.

code_eval.py

@@ -181,10 +181,8 @@ class CodeEval(evaluate.Metric):
 
         with ThreadPoolExecutor(max_workers=num_workers) as executor:
             results = {}
-            for tid, pred_d, ref_d in zip(ids, predictions, references):
+            for tid, pred, ref in zip(ids, predictions, references):
                 results[tid] = []
-                pred = pred_d[pred_key]
-                ref = ref_d[ref_key]
                 for pid, p in enumerate(pred):
                     result = Result(task_id=tid, prediction_id=pid)
                     body = Template(pred_template).safe_substitute(prediction=p)