Fix var name in SE-RexNeXt

fluid/image_classification/se_resnext.py

@@ -145,7 +145,7 @@ def train(learning_rate, batch_size, num_passes, model_save_dir='model'):
 
         test_accuracy.reset(exe)
         for data in test_reader():
-            out, acc = exe.run(inference_program,
+            loss, acc = exe.run(inference_program,
                                 feed=feeder.feed(data),
                                 fetch_list=[avg_cost] + test_accuracy.metrics)
         test_pass_acc = test_accuracy.eval(exe)