fix: GermanDPR Dataset Causes Cross-Encoder Failure Due to Unexpected…

… dict (#1621) Fixes #1609
embeddings-benchmark · Dec 22, 2024 · 748033e · 748033e
1 parent e1b74f2
commit 748033e
Showing 1 changed file with 1 addition and 0 deletions.
diff --git a/mteb/tasks/Retrieval/deu/GermanDPRRetrieval.py b/mteb/tasks/Retrieval/deu/GermanDPRRetrieval.py
@@ -83,6 +83,7 @@ def load_data(self, **kwargs):
             )
             corpus.update(neg_docs)
             relevant_docs[q_id] = {k: 1 for k in pos_docs}
+        corpus = {doc["id"]: doc.get("title", "") + " " + doc["text"] for doc in corpus}
         self.queries = {self._EVAL_SPLIT: queries}
         self.corpus = {self._EVAL_SPLIT: corpus}
         self.relevant_docs = {self._EVAL_SPLIT: relevant_docs}