Two ongoing works presented at MIRU 2025

Two ongoing works were presented as posters at MIRU 2025¹:

• Design Choices for Labeling Sparse Autoencoder Concepts with Vision Language Models

  Patrick Ramos, Yuta Nakashima, Noa Garcia

• Pre-training Bias Does Not Transfer in CLIP

  Ryan Ramos, Yuta Nakashima, Noa Garcia

---