Update app.py

app.py

@@ -179,9 +179,6 @@ def evaluate(
                 eval_results[result["task_id"]].append(result)
                 del future, result
                 gc.collect()
-
-        del problems, futures
-        gc.collect()
         
         # sort the results for each problem by completion_id
         for task_id, task_results in eval_results.items():
@@ -215,7 +212,10 @@ def evaluate(
             for k in pass_k
             if total.min() >= k
         })
-    
+
+        del problems, futures
+        gc.collect()
+        
     pass_at_k["model"] = os.path.basename(samples).split("--bigcodebench-")[0]
     pass_at_k["split"] = split
     pass_at_k["subset"] = subset