🔥Telegram İfşa

👀At this year's ECML PKDD'2023, Timo Kaufmann @ ICLR 2025 and Sarah Ball gave a talk on the Challenges and Practices of Reinforcement Learning from Real Human Feedback #HLDM. 📷Check out the presentation on our YouTube channel youtu.be/6BxYMbsJavg!

Very cool work! I love the focus on control tasks over LLMs, the creative use of AI feedback and NetHack as a benchmark.

Really happy with this collaboration. Great fun in the process, satisfying outcome. Give it a read!

🚀 We’re excited to present at the MHFAIA@ICML2024 Workshop at ICML Conference! Dive into our novel approach to human feedback in reinforcement learning based on distinguishability queries. #RLHF #ICML2024 🧵1/6

I'm at #ICML2024, reach out if you want to chat!

Excited to attend the first RLC, the @RLBRew_2024 workshop in particular, and the beyond rewarding post-workshop beers in very particular! 🍺

I'll be presenting this at @RLBRew_2024 this morning, join us at the workshop!

Had a good time at the @RLBRew_2024 workshop at RLC yesterday, thanks to the organizers!

Learning about the history of RL at the birthplace of RL from the founders of RL at the first RL_Conference! What a privilege!

I love how green the #RLC2024 venue (UMass Amherst) is

🚀 Thrilled that our work on Inverse Constitutional AI has been accepted at ICLR 2025! We are continuing to extend the implementation, just released the latest version (v0.1.2) at github.com/rdnfn/icai

Our paper on query-efficient reward learning will be at AAAI! Unfortunately I won’t be attending, but Xuening will be at the poster - stop by or reach out online!

🕵🏻💬 Introducing Feedback Forensics: a new tool to investigate pairwise preference data. Feedback data is notoriously difficult to interpret and has many known issues – our app aims to help! Try it at app.feedbackforensics.com Three example use-cases 👇🧵

Ever wondered how models on chatbot arena differ? Feedback forensics gives some answers, check it out!

Currently visiting Arduin Findeis in Cambridge. I didn't realize it's this beautiful! Do I know anyone here that I haven't met up with yet?

Looking forward to present tomorrow. Come by our poster #520 if you're at ICLR!

This was a lot of fun!

Just noticed the key deadlines for #ICLR2026 out! PSA for everyone else who's been waiting. Full paper: Sept 24 AoE.

Nice demo of feedback forensics, check out our tool and Arduin's analysis!