From e76fdb5d2f334ce1341b0cefa5a60ebd575df3f1 Mon Sep 17 00:00:00 2001
From: Shobana Marannan <51772793+mshobana@users.noreply.github.com>
Date: Fri, 18 Apr 2025 17:19:56 +0530
Subject: [PATCH] Update the logic of answer_relevance as per the definition

Calling generate 3 times gives the questions with exact words every time. generate_multiple tends to give semantically matching questions.
---
 src/ragas/metrics/_answer_relevance.py | 9 +++------
 1 file changed, 3 insertions(+), 6 deletions(-)

diff --git a/src/ragas/metrics/_answer_relevance.py b/src/ragas/metrics/_answer_relevance.py
index 03b6ea4f9..ab0d6f79c 100644
--- a/src/ragas/metrics/_answer_relevance.py
+++ b/src/ragas/metrics/_answer_relevance.py
@@ -139,15 +139,12 @@ async def _ascore(self, row: t.Dict, callbacks: Callbacks) -> float:
         assert self.llm is not None, "LLM is not set"
 
         prompt_input = ResponseRelevanceInput(response=row["response"])
-        tasks = [
-            self.question_generation.generate(
+        responses = await self.question_generation.generate_multiple(
+                n=self.strictness,
                 data=prompt_input,
                 llm=self.llm,
                 callbacks=callbacks,
-            )
-            for _ in range(self.strictness)
-        ]
-        responses = await asyncio.gather(*tasks)
+        )
 
         return self._calculate_score(responses, row)