Upload README.md with huggingface_hub

README.md

@@ -8,7 +8,7 @@ tags:
 - lstm
 license: mit
 datasets:
-- leonvanbokhorst/topic-drift
+- leonvanbokhorst/topic-drift-v2
 metrics:
 - rmse
 - r2_score
@@ -24,10 +24,10 @@ model-index:
     metrics:
       - name: Test RMSE
         type: rmse
-        value: 0.0153
+        value: 0.0139
       - name: Test R²
         type: r2
-        value: 0.8500
+        value: 0.8766
       - name: Test Loss 
         type: loss
         value: 0.0002
@@ -35,7 +35,7 @@ model-index:
 
 # Topic Drift Detector Model
 
-## Version: v20241225_160448
+## Version: v20241225_162244
 
 This model detects topic drift in conversations using an enhanced attention-based architecture. Trained on the [leonvanbokhorst/topic-drift](https://huggingface.co/datasets/leonvanbokhorst/topic-drift) dataset.
 
@@ -50,18 +50,18 @@ This model detects topic drift in conversations using an enhanced attention-base
 ## Performance Metrics
 ```txt
 === Full Training Results ===
-Best Validation RMSE: 0.0145
-Best Validation R²: 0.8656
+Best Validation RMSE: 0.0133
+Best Validation R²: 0.8873
 
 === Test Set Results ===
 Loss: 0.0002
-RMSE: 0.0153
-R²: 0.8500
+RMSE: 0.0139
+R²: 0.8766
 
 ```
 
 ## Training Curves
-![Training Curves](plots/v20241225_160448/training_curves.png)
+![Training Curves](plots/v20241225_162244/training_curves.png)
 
 ## Usage
 ```python
@@ -73,7 +73,7 @@ base_model = AutoModel.from_pretrained('BAAI/bge-m3')
 tokenizer = AutoTokenizer.from_pretrained('BAAI/bge-m3')
 
 # Load topic drift detector
-model = torch.load('models/v20241225_160448/topic_drift_model.pt')
+model = torch.load('models/v20241225_162244/topic_drift_model.pt')
 model.eval()
 
 # Prepare conversation window (8 turns)