diff --git a/libs/langchain/langchain/smith/evaluation/string_run_evaluator.py b/libs/langchain/langchain/smith/evaluation/string_run_evaluator.py index 4016482c191..0319433808c 100644 --- a/libs/langchain/langchain/smith/evaluation/string_run_evaluator.py +++ b/libs/langchain/langchain/smith/evaluation/string_run_evaluator.py @@ -375,7 +375,11 @@ class StringRunEvaluatorChain(Chain, RunEvaluator): ) # Configure how example rows are fed as a reference string to the evaluator - if reference_key is not None or data_type in (DataType.llm, DataType.chat): + if ( + reference_key is not None + or data_type in (DataType.llm, DataType.chat) + or evaluator.requires_reference + ): example_mapper = StringExampleMapper(reference_key=reference_key) elif evaluator.requires_reference: raise ValueError(