test=develop

python/paddle/fluid/metrics.py

@@ -243,6 +243,7 @@ class Precision(MetricBase):
             raise ValueError("The 'labels' must be a numpy ndarray.")
         sample_num = labels.shape[0]
         preds = np.rint(preds).astype("int32")
+
         for i in range(sample_num):
             pred = preds[i]
             label = labels[i]