Gabriele Sarti

gsarti.com

01 July 2026 13:07

🚨New paper led by @danielsc4.it & @saracandussio.bsky.social studying answer commitment across CoT for math & logic reasoning! Turns out, models often converge to final answers early in the CoT, then tend to "fake" hedging & re-checks - and we can tell apart mid/final guesses quite robustly! See 👇

Sara Candussio

saracandussio.bsky.social

01 July 2026 13:07

Are all the CoT steps necessary? In our latest paper, we find evidence for the existence of a commitment boundary, marking a sharp transition from no/mid guesses to the model final answer across various reasoning tasks and model families. Thread 🧵👇

BlackboxNLP

blackboxnlp.bsky.social

01 July 2026 12:40

🏆 Announcing the NDIF Best Paper Award for the BlackboxNLP 2026 Reproducibility Challenge! @ndif-team.bsky.social Reproduce an interp. finding with nnsight + NDIF, open-source it, and push it further. Winners will receive a $500 prize, and will be invited to present at the workshop!

Gabriele Sarti

gsarti.com

04 January 2026 18:08

Our work on contrastive SAE steering for personalizing literary machine translation was accepted to EACL main! 🎉 Check it out! ⬇️

Jaap Jumelet

jumelet.bsky.social

16 October 2025 11:23

🌍Introducing BabyBabelLM: A Multilingual Benchmark of Developmentally Plausible Training Data! LLMs learn from vastly more data than humans ever experience. BabyLM challenges this paradigm by focusing on developmentally plausible data We extend this effort to 45 new languages!