diff --git a/utils.py b/utils.py
index 0a1ddf2..30ba957 100644
--- a/utils.py
+++ b/utils.py
@@ -34,7 +34,16 @@ def evaluate_by_chatgpt(data, output_entry, correctness_entry, gpt_model="gpt-4"
         prompt += sample[output_entry]
         prompt += '\nOutput:'
 
-        response = openai.ChatCompletion.create(model=gpt_model, messages=[{"role": "user", "content": prompt}], api_key=api_key)
+        while True:
+            try:
+                response = openai.ChatCompletion.create(
+                    model=gpt_model, 
+                    messages=[{"role": "user", "content": prompt}], 
+                    api_key=api_key, 
+                    timeout=5)
+                break
+            except:
+                time.sleep(5)  # Wait for 5 seconds before retrying
 
         output_text = response['choices'][0]['message']['content']
 
@@ -68,10 +77,8 @@ 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"])])
-            try:
-                response2 = orig_response[key]
-            except:
-                from IPython import embed;embed()
+            response2 = orig_response[key]
+
             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:'
@@ -80,7 +87,18 @@ def check_same_by_chatgpt(data, output_entry, gpt_model="gpt-4", load_json=False
             prompt += response2
             prompt += '\nOutput:'
 
-            response = openai.ChatCompletion.create(model=gpt_model, messages=[{"role": "user", "content": prompt}], api_key=api_key)
+            while True:
+                try:
+                    response = openai.ChatCompletion.create(
+                        model=gpt_model, 
+                        messages=[{"role": "user", "content": prompt}], 
+                        api_key=api_key, 
+                        timeout=5)
+
+                    break
+                except:
+                    time.sleep(5)  # Wait for 5 seconds before retrying
+
 
             output_text = response['choices'][0]['message']['content']