Skip to content

README.md

Latest commit

 

History

History
42 lines (28 loc) · 1.17 KB

README.md

File metadata and controls

42 lines (28 loc) · 1.17 KB
Raw

DINOv3

Meta's DINOv3 vision foundation model for feature extraction.

Requirements

HuggingFace Token Required - DINOv3 models are gated and require authentication.

  1. Accept license at https://huggingface.co/facebook/dinov3-vits16-pretrain-lvd1689m (click on the Files and versions tab)
  2. Get token from https://huggingface.co/settings/tokens
  3. Update packages/vision/dinov3/config.yaml:
environment_variables:
- "HF_TOKEN=hf_your_actual_token_here"

Build and Run

ryzers build dinov3
ryzers run

Model

By default test.py uses facebook/dinov3-vits16-pretrain-lvd1689m (21.6M parameters - smallest variant).

Find list of models on Hugging Face, e.g:

  • facebook/dinov3-vitb16-pretrain-lvd1689m (85.7M)
  • facebook/dinov3-vitl16-pretrain-lvd1689m (0.3B)
  • facebook/dinov3-vit7b16-pretrain-lvd1689m (7B)

References

Copyright(C) 2025 Advanced Micro Devices, Inc. All rights reserved.