handle_new_names

data/dataset_handler.py

@@ -34,8 +34,12 @@ DEPRECATED_VIDORE_2_DATASETS_KEYWORDS = [
     "rse_restaurant",
     "mit_biomedical",
     "economics_macro",
+    "biomedical_lectures",
+    "esg_reports",
+    "economics_reports",
 ]
 
+
 def get_datasets_nickname(dataset_name) -> str:
     if dataset_name == "VidoreArxivQARetrieval":
         return "ArxivQA"
@@ -70,6 +74,7 @@ def get_datasets_nickname(dataset_name) -> str:
     else:
         raise ValueError(f"Dataset {dataset_name} not found in ViDoRe")
 
+
 def deprecated_get_datasets_nickname(dataset_name) -> str:
     if "arxivqa" in dataset_name:
         return "ArxivQA"
@@ -99,25 +104,31 @@ def deprecated_get_datasets_nickname(dataset_name) -> str:
     elif "healthcare_industry" in dataset_name:
         return "Healthcare Industry"
 
-    elif "restaurant_esg" in dataset_name:
+    elif ("restaurant_esg" in dataset_name) or ("esg_reports_human" in dataset_name):
         return "ESG Restaurant Human"
 
-    elif "rse_restaurant" in dataset_name and "multilingual" in dataset_name:
+    elif ("rse_restaurant" in dataset_name and "multilingual" in dataset_name) or (
+        "esg_reports" in dataset_name and not "_eng_" in dataset_name
+    ):
         return "ESG Restaurant Synthetic Multilingual"
 
-    elif "rse_restaurant" in dataset_name:
+    elif ("rse_restaurant" in dataset_name) or ("esg_reports" in dataset_name and "_eng_" in dataset_name):
         return "ESG Restaurant Synthetic"
 
-    elif "mit_biomedical" in dataset_name and "multilingual" in dataset_name:
+    elif ("mit_biomedical" in dataset_name and "multilingual" in dataset_name) or (
+        "biomedical_lectures" in dataset_name and not "_eng_" in dataset_name
+    ):
         return "MIT Biomedical Multilingual"
 
-    elif "mit_biomedical" in dataset_name:
+    elif ("mit_biomedical" in dataset_name) or ("biomedical_lectures" in dataset_name and "_eng_" in dataset_name):
         return "MIT Biomedical"
 
-    elif "economics_macro" in dataset_name and "multilingual" in dataset_name:
+    elif ("economics_macro" in dataset_name and "multilingual" in dataset_name) or (
+        "economics_reports" in dataset_name and not "_eng_" in dataset_name
+    ):
         return "Economics Macro Multilingual"
 
-    elif "economics_macro" in dataset_name:
+    elif ("economics_macro" in dataset_name) or ("economics_reports" in dataset_name and "_eng_" in dataset_name):
         return "Economics Macro"
 
     else: