Skip to content

Latest commit

 

History

History
22 lines (15 loc) · 916 Bytes

README.md

File metadata and controls

22 lines (15 loc) · 916 Bytes

Linear Evaluation on ImageNet-1K

Installation

Please follow the installation instructions in pretrain.

Dataset

We prepare ImageNet-1K following the instructions in DINOv2 (there are additional files needed compared to the pre-training stage).

Training

  1. Specify the config file with config-file in the training script run_probing.sh.

  2. Use the pretrained-weights parameter to provide the path to your pre-trained model.

  3. Replace imagenet_path in the train-dataset and val-dataset parameters with the directory where your datasets are located.

  4. Simply run the training script as follows:

    bash run_probing.sh

Acknowledgment

This part is heavily build upon DINOv2. We gratefully thank the authors for their wonderful works.