diff --git a/lmeval/benchmark.py b/lmeval/benchmark.py
index 885cca6..2d3a0bb 100644
--- a/lmeval/benchmark.py
+++ b/lmeval/benchmark.py
@@ -490,7 +490,8 @@ def load_benchmark(path: str, archive = None, use_tempfile: bool | None = None)
     media_to_load = []
     for category in benchmark.categories:
         for task in category.tasks:
-            if isinstance(task.scorer.type, ScorerType):
+            scorer_values = set(item.value for item in ScorerType)
+            if task.scorer.type in scorer_values:
                 scorer_name = str(task.scorer.type)
                 stype = ScorerType[scorer_name]
             else:
diff --git a/lmeval/question.py b/lmeval/question.py
index 1db0c07..1970e6e 100644
--- a/lmeval/question.py
+++ b/lmeval/question.py
@@ -37,10 +37,13 @@ class Question(CustomModel):
     # we need it for (de)serialization - automated added by Task.add()
     id: int = Field(default=-1)
     question: str | None = Field(default=None)
+    description: str = Field(default="")
     language: str = Field(default="en")
 
     # answer
     answer: str | None = Field(default=None)
+    # old answer (e.g. in case the anwere (label) changed due to a manual review)
+    answer_old: str | None = Field(default=None)
     # additional answers e.g. for multiple choice questions
     additional_answers: List[str] = Field(default_factory=list)
 
@@ -94,6 +97,7 @@ def add_media(self, path: str| Path, filetype: FileType = FileType.auto,
                      ".mp4": [FileType.mp4, Modality.video],
                      ".py": [FileType.python, Modality.code],
                      ".pdf": [FileType.pdf, Modality.document],
+                     ".html": [FileType.html, Modality.code],
                     }
 
         # auto detection if needed
@@ -124,4 +128,4 @@ def _compute_file_hash(self, path: Path,
     
 class GroupedQuestion(Question):
     metadata: Dict[str, Any]
-    question_set: List[Question] 
\ No newline at end of file
+    question_set: List[Question]