maintain schema

app.py

@@ -124,7 +124,18 @@ def push_results_to_hub(user_answers, token: gr.OAuthToken | None):
     repo_id = f"{EXAM_DATASET_ID}_student_responses"
     submission_time = datetime.now().strftime("%Y-%m-%d %H:%M:%S")
 
-    new_ds = Dataset.from_list([{"is_correct":row["is_correct"]} for row in user_answers])
+    # filter down user answers to only include first character of the question and the answer
+    user_answers = [
+        {
+            "question": answer["question"][:1],
+            "selected_answer": answer["selected_answer"][:1],
+            "is_correct": answer["is_correct"],
+            "correct_answer": answer["correct_answer"][:1],
+            "correct_reference": answer["correct_reference"],
+        }
+        for answer in user_answers
+    ]
+    new_ds = Dataset.from_list(user_answers)
     new_ds = new_ds.map(
         lambda x: {
             "username": user_info["name"],