return list for single eval

openpo/client.py

@@ -73,6 +73,7 @@ def _get_model_consensus(
         res_a: List[Dict],
         res_b: List[Dict],
     ) -> List[int]:
+        # TODO: implement error handling to skip?
         if len(res_a) != len(res_b):
             raise ValueError(
                 "responses for evaluation do not have identical number of responses."
@@ -169,7 +170,7 @@ def eval_single(
                 result = res.content[0].input['"evaluation']
             result = json.loads(res.choices[0].message.content)["evaluation"]
 
-            return {"evaluation": result}
+            return {"evaluation": [result]}
         except (AuthenticationError, ValueError) as e:
             raise e
         except Exception as e: