That's great to hear, Kosta!

The Conference on #Robot #Learning 2021 will be held on Nov 8-11 in London, UK & virtually. Exciting new changes, including: - both regular and blue sky paper submission types, and - use of #openreview June 18th is the submission date! Find out more: robot-learning.org

Congratulations to Petar Veličković, Taco Cohen Michael Bronstein, and @joanbruna! An insightful and beautifully written manuscript on the 5 G's of geometric DL. It's gggggreat! (sorry)

I love this video demonstrating how learnt motor control can enable team level coordination in a 2v2 MuJoCo humanoid soccer domain (the paper's a great read too :). Congratulations to the whole @DeepMind Embodied Intelligence team! youtube.com/watch?v=KHMwq9…

We're in June already - submissions for the Conference on #Robot #Learning 2021 are due June 18th! Remember you can submit both regular and blue sky paper types, and that the #OpenReview process will be used. Find out more: robot-learning.org #robots #learning #robotics

Stacking blocks may be child's play, but to see a robot do it this well brings a tear to my eye! Enjoy the blogpost, code, paper, and RGB-stacking benchmark and stay tuned - more great research from DeepMind Robotics on the way.

ML/AI/Robotics folks: you didn't even realise it was your birthday, did you? Github: github.com/deepmind/mujoco Stay tuned for open sourcing and feature updates, and congratulations to the DeepMind Robotics Simulation team!

The first hybrid Conference on Robot Learning is underway in London and the World! Looking forward to the first hybrid poster session. #CoRL2021 Conference on Robot Learning

Meet Gopher, a 280B parameter language model from DeepMind! deepmind.com/blog/article/l…

Happy to be a founding editor of this new journal, TMLR: Transactions on Machine Learning Research! Please read all about it in the blog post, and chime in with your questions and feedback on Hugo's thread below.

My first real in-person post-pandemic talk! Hopefully I still remember how..

We're always looking for more general agents, but this one takes the cake - it can play Atari, stack blocks with a real robot arm, and caption images. With the same model. And the same outputs. At the same time. Awesome! 🐈‍⬛😻 Congrats Scott Reed and Nando de Freitas and team!

There are legitimate and scientifically valuable reasons to train a language model on toxic text, but the deployment of GPT-4chan lacks them. AI researchers: please look at this statement and see what you think: forms.gle/ikiYE6ArLpWYz7…

I absolutely enjoyed talking robots and AI with Scott Pelley and 60 minutes, but full credit goes to the @DeepMind Robotics team for the great research that turned these little OP3s into soccer pros!

To everyone who couldn't seem to get the hang of 'DeepMind' and insisted on saying 'Google Deepmind'... don't worry about it. You're good. Seriously, thrilled to join up forces and do amazing things with our Google friends!

💙

Pretty good explanation! I like that it honored his official labeling of the "grabber hands."