Update mmlu_eval_original.py

mmlu_eval_original.py

@@ -106,7 +106,6 @@ def eval (subject, model, tokenizer, dev_df, test_df, num_questions_per_subject=
         
         logger.info (f"Sample: {i}")
 
-        #logger.info (f"Prompt: {prompt}")
 
         label = test_df.iloc[i, test_df.shape[1] - 1]
 
@@ -132,12 +131,13 @@ def eval (subject, model, tokenizer, dev_df, test_df, num_questions_per_subject=
         pred = {0: "A", 1: "B", 2: "C", 3: "D"}[np.argmax(probs)]
 
         cor = pred == label
-
-        #logger.info(f"Label: {label}")
-        #logger.info(f"Logits: {logits}")
-        #logger.info(f"Probabilities: {probs}")
-        #logger.info(f"Prediction: {pred}")
-        #logger.info(f"Correct: {cor}")
+        if (i == 0):
+            logger.info (f"Prompt: {prompt}")
+            logger.info(f"Label: {label}")
+            logger.info(f"Logits: {logits}")
+            logger.info(f"Probabilities: {probs}")
+            logger.info(f"Prediction: {pred}")
+            logger.info(f"Correct: {cor}")
 
         cors.append(cor)
         all_probs.append(probs)