Allow benchmarking tool to handle empty response (#12174)

Co-authored-by: Claude <noreply@anthropic.com>
2025-10-29 20:26:23 -07:00
parent 25c5049870
commit 2b71531a70
5 changed files with 9 additions and 3 deletions
@@ -204,6 +204,7 @@ class MMMUVLMEval(Eval):

            # Sample
            response_text = sampler(prompt_messages)
+            response_text = response_text or ""

            # Parse and score
            gold = sample["answer"]