publications

publications by categories in reversed chronological order. generated by jekyll-scholar.

2025

  1. ICLR
    SAE.png
    Rethinking Evaluation of Sparse Autoencoders through the Representation of Polysemous Words
    Gouki Minegishi, Hiroki Furuta, Yusuke Iwasawa, and Yutaka Matsuo
    In The Thirteenth International Conference on Learning Representations, ICLR, 2025
  2. TMLR
    Grokking.png
    Bridging Lottery Ticket and Grokking: Understanding Grokking from Inner Structure of Networks
    Gouki Minegishi, Yusuke Iwasawa, and Yutaka Matsuo
    Transactions on Machine Learning Research, TMLR, 2025
  3. ICLR WS
    In-Context Meta Learning Induces Multi-Phase Circuit Emergence
    Gouki Minegishi, Hiroki Furuta, Shohei Taniguchi, Yusuke Iwasawa, and Yutaka Matsuo
    In ICLR 2025 Workshop on Building Trust in Language Models and Applications, 2025
  4. ICML
    ICL.gif
    Beyond Induction Heads: In-Context Meta Learning Induces Multi-Phase Circuit Emergence
    Gouki Minegishi, Hiroki Furuta, Shohei Taniguchi, Yusuke Iwasawa, and Yutaka Matsuo
    In Forty-second International Conference on Machine Learning, ICML, 2025
  5. Preprint
    Reasoning_Graph.gif
    Topology of Reasoning: Understanding Large Reasoning Models through Reasoning Graph Properties
    Gouki Minegishi, Hiroki Furuta, Takeshi Kojima, Yusuke Iwasawa, and Yutaka Matsuo
    2025

2024

  1. ICLR WS
    Interpreting Grokked Transformers in Complex Modular Arithmetic
    Hiroki Furuta, Minegishi Gouki, Yusuke Iwasawa, and Yutaka Matsuo
    In ICLR 2024 Workshop on Bridging the Gap Between Practice and Theory in Deep Learning, 2024
  2. ICLR WS
    Bridging Lottery ticket and Grokking: Is Weight Norm Sufficient to Explain Delayed Generalization?
    Minegishi Gouki, Yusuke Iwasawa, and Yutaka Matsuo
    In ICLR 2024 Workshop on Bridging the Gap Between Practice and Theory in Deep Learning, 2024
  3. Neurips
    ADOPT: Modified Adam Can Converge with Any {}beta_2 with the Optimal Rate
    Shohei Taniguchi, Keno Harada, Gouki Minegishi, Yuta Oshima, Seong Cheol Jeong, and 5 more authors
    In The Thirty-eighth Annual Conference on Neural Information Processing Systems, Neurips, 2024
  4. TMLR
    Towards Empirical Interpretation of Internal Circuits and Properties in Grokked Transformers on Modular Polynomials
    Hiroki Furuta, Gouki Minegishi, Yusuke Iwasawa, and Yutaka Matsuo
    Transactions on Machine Learning Research, TMLR, 2024