who are currently on the academic and industry job markets. Check it out here: https://lnkd.in/g-AzpsxK

CITRIS and the Berkeley Space Center are excited to welcome former BAIR graduate and current NASA - National Aeronautics and Space Administration #Astronaut Woody Hoburg on Feb. 6, at 4 p.m. for a special lecture. Free and open to the public! Bookmark our YouTube channel so you can watch the livestream:?https://lnkd.in/gZVjvjjZ

How can we make LLM agents work together efficiently on complex tasks at a large scale? Introducing LLMCompiler, a tool that compiles an effective plan for executing multiple tasks in parallel. It helps create scalable LLM applications, identifies tasks for parallel execution, and manages dependencies. LLMCompiler is compatible with both open-source and OpenAI models, marking a stride towards more efficient and intelligent software systems. ??Up to 1.3x faster than OpenAI’s recent Parallel Function Calling ?? Up to 9% higher accuracy as compared to ReAct ?? Up to 6x lower cost due to efficient token usage See this twitter post for a quick TLDR: https://lnkd.in/gxs2R3ur Link to Paper: https://lnkd.in/gA-4y8RA Link to Code: https://lnkd.in/gjY6uyiv Joint work with Sehoon Kim Suhong Moon Ryan Tabrizi Nicholas Lee Michael Mahoney Kurt Keutzer

Check out this new language model from Jiantao Jiao's group at BAIR! https://lnkd.in/dUmmfR49