Publications

2025

  1. mib.png
    MIB: A Mechanistic Interpretability Benchmark
    Aaron Mueller, Atticus Geiger, Sarah Wiegreffe, Dana Arad, Iván Arcuschin, Adam Belfki, Yik Siu Chan, Jaden Fiotto-Kaufman, Tal Haklay, Michael Hanna, and 13 more authors
    In ICML, 2025
    Website
  2. arithmetic_boh.png
    Arithmetic Without Algorithms: Language Models Solve Math With a Bag of Heuristics
    Yaniv Nikankin, Anja Reusch, Aaron Mueller, and Yonatan Belinkov
    In ICLR, 2025
    Code Website

2023

  1. deconstructing.png
    Deconstructing Data Reconstruction: Multiclass, Weight Decay and General Losses
    Gon Buzaglo, Niv Haim, Gilad Yehudai, Gal Vardi, Yakir Oz, Yaniv Nikankin, and Michal Irani
    In NeurIPS, 2023
  2. sinfusion.gif
    SinFusion: Training Diffusion Models on a Single Image or Video
    Yaniv Nikankin, Niv Haim, and Michal Irani
    In ICML, 2023
    Code Website