Dahye Kim

Contact. dahye [at] bu [dot] edu

prof_pic.jpg

665 Commonwealth Ave.

Boston, MA 02215

Hi, thanks for stopping by! šŸ‘‹

I am a third-year CS PhD student at Boston University, where I am fortunate to be advised by Prof. Deepti Ghadiyaram. Prior to joining BU, I completed my Masterā€™s degree at Yonsei University under the guidance of Prof. Kwanghoon Sohn.

Iā€™m interested in understanding how generative models work and using those insights to make them more useful, efficient, and controllable.

Iā€™m always open to collaborations and research discussions. Feel free to reach out!

news

Aug 31, 2026 Excited to join Adobe as a Research Scientist Intern in Seattle! šŸŽØ
Apr 09, 2026 DDiT is selected as a Highlight at CVPR 2026! šŸŒŸ
Feb 21, 2026 DDiT is accepted at CVPR 2026! šŸžļø
Jun 30, 2025 I joined Amazon as an Applied Scientist Intern in Seattle! šŸ™ļø
Jun 25, 2025 Revelio is accepted at ICCV 2025! šŸŒŗ
May 03, 2025 Revelio is selected for an oral presentation at the CVPR 2025 MIV Workshop ! šŸŽø
Feb 03, 2025 New preprint is out: check out Concept Steerers šŸŖ„
Nov 26, 2024 New preprint is out: check out Revelio šŸ”Ž
Sep 01, 2024 šŸš€ I started my PhD journey at Boston University!

publications

  1. swift_sampling.png
    Swift Sampling: Selecting Temporal Surprises via Taylor Series
    Dahye Kim,Ā Bhuvan Sachdeva*,Ā Karan Uppal*,Ā Naman Gupta*,Ā Vineeth N. Balasubramanian,Ā andĀ Deepti Ghadiyaram
    arXiv preprint arXiv:2605.22678, 2026
    arXiv Code Website
  2. ddit.png
    DDiT: Dynamic Patch Scheduling for Efficient Diffusion Transformers
    Dahye Kim,Ā Deepti Ghadiyaram,Ā andĀ Raghudeep Gadde
    CVPR, 2026 (Highlight)
    arXiv Website
  3. steerers.png
    Concept Steerers: Leveraging K-Sparse Autoencoders for Controllable Generations
    Dahye Kim,Ā andĀ Deepti Ghadiyaram
    arXiv preprint arXiv:2501.19066, 2025
    arXiv Code Website
  4. revelio.png
    Revelio: Interpreting and leveraging semantic information in diffusion models
    Dahye Kim*,Ā Xavier Thomas*,Ā andĀ Deepti Ghadiyaram
    ICCV, 2025
    CVPRW MIV, 2025 (Oral)
    arXiv Code Poster Website
  5. kim2023language.png
    Language-free training for zero-shot video grounding
    Dahye Kim,Ā Jungin Park,Ā Jiyoung Lee ,Ā Seongheon Park,Ā andĀ Kwanghoon Sohn
    WACV, 2023
    arXiv PDF
  6. park2023normality.png
    Normality guided multiple instance learning for weakly supervised video anomaly detection
    Seongheon Park,Ā Hanjae Kim ,Ā Minsu Kim ,Ā Dahye Kim,Ā andĀ Kwanghoon Sohn
    WACV, 2023
    PDF