AutoRL Workshop

We have a speaker change: instead of @jparkerholder we'll hear from @MichaelD1729 - the focus of the talk is the same, though, so join if you're interested in generating any environments you can imagine!

After a day of presentations, posters and a breakout session, we'll close with a panel discussion. @pcastr @AlexDGoldie @jakeABeck and Doina Precup will tell us their views on the present and future of AutoRL - join us for an exciting finale to @icmlconf 2024! 👑

Last but not least: Pablo Samuel Castro @pcastr is a senior researcher at @GoogleDeepMind known for his musical endeavors and pushing the limits of the ALE with algorithmic innovations and design decisions. He’ll talk about why the ALE is a great benchmark for AutoRL 👾🕹️

Up next: Jack Parker-Holder @jparkerholder works on open-endedness as a research scientist at @GoogleDeepMind and honorary lecturer with @UCL_DARK . His focus is on unlimited training data for open-ended RL - being able to generate interactive tasks controllably and at will 🔮

Our third speaker is Pierluca D'Oro @proceduralia, researcher at @AIatMeta and PhD student at @Mila_Quebec. You likely know his work combining RL with LLMs to create capable AI assistants. We’re excited to hear how he envisions the future of LLMs in RL and vice versa! 🗣️🧠

Speaker number two! 🔥 Roberta Raileanu @robertarail, research scientist at @AIatMeta, has worked on different aspects of RL, most recently teaching LLMs to make better decisions. She’ll discuss generalization for robust and capable agents in practice 💪

Chelsea Finn @chelseabfinn barely needs an introduction: If you’re interested in robotics or meta-learning, you almost certainly know her work. She’s an assistant professor at @Stanford, co-founder of @physical_int and expert in making RL work for real-world robotics tasks 🤖💫

ICML in Vienna is right around the corner! ☕️🍰 As our own countdown, we’re introducing one of our invited speakers each day this week. First up:

Stop having conferences in the US.

In search of some Friday entertainment? 🎥 Some of our accepted papers' summary videos are already online for you to enjoy 🍿🍹 youtube.com/playlist?list=…

Please send us your CV and a short statement as to why you should receive the complimentary registration to autorlworkshop@ai.uni-hannover.de by 8 July AOE in case you want one.

If you are attending the AutoRL workshop, we are offering two complimentary registrations to persons looking for financial support (you will be reimbursed in case you have already registered).

All of these should be on your reading list and we're looking forward to hearing more from the authors in Vienna next month! 🎉

"Self-Exploring Language Models: Active Preference Elicitation for Online Alignment" adds optimism to the RLHF objective for better out-of-distribution sampling. By @ShenaoZhang @Yudh9662, Hiteshi Sharma, @yzy_ai @shuohangw @hany_hassan & @zhaoran_wang openreview.net/forum?id=W4vHm…

Our accepted papers are now available on our website - as are our paper spotlights, three excellent works we think deserve extra recognition: ⭐️🏆

It has been great seeing all the interesting submissions to the workshop - acceptance notifications will be posted next week. Stay tuned for the public list of papers, there are some excellent contributions from all corners of AutoRL we're excited to talk about in Vienna! 🚀

Are you as exhausted as us from the NeurIPS deadline? If yes, we have good news: we extended the submission deadline for the workshop by a week, so you now have until the 31.05. AOE 🎉 Submit here: #tab-your-consoles" target="_blank" rel="nofollow noopener">openreview.net/group?id=ICML.…

Reminder: our submission deadline is coming up on May 23rd (AOE) - that's only three more days to go! 🕙💨 CfP: #cfp" target="_blank" rel="nofollow noopener">autorlworkshop.github.io/#cfp