Ke Wang

prof_pic.jpg
k.wang@epfl.ch

Hi! 👋

I am a PhD student at at EPFL 🇨🇭 (Swiss Federal Institute of Technology in Lausanne), under the supervision of Prof. Pascal Frossard.

My research interests are centered around model post-training topics such as model merging, model editing, and model fine-tuning. You can check out my research from my Scholar page.

Before starting my PhD, I completed my MSc from EPFL, and my BSc from University of Electronic Science and Technology of China.

During my study, I was also an intern in Oracle Labs and Google Deepmind.

I am from Chengdu, Sichuan in Southwest China (hometown to the pandas 🐼). In my free time, I like football ⚽️, spicy food 🌶️, hiking 🏔️ and skiing 🎿.

selected publications

  1. arxiv-logo.png
    PriFT: Prior-Support Guided Supervised Fine-Tuning
    Ke Wang* , Shuangqi Li* , Mathieu Salzmann , and 1 more author
    In Arxiv , 2026
  2. Neurips-logo.png
    MEMOIR: Lifelong Model Editing with Minimal Overwrite and Informed Retention for LLMs
    Ke Wang* , Yiming Qin* , Nikolaos Dimitriadis , and 2 more authors
    In NeurIPS , 2025
  3. ICLR-logo.svg
    LiNeS: Post-training Layer Scaling Prevents Forgetting and Enhances Model Merging
    Ke Wang* , Nikolaos Dimitriadis* , Alessandro Favero , and 3 more authors
    In International Conference on Learning Representations (ICLR) , 2025
  4. ICML-logo.svg
    Localizing Task Information for Improved Model Merging and Compression
    Ke Wang* , Nikolaos Dimitriadis* , Guillermo Ortiz-Jimenez , and 2 more authors
    In International Conference on Machine Learning (ICML) , 2024
  5. ICML-logo.svg
    Pi-DUAL: Using Privileged Information to Distinguish Clean from Noisy Labels
    Ke Wang , Guillermo Ortiz-Jimenez , Rodolphe Jenatton , and 3 more authors
    In International Conference on Machine Learning (ICML) , 2024