Update SPC-UQ/Image_Classification/README.md
Browse files
SPC-UQ/Image_Classification/README.md
CHANGED
|
@@ -25,6 +25,17 @@ The implementation is based on [DDU](https://github.com/omegafragger/DDU), with
|
|
| 25 |
|
| 26 |
**Note:** All datasets should be placed in the [data](data) directory.
|
| 27 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 28 |
## Pretrained model weights
|
| 29 |
We provide several pretrained model weights to facilitate **reproducibility** of our experiments.
|
| 30 |
|
|
|
|
| 25 |
|
| 26 |
**Note:** All datasets should be placed in the [data](data) directory.
|
| 27 |
|
| 28 |
+
|
| 29 |
+
## Additional Environment Requirements
|
| 30 |
+
Since the application of Laplace approximation depends on the [Laplace library](https://github.com/aleximmer/Laplace). It requires extra dependencies beyond the default `environment.yml`.
|
| 31 |
+
|
| 32 |
+
If you want to run the Laplace approximation baseline, please install the following libraries manually:
|
| 33 |
+
```
|
| 34 |
+
pip install laplace-torch
|
| 35 |
+
pip install git+https://[email protected]/wiseodd/asdl@asdfghjkl
|
| 36 |
+
```
|
| 37 |
+
|
| 38 |
+
|
| 39 |
## Pretrained model weights
|
| 40 |
We provide several pretrained model weights to facilitate **reproducibility** of our experiments.
|
| 41 |
|