Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
pipeline_tag: sentence-similarity
|
| 3 |
tags:
|
| 1 |
+
|
| 2 |
+
The SBERT model was trained on the dataset of UNO sustainable development goals. The total dataset size is 20000 records. 16000 were used for training and 4000 for evaluation.
|
| 3 |
+
|
| 4 |
+
The similarity between records was calculated based on the class similarity:
|
| 5 |
+
0 (case 1 - no common classes)
|
| 6 |
+
(number of common classes)/(number of all classes) (case 2)
|
| 7 |
+
(number of common classes)/(maximal number of record classes)+(number of common classes)/(number of all classes) (case 3)
|
| 8 |
+
|
| 9 |
---
|
| 10 |
pipeline_tag: sentence-similarity
|
| 11 |
tags:
|