publications

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

See the list of publications below, or check my Google Scholar for an up-to-date-list.

2026

  1. ICML
    Inverse Depth Scaling From Most Layers Being Similar
    Yizhou Liu, Sara Kangaslahti, Ziming Liu, and Jeff Gore
    Forthcoming at International Conference on Machine Learning, 2026
    arXiv
  2. ICLR
    Boomerang Distillation Enables Zero-Shot Model Size Interpolation
    Sara Kangaslahti, Nihal V Nayak, Jonathan Geuter, Marco Fumero, Francesco Locatello, and David Alvarez-Melis
    International Conference on Learning Representations, 2026
    arXiv
  3. ICLR
    Hidden Breakthroughs in Language Model Training
    Sara Kangaslahti, Elan Rosenfeld, and Naomi Saphra
    International Conference on Learning Representations, 2026
    arXiv

2025

  1. NeurIPS-W
    Boomerang Distillation Enables Zero-Shot Model Size Interpolation
    Sara Kangaslahti, Nihal V Nayak, Jonathan Geuter, Marco Fumero, Francesco Locatello, and David Alvarez-Melis
    NeurIPS UniReps Workshop Blogpost Track, 2025
  2. TMLR
    Continuous Language Model Interpolation yields Dynamic and Controllable Text Generation
    Sara Kangaslahti and David Alvarez-Melis
    Transactions on Machine Learning Research, 2025
    arXiv

2024

  1. NeurIPS-W
    Continuous Language Model Interpolation for Dynamic and Controllable Text Generation
    Sara Kangaslahti and David Alvarez-Melis
    NeurIPS Workshop on Adaptive Foundation Models, 2024
  2. ICML-W
    Loss in the Crowd: Hidden Breakthroughs in Language Model Training
    Sara Kangaslahti, Elan Rosenfeld, and Naomi Saphra
    Spotlight at the ICML Workshop on Mechanistic Interpretability, 2024
  3. EMNLP
    Fast Forwarding Low-Rank Training
    Adir Rahamim, Naomi Saphra, Sara Kangaslahti, and Yonatan Belinkov
    In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, 2024
    arXiv

2023

  1. NeurIPS-W
    Can you label less by using out-of-domain data? Active & transfer learning with few-shot instructions
    Rafal Kocielnik, Sara Kangaslahti, Shrimai Prabhumoye, Meena Hari, Michael Alvarez, and Anima Anandkumar
    In Transfer Learning for Natural Language Processing Workshop, 2023
    arXiv