Instructions to use apple/DFN5B-CLIP-ViT-H-14-378 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- OpenCLIP
How to use apple/DFN5B-CLIP-ViT-H-14-378 with OpenCLIP:
import open_clip model, preprocess_train, preprocess_val = open_clip.create_model_and_transforms('hf-hub:apple/DFN5B-CLIP-ViT-H-14-378') tokenizer = open_clip.get_tokenizer('hf-hub:apple/DFN5B-CLIP-ViT-H-14-378') - Notebooks
- Google Colab
- Kaggle
Update README.md
Browse files
README.md
CHANGED
|
@@ -41,7 +41,7 @@ These weights are directly usable in OpenCLIP (image + text).
|
|
| 41 |
| ImageNet-R | 0.937633 |
|
| 42 |
| KITTI Vehicle Distance | 0.38256 |
|
| 43 |
| MNIST | 0.8372 |
|
| 44 |
-
| ObjectNet
|
| 45 |
| Oxford Flowers-102 | 0.896834 |
|
| 46 |
| Oxford-IIIT Pet | 0.966841 |
|
| 47 |
| Pascal VOC 2007 | 0.826255 |
|
|
@@ -63,7 +63,7 @@ These weights are directly usable in OpenCLIP (image + text).
|
|
| 63 |
| **Average** | **0.709421** |
|
| 64 |
|
| 65 |
|
| 66 |
-
[
|
| 67 |
## Model Usage
|
| 68 |
### With OpenCLIP
|
| 69 |
```
|
|
|
|
| 41 |
| ImageNet-R | 0.937633 |
|
| 42 |
| KITTI Vehicle Distance | 0.38256 |
|
| 43 |
| MNIST | 0.8372 |
|
| 44 |
+
| ObjectNet <sup>1</sup> | 0.796867 |
|
| 45 |
| Oxford Flowers-102 | 0.896834 |
|
| 46 |
| Oxford-IIIT Pet | 0.966841 |
|
| 47 |
| Pascal VOC 2007 | 0.826255 |
|
|
|
|
| 63 |
| **Average** | **0.709421** |
|
| 64 |
|
| 65 |
|
| 66 |
+
[1]: Center-crop pre-processing used for ObjectNet (squashing results in 6pp lower accuracy)
|
| 67 |
## Model Usage
|
| 68 |
### With OpenCLIP
|
| 69 |
```
|