1 dépôt
Awesome GitHub RepositoriesHypersphere Embedding Regularization
Loss functions that constrain feature embeddings to a uniform distribution on a hypersphere to prevent representation collapse.
Distinct from Regularization Techniques: Distinct from general regularization: focuses specifically on embedding uniformity on hyperspheres during training.
Explore 1 awesome GitHub repository matching artificial intelligence & ml · Hypersphere Embedding Regularization. Refine with filters or upvote what's useful.
Awesome Hypersphere Embedding Regularization GitHub Repositories
Trouvez les meilleurs dépôts grâce à l'IA.Nous recherchons les dépôts les plus pertinents grâce à l'IA.
- facebookresearch/dinov2
facebookresearch/dinov2
12,987Voir sur GitHubDINOv2 is a self-supervised vision transformer foundation model designed to generate high-quality visual representations from raw image data. By leveraging large-scale unlabelled datasets, the framework learns to extract robust numerical embeddings that serve as inputs for various machine learning and analysis workflows. The model distinguishes itself through a teacher-student training framework that utilizes centered and sharpened soft probability distributions to align feature maps across multiple image crops. It incorporates a masking strategy that forces the model to reconstruct missing i
Applies regularization to encourage uniform distribution of feature embeddings and prevent representation collapse.
Jupyter Notebook
