Skip to content

v1.1.0

Compare
Choose a tag to compare
@ZexinHe ZexinHe released this 04 Mar 16:51
· 10 commits to main since this release

Full Changelog: v1.0.0...v1.1.0

  • Improved code structure for enhanced usability and extensibility.
  • For detailed information on the new model architectures, please refer to the updated model card.
    • The image encoder now utilizes DINOv2 with register tokens by default.
    • The small, base, and large model sizes have been redefined to provide clearer distinctions.
    • Updated training settings are documented in the model card.
  • Version 1.1 of the model weights has been released with the following updates.
    • Release model weights trained only on Objaverse with the refactored codebase.
    • Release new model weights trained on both Objaverse and MVImgNet.
  • Huggingface demo defaults to openlrm-mix-base-1.1 model.