Upload folder using huggingface_hub

README.md

@@ -11,8 +11,8 @@ This dataset contains 1,000 web pages with human-annotated ground truth for eval
 CSV format with columns:
 - `id`: Sequential identifier
 - `url`: Full URL of the web page  
-- `important_content`: ~100-word core snippet (main content)
-- `not_important_content`: ~10-word non-core snippet (navigation/footer/ads)
+- `truth_text`: ~100-word core snippet (main content)
+- `lie_text`: ~10-word non-core snippet (navigation/footer/ads)
 - `error`: Optional error message if page retrieval failed
 
 ## Limitations