RoboGen: Towards Unleashing Infinite Data for Automated Robot Learning via Generative Simulation paper page: huggingface.co/papers/2311.01… present RoboGen, a generative robotic agent that automatically learns diverse robotic skills at scale via generative simulation. RoboGen

🔥Generate and solve robot simulation tasks with multi-modal LLMs! 🤖Introduce our work to be presented at #CoRL2024 in Munich, Germany: GenSim2: Scaling Robot Data Generation with Multi-modal and Reasoning LLMs! It "amplifies" simulation tasks and data with foundation models.

Giving a plenary speech at the Chinese pattern recognition and computer vision conference prcv.cn in Urumqi this weekend. The publisher also organised a book-signing event for the Chinese version of my new book.

【IDS PanDA-TAs had fun - Thank you for joining us today @ HKU Info Day 2024🐼】 Hope you had enjoyed our activities, talks, and exclusive souvenirs prepared for you. Can’t wait to meet you at other initiatives! 🥳🎁🙂‍↕️🤖 #HKU #hkudatascience #idsonhkuinfoday2024

【Grand Opening of IDS New Premises on Oct 29 – Share the Joy with Us for Its Promising Future Dev't amidst Its 3rd Birthday!🎉🙌🏻】 Let’s revisit the memorable day with our team - hope to see your faces at different research collaborative opportunities😉 #HKU #hkudatascience

Everything you love about generative models — now powered by real physics! Announcing the Genesis project — after a 24-month large-scale research collaboration involving over 20 research labs — a generative physics engine able to generate 4D dynamical worlds powered by a physics

If the robotics community lacks a simulator enough to convince everyone to use it by now, then from now on, we will reach a consensus that Genesis will be the best one.

In our initial release, we claimed that Genesis is over an order of magnitude faster than existing GPU-accelerated robotic simulators. However, our reported benchmarking results only included three relatively simple and static scenes, which were not comprehensive enough to fully

[1/n] 🧐DeepSeek #DeepSeekR1 has shown the power of RL without SFT. But what does RL learns differently than SFT? Our answer is: 📉SFT Memorizes, RL Generalizes.📈 tianzhechu.com/SFTvsRL/

I'm at ICLR this week! I'll be presenting ToST, a (provably) computationally efficient high-performance deep architecture derived from information theory and convex analysis principles. 📅 Saturday April 26, 10AM-12:30PM 📌 Hall 3 + Hall 2B #145 💡Awarded a Spotlight! (1/3)

At ICLR! Welcome to chat with me for dinner and discuss the latest work.

Our latest book on the mathematical principles of deep learning and intelligence has been released publicly at: ma-lab-berkeley.github.io/deep-represent… It also comes with a customized Chatbot that helps readers study and a Chinese version translated mainly by AI. This is an open-source project.

It's a great honor to be the TA of this course.