Upload README.md

README.md

@@ -12,28 +12,29 @@ task_categories:
 - image-classification
 pretty_name: CleanPatrick - Data Cleaning Benchmark
 configs:
-- config_name: Combined Annotations
+- config_name: combined_annotations
   data_files:
   - split: test
     path: "combined_annotations.csv"
-- config_name: Label Errors
+- config_name: label_errors
   data_files:
   - split: aggregated
     path: "label_errors.csv"
   - split: metadata
     path: "label_errors_meta.csv"
-- config_name: Near Duplicates
+- config_name: near_duplicates
   data_files:
   - split: aggregated
     path: "near_duplicates.csv"
   - split: metadata
     path: "near_duplicates_meta.csv"
-- config_name: Off-Topic Samples
+- config_name: off_topic_samples
   data_files:
   - split: aggregated
     path: "off_topic_samples.csv"
   - split: metadata
     path: "off_topic_samples_meta.csv"
+default_config_name: combined_annotations
 ---
 # CleanPatrick: A Benchmark for Data Cleaning
 
@@ -81,7 +82,9 @@ To load the dataset using the HuggingFace `datasets` library:
 ```python
 from datasets import load_dataset
 
-dataset = load_dataset("Digital-Dermatology/CleanPatrick")
+# Load a specific configuration
+dataset = load_dataset("Digital-Dermatology/CleanPatrick", "label_errors")
+# Available configs: combined_annotations, label_errors, near_duplicates, off_topic_samples
 ```
 
 ## Citation