Commit
•
2824fa9
1
Parent(s):
94fa102
Update README.md
Browse files
README.md
CHANGED
|
@@ -11,4 +11,28 @@ Model was trained by [Sber AI](https://github.com/sberbank-ai) and [SberDevices]
|
|
| 11 |
* Num Parameters: `150M`
|
| 12 |
* Training Data Volume: `240 million text-image pairs`
|
| 13 |
* Language: `Russian`
|
| 14 |
-
* Image Size: `224`
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 11 |
* Num Parameters: `150M`
|
| 12 |
* Training Data Volume: `240 million text-image pairs`
|
| 13 |
* Language: `Russian`
|
| 14 |
+
* Image Size: `224`
|
| 15 |
+
|
| 16 |
+
## Performance
|
| 17 |
+
We have evaluated the performance on the following datasets:
|
| 18 |
+
|
| 19 |
+
| dataset | metric name | metric result |
|
| 20 |
+
|:--------------|:---------------|:--------------------|
|
| 21 |
+
| Food101 | acc | 0.505 |
|
| 22 |
+
| CIFAR10 | acc | 0.818 |
|
| 23 |
+
| CIFAR100 | acc | 0.504 |
|
| 24 |
+
| Birdsnap | acc | 0.115 |
|
| 25 |
+
| SUN397 | acc | 0.452 |
|
| 26 |
+
| Stanford Cars | acc | 0.433 |
|
| 27 |
+
| DTD | acc | 0.380 |
|
| 28 |
+
| MNIST | acc | 0.447 |
|
| 29 |
+
| STL10 | acc | 0.932 |
|
| 30 |
+
| PCam | acc | 0.501 |
|
| 31 |
+
| CLEVR | acc | 0.148 |
|
| 32 |
+
| Rendered SST2 | acc | 0.489 |
|
| 33 |
+
| ImageNet | acc | 0.375 |
|
| 34 |
+
| FGVC Aircraft | mean-per-class | 0.033 |
|
| 35 |
+
| Oxford Pets | mean-per-class | 0.560 |
|
| 36 |
+
| Caltech101 | mean-per-class | 0.786 |
|
| 37 |
+
| Flowers102 | mean-per-class | 0.401 |
|
| 38 |
+
| HatefulMemes | roc-auc | 0.564 |
|