Add pipeline tag and links to paper/repository/project page
#1
by nielsr HF Staff - opened
README.md
CHANGED
|
@@ -1,33 +1,48 @@
|
|
| 1 |
---
|
| 2 |
license: apache-2.0
|
|
|
|
| 3 |
---
|
| 4 |
|
| 5 |
-
|
| 6 |
|
| 7 |
-
|
| 8 |
-
The knowledge distillation process follows the procedure proposed in "LEAP: Layer-skipping Efficiency via Adaptive
|
| 9 |
-
Progression for Vision Transformer Distillation"
|
| 10 |
|
| 11 |
-
**
|
| 12 |
|
| 13 |
-
|
| 14 |
|
| 15 |
-
|
| 16 |
|
| 17 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 18 |
|
| 19 |
-
|
|
|
|
|
|
|
| 20 |
|
| 21 |
|
| 22 |
|
| 23 |
|
| 24 |
|
| 25 |
|
| 26 |
-
ImageNet-1K:
|
| 27 |
|
| 28 |
|
| 29 |
|
| 30 |
|
| 31 |
|
| 32 |
-
|
| 33 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
license: apache-2.0
|
| 3 |
+
pipeline_tag: image-classification
|
| 4 |
---
|
| 5 |
|
| 6 |
+
# LEAP: Layer-skipping Efficiency via Adaptive Progression for Vision Transformer Distillation
|
| 7 |
|
| 8 |
+
[Paper](https://huggingface.co/papers/2606.19483) | [GitHub](https://github.com/KevinZ0217/LEAP) | [Project Page](https://kevinz0217.github.io/LEAP_page/)
|
|
|
|
|
|
|
| 9 |
|
| 10 |
+
This repository contains the ViT-Tiny and ViT-S checkpoints (No Register) distilled from ViT-G DINOv2 on ImageNet-100 and ImageNet-1K. The knowledge distillation process follows the procedure proposed in the paper **"LEAP: Layer-skipping Efficiency via Adaptive Progression for Vision Transformer Distillation"**.
|
| 11 |
|
| 12 |
+
### Introduction
|
| 13 |
|
| 14 |
+
Vision Foundation Models (VFMs) with ViT backbones, such as DINOv2, are computationally demanding. LEAP (Layer-skipping Efficiency via Adaptive Progression) is a training curriculum for ViT feature-based knowledge distillation. Instead of supervising the student against a fixed teacher block, LEAP advances the supervisory target through the teacher's feature maps (shallow-to-deep) based on online CKA alignment. This allows the student to build a foundational representation before tackling higher-level abstractions.
|
| 15 |
|
| 16 |
+
### Use cases
|
| 17 |
+
The ViT models output feature maps that can be used for a variety of downstream tasks, including:
|
| 18 |
+
- Image Classification
|
| 19 |
+
- Instance Retrieval
|
| 20 |
+
- Semantic Segmentation
|
| 21 |
|
| 22 |
+
### Performance
|
| 23 |
+
|
| 24 |
+
#### ImageNet-100:
|
| 25 |
|
| 26 |
|
| 27 |
|
| 28 |
|
| 29 |
|
| 30 |
|
| 31 |
+
#### ImageNet-1K:
|
| 32 |
|
| 33 |
|
| 34 |
|
| 35 |
|
| 36 |
|
|
|
|
| 37 |
|
| 38 |
+
|
| 39 |
+
### Citation
|
| 40 |
+
|
| 41 |
+
```bibtex
|
| 42 |
+
@article{leap2024,
|
| 43 |
+
title={LEAP: Layer-skipping Efficiency via Adaptive Progression for Vision Transformer Distillation},
|
| 44 |
+
author={Zheng, Kevin and others},
|
| 45 |
+
journal={arXiv preprint arXiv:2606.19483},
|
| 46 |
+
year={2024}
|
| 47 |
+
}
|
| 48 |
+
```
|