1.1

README.md

@@ -6,18 +6,18 @@ language:
 - en
 metrics:
   - type: f1         
-    value: 0.8188 
+    value: 0.8345 
     name: micro F1
     args:
-      threshold: 0.41
+      threshold: 0.46
   - type: NDCG@3         
-    value: 0.8778  
+    value: 0.8819  
     name: NDCG@5
   - type: NDCG@5         
-    value: 0.8625 
+    value: 0.8689 
     name: NDCG@5
   - type: NDCG@10         
-    value: 0.8711 
+    value: 0.8780 
     name: NDCG@10
 tags:
 - eurovoc
@@ -105,7 +105,7 @@ This classification model is built on top of [EUBERT](https://huggingface.co/Eur
 With less than 100 million parameters, it can be deployed on commodity hardware without GPU acceleration (around 200 ms per inference for 2000 characters).
 
 Parameters :
-- Number of epochs 10
+- Number of epochs 16
 - Batch size  10
 - Max lenght 512
 - Learning Rate 5e-05
@@ -126,10 +126,10 @@ On Eurovoc Dataset version 23.08 with a stratification ratio 90/10 for training/
 
 | Metric     | Value      | Threshold Value |
 |------------|------------|-----------------|
-| Micro F1   | 0.8188     | 0.41            |
-| NDCG@3     | 0.8778     | -               |
-| NDCG@5     | 0.8625     | -               |
-| NDCG@10    | 0.8711     | -               |
+| Micro F1   | 0.8345     | 0.46            |
+| NDCG@3     | 0.8819     | -               |
+| NDCG@5     | 0.8689     | -               |
+| NDCG@10    | 0.8780     | -               |
 
 These values are higher than the state of the art previously known in the field, see publications: