fix groupby

code_eval.py

@@ -212,13 +212,14 @@ class CodeEval(evaluate.Metric):
         new_results = {}
         for key, result in results.items():
             new_result = []
+            result.sort(key=lambda x: x[0])
             for completion_id, group in itertools.groupby(result, key=lambda x: x[0]):
                 group = list(group)
                 new_result.append(
                     (
-                        group[0][0],
+                        completion_id,
                         dict(
-                            task_id=group[0][0],
+                            task_id=key,
                             passed=all(r[1]["passed"] for r in group),
                             result=[r[1]["result"] for r in group],
                             completion_id=completion_id,