Upload README.md with huggingface_hub
Browse files
README.md
CHANGED
|
@@ -1,3 +1,32 @@
|
|
| 1 |
-
---
|
| 2 |
-
license:
|
| 3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: other
|
| 3 |
+
library_name: cra5
|
| 4 |
+
tags:
|
| 5 |
+
- climate
|
| 6 |
+
- weather
|
| 7 |
+
- compression
|
| 8 |
+
- vaeformer
|
| 9 |
+
---
|
| 10 |
+
|
| 11 |
+
# CRA5: Extreme Compression of ERA5
|
| 12 |
+
|
| 13 |
+
This repository contains the pretrained weights for the CRA5 project.
|
| 14 |
+
|
| 15 |
+
## Usage
|
| 16 |
+
|
| 17 |
+
You can use these weights with the CRA5 library.
|
| 18 |
+
|
| 19 |
+
```python
|
| 20 |
+
from cra5.models.compressai.zoo import vaeformer_pretrained
|
| 21 |
+
# Note: You need to update the model URLs in your library to point here
|
| 22 |
+
# or manually load the state dict.
|
| 23 |
+
|
| 24 |
+
# Example if using updated library:
|
| 25 |
+
model = vaeformer_pretrained(quality=268, pretrained=True)
|
| 26 |
+
```
|
| 27 |
+
|
| 28 |
+
## Models
|
| 29 |
+
- `cra5_268v_300k.pth`: Model with 268 variables.
|
| 30 |
+
- `cra5_159v_150k.pth`: Model with 159 variables.
|
| 31 |
+
|
| 32 |
+
See [GitHub Repository](https://github.com/taohan10200/CRA5) for more details.
|