README.md

---
license: cc-by-4.0
---

# TACDEC-model
This is a simple model with weights and reproducible code for the results in TACDEC-paper.

What you can find in this repo is:
- The simple [model](https://huggingface.co/SimulaMet-HOST/TACDEC-model/resolve/main/model.py?download=true) used in the TACDEC-paper
- The [weights](https://huggingface.co/SimulaMet-HOST/TACDEC-model/resolve/main/simple_model_weights.pt?download=true) used in the proof-of-concept section in the TACDEC-paper
- A first notebook, [feature_extraction.ipynb](https://huggingface.co/SimulaMet-HOST/TACDEC-model/resolve/main/feature_extraction.ipynb?download=true), that contains a feature extraction process using DINOv2.
- A second notebook, [train_classifier.ipynb](https://huggingface.co/SimulaMet-HOST/TACDEC-model/resolve/main/train_classifier.ipynb?download=true), that uses the features that were either extracted using the first notebook, or downloaded directly from [TACDEC repo](https://huggingface.co/datasets/SimulaMet-HOST/TACDEC).

We highly recommend downloading the already extracted and concatenated (features)[https://huggingface.co/datasets/SimulaMet-HOST/TACDEC/resolve/main/sorted_cls_tokens_features.pt] and the concatenated (labels)[https://huggingface.co/datasets/SimulaMet-HOST/TACDEC/resolve/main/sorted_cls_tokens_labels.npy] if you wish to try the dataset/model. You would then just have to run the second notebook.


If you hold more interest in DINOv2, the **feature_extraction.ipynb** could hold good value.

In both notebooks, there should be good enough documentation, but should you have any questions, see [TACDEC](https://huggingface.co/datasets/SimulaMet-HOST/TACDEC).

## More information
For any other information or information about the dataset: 
[TACDEC](https://huggingface.co/datasets/SimulaMet-HOST/TACDEC)