Check out the recording of the workshop at: iclr.cc/virtual/2025/worksho… 🎥

🚨 New Preprint! Large reasoning models rely on long chain-of-thought generation to solve complex problems, but this comes at a substantial compute cost. We introduce CoDE-Stop, a simple, training-free method that decides when to stop reasoning by tracking how model confidence evolves over time.

Cool suggestions and results from @AndrewLampinen on how semantic plausibility “shortcuts”/“spurious features” can effectively drive the more difficult learning of human language syntax!

Had fun talking at the @SCSLWorkshop at ICLR! One example I brought up, which I think provides an uncommon perspective: a case where spurious shortcuts can improve generalization... even to out-of-distribution sets where the spurious feature doesn't generalize! Thread:

📢 Now live: Catastrophic overtraining of language models @AdtRaghunathan

Looking forward to participating in the #ICLR2025 workshop and the panel. Great speakers!

🌟📢 Live from ICLR 2025! @iclr_conf Our Workshop is underway at Singapore EXPO, Garnet 214-215, featuring presentations and rigorous discussions. Exploring the foundations of robustness and generalization. 🔥 Full schedule here: scslworkshop.github.io/sched… #ICLR2025

💥 See you tomorrow at the Workshop on Spurious Correlation & Shortcut Learning (SCSL) at #ICLR2025! @iclr_conf 🗓️ April 28, 2025 📍 Garnet 214-215, Singapore EXPO 🌐 Schedule and More Info: scslworkshop.github.io #ICLR2025

See you all at the ICLR SCSL workshop today in Garnet 214-215!

🚨 Join us at the Workshop on Spurious Correlation & Shortcut Learning (SCSL) at #ICLR2025! @iclr_conf 🗓️ April 28, 2025 📍 Garnet 214-215, Singapore EXPO 🌐 More info: scslworkshop.github.io #ICLR2025

Also @iclr_conf and it is going to be a busy week! 1. Today at 10am, Optimal Protocols for Continual Learning via Statistical Physics and Control Theory #342 2. Tomorrow at 10am, A Theory of Initialisation's Impact on Specialisation #342 3. Speaking at @SCSLWorkshop o Monday

Excited to head to ICLR this week! Feel free to reach out if you'll be there and would like to chat. I'll be giving a talk at the workshop on spurious correlations and shortcut learning, and we'll have some posters scattered across various other workshops.

🚀 SCSL Workshop at ICLR 2025 (Singapore, April 28) is seeking sponsors! Join us in tackling spurious correlations & shortcut learning with top researchers from NYU, UdeM, UCLA, UMD, SUT, and Google DeepMind. Support cutting-edge AI research & gain visibility! 🔗 Learn more: