Fix accuracy metric in sentiment anlaysis

src/deepeval/sentiment_analysis_task.py

@@ -18,7 +18,7 @@ class SentimentAnalysisTask(BaseTask):
             messages = prompt
             answer = self.generate_response_mcqa(messages, choices=["positive", "negative", "neutral"])
             responses.append(answer)
-            if row["sentiment"] == "positive":
+            if row["sentiment"] == answer:
                 true += 1
 
         print(responses)