heyongxin233/FiSeR-DINOv3-ViT-L16

Full fine-tune vs facebook/dinov3-vitl16-pretrain-lvd1689m. Fine-tune: 100% of the weights changed, across 24 of 24 layers.

Tensors changed: Embeddings 4/5, Attention 168/168, other 48/48, MLP 96/96, Norms 98/98

On the Hub · Downloads on Model Pulse