Merge branch 'main' into convert_gen

lm_eval/api/task.py

@@ -1626,7 +1626,7 @@ class ConfigurableTask(Task):
                     # This allows for multiple metrics to be returned from the same function
                     for k, v in result_score.items():
                         result_dict[k] = v
-                        return result_dict
+                else:
                     result_dict[metric] = result_score
         else:
             raise ValueError(