@misc{indiciae36213395cd76,
  title = {Embedding Shift Dissection on CLIP: Effects of Augmentations on VLM's Representation Learning},
  author = {Ashim Dahal and Saydul Akbar Murad and Nick Rahimi},
  year = {2025},
  doi = {10.1109/cvprw67362.2025.00469},
  url = {https://arxiv.org/abs/2503.23495},
  note = {Source identifier: 2503.23495}
}