Update README.md

README.md

@@ -34,16 +34,16 @@ model-index:
     metrics:
     - name: F1
       type: f1
-      value: 0.932
+      value: 0.936
     - name: Precision
       type: precision
       value: 0.960
     - name: Recall
       type: recall
-      value: 0.905
+      value: 0.914
     - name: Accuracy
       type: accuracy
-      value: 0.998
+      value: 0.997
   - task:
       name: Multi-label Text Classification
       type: text-classification
@@ -55,16 +55,16 @@ model-index:
     metrics:
     - name: F1
       type: f1
-      value: 0.789
+      value: 0.780
     - name: Precision
       type: precision
-      value: 0.859
+      value: 0.837
     - name: Recall
       type: recall
-      value: 0.730
+      value: 0.729
     - name: Accuracy
       type: accuracy
-      value: 0.971
+      value: 0.975
 widget:
 - text: "Goedemiddag, ik heb al drie keer gebeld over mijn uitkering en krijg geen duidelijkheid."
 ---
@@ -81,7 +81,7 @@ Trained with a combined objective: alpha · (1 − Soft‑F1) + (1 − alpha) ·
 ## Overview
 - Encoder: mmBERT‑base (multilingual)
 - Heads: 2× MLP (Linear → Dropout → ReLU → Linear)
-- Labels: 96 onderwerp, 26 beleving
+- Labels: 65 onderwerp, 33 beleving
 - Task: Multi‑label classification (sigmoid per class)
 - Thresholds: Disabled (fixed 0.5 used for evaluation/inference)
 
@@ -90,12 +90,12 @@ Trained with a combined objective: alpha · (1 − Soft‑F1) + (1 − alpha) ·
 - Useful for analytics, routing, and trend insights. Not intended for legal or benefit decisions without human review.
 
 ## Training Data
-- Source: `UWV/wim-synthetic-data-rd` (train split)
-- Samples: 9,351
-- Labels: 96 onderwerp, 26 beleving
-- Avg labels per sample: onderwerp 1.75, beleving 1.89
-- Shapes: onderwerp (9351, 96), beleving (9351, 26)
-- Train/Val split: 7,480 / 1,871 (80/20)
+- Source: `UWV/wim-synthetic-data-rd` (train split) + `UWV/wim_synthetic_data_for_testing_split_labels` (train split, 3,500 samples)
+- Samples: 12,491
+- Labels: 65 onderwerp, 33 beleving
+- Avg labels per sample: onderwerp 1.85, beleving 2.08
+- Shapes: onderwerp (12491, 65), beleving (12491, 33)
+- Train/Val split: 11,866 / 625
 
 ## Training Setup
 - Date: 2025‑10‑20
@@ -120,18 +120,18 @@ Trained with a combined objective: alpha · (1 − Soft‑F1) + (1 − alpha) ·
 ## Metrics
 Final validation (500 samples):
 - Onderwerp:
-  - Accuracy: 99.8%
+  - Accuracy: 99.7%
   - Precision: 0.960
-  - Recall: 0.905
-  - F1: 0.932
+  - Recall: 0.914
+  - F1: 0.936
 - Beleving:
-  - Accuracy: 97.1%
-  - Precision: 0.859
-  - Recall: 0.730
-  - F1: 0.789
+  - Accuracy: 97.5%
+  - Precision: 0.837
+  - Recall: 0.729
+  - F1: 0.780
 - Combined:
-  - Average Accuracy: 98.4%
-  - Average F1: 0.861
+  - Average Accuracy: 98.6%
+  - Average F1: 0.858
 
 ## Saved Artifacts
 - HF‑compatible files: