Pinned
Check out our new paper 🔥
It’s been so much fun working on this project!
Neural networks do math by rotating shapes.
We found a shape-rotating calculator hidden inside an LLM – and it’s used for more than just math! (1/6)
- I’m excited to share that I’ve started my PhD! continuing my research on interpretability of LLMs. After enjoying my Master’s beyond expectations, I decided my journey wasn’t over yet. It’s a great opportunity to share a few things I’ve learned about research🔍
- 1/13 LLM circuits tell us where the computation happens inside the model—but the computation varies by token position, a key detail often ignored! We propose a method to automatically find position-aware circuits, improving faithfulness while keeping circuits compact. 🧵👇
- Our paper "Position-Aware Circuit Discovery" got accepted to ACL! 🎉 Huge thanks to my collaborators🙏 @OrgadHadas @davidbau @amuuueller @boknilev See you in Vienna! 🇦🇹 #ACL2025 @aclmeeting
- 🚨 Call for Papers is Out! The First Workshop on 𝐀𝐜𝐭𝐢𝐨𝐧𝐚𝐛𝐥𝐞 𝐈𝐧𝐭𝐞𝐫𝐩𝐫𝐞𝐭𝐚𝐛𝐢𝐥𝐢𝐭𝐲 will be held at ICML 2025 in Vancouver! 📅 Submission Deadline: May 9 Follow us >> @ActInterp 🧠Topics of interest include: 👇
- Amazing news: our workshop was accepted to ICML 2025! Interpretability research sheds light on how models work—but too often, those insights don’t translate into actions that improve them. Our workshop aims to challenge the interpretability community to go further.🎉 Our Actionable Interpretability workshop has been accepted to #ICML2025! 🎉 >> Follow @ActInterp @tal_haklay @anja_reu @mariusmosbach @sarahwiegreffe @iftenney @megamor2 Paper submission deadline: May 9th!
- Replying to @tal_haklay2/13 Check out the full paper >> arxiv.org/abs/2502.04577 Project page >> peap-circuits.github.io Or continue in this thread for paper highlights! 🧵👇
- We knew many of you wanted to submit to our Actionable Interpretability workshop, but we didn’t expect to crash Overleaf! 😏🍃 Only 5 days left ⏰! Got a paper accepted to ICML that fits our theme? Submit it to our conference track! 👉 @ActInterp
- We received more submissions to our Actionable Interpretability workshop (@ActInterp) than expected, and we're now looking for additional reviewers! We're seeking reviewers to handle 2–3 papers between May 24 – June 7. Sign up here: forms.gle/FLToWY3keb832n… Thank you! 🙏
- Replying to @tal_haklay(8/8) When I started my Bachelor’s, I never imagined I’d do a Master’s. When I started my Master’s, I never imagined I’d go on to do a PhD. I’m not sure what I’ll do after the PhD. Life is full of surprises 🙏
- Replying to @tal_haklay3/13 What is a circuit? A circuit is a minimal subgraph of a model’s computation graph that executes a specific task. Circuit analysis helps us understand how the model operates and which components (e.g., MLPs, attention heads) are involved.
- Replying to @tal_haklay13/13 This work was done in collaboration with @OrgadHadas, @davidbau , @amuuueller and @boknilev . 💡 Thoughts? Questions? Let’s discuss! Project page >> peap-circuits.github.io Arxiv >> arxiv.org/abs/2502.04577
- Replying to @tal_haklay(6/8) 4. Being a researcher in academia isn’t just about doing research. You’re often also a product manager, a graphic designer, a PR person, a fundraiser (and a bunch of other things I’m definitely not qualified for 😅)
- Replying to @tal_haklay(2/8) 1. Research can put you in very uncomfortable positions. Most of the time you’re not the smartest person in the room. You don’t have all the answers. You make many mistakes. At the beginning of my Master’s that used to paralyze me.
