[Fix] removal of not working benchmarks

src/dataloading.py

@@ -87,7 +87,7 @@ def get_leaderboard_datasets(model_ids):
         common_datasets = set.intersection(*model_datasets.values())
 
     # Filter datasets that are not MCQ or currently do not work
-    ignore = ["_bbh_", "_gpqa_", "_math_", "_ifeval_"]
+    ignore = ["bbh_", "gpqa_", "math_", "ifeval"]
     discard = []
     for dataset in common_datasets:
         for ignore_data in ignore:
@@ -104,7 +104,6 @@ def filter_labels(doc):
         for d in doc:
             labels.append(d["answer_index"])
     else:
-        print("Target:", doc[0]["target"])
         for d in doc:
             if d["target"] == "False":
                 labels.append(0)
@@ -147,6 +146,3 @@ def load_run_data(model_name, dataset_name):
 def load_run_data_cached(model_name, dataset_name):
     return load_run_data(model_name, dataset_name)
 
-
-if __name__ == "__main__":
-    get_leaderboard_models_reload()