fix json sheet bugs2

tsv_to_json.py

@@ -33,7 +33,9 @@ def get_image_file_location(root, row):
     tsv_file = csv.reader(file, delimiter="\t")
     flag = 0
     for line in tsv_file:
-        if line[0] not in ["VD", "VS"]:
+        # if line[0] not in ["VD", "VS"]:
+        # if line[0] in ["NOTE", "category"]:
+        if "VD" not in line[0] and "VS" not in line[0]:
             continue
         data_dict = {}
         try:

utils.py

@@ -68,7 +68,10 @@ def check_same_by_chatgpt(data, output_entry, gpt_model="gpt-4", load_json=False
     for sample in tqdm(data):
         if "same" not in sample.keys():
             key = "_".join([sample["category"], sample["subcategory"], str(sample["set_id"]), str(sample["question_id"])])
-            response2 = orig_response[key]
+            try:
+                response2 = orig_response[key]
+            except:
+                from IPython import embed;embed()
             prompt = 'Imagine you are an intelligent teacher. Thoroughly read the two responses to two different questions. Assess the consistency of the information provided within those two responses. '
             prompt += 'You do not know the specific questions, but you can asssess the consistency among the two responses by checking for logical conflicts if both responses are correct. '
             prompt += 'If response1 does not conflict with response2, please generate “same”. Otherwise, generate "different". \n\n response1:'