diff --git a/modules/deeplearning/cnn_cld_frac.py b/modules/deeplearning/cnn_cld_frac.py
index 6c36a4d19948d5f911e2c541a5ae948e902b7102..951d8c0bf2f370e5161449851ad5314405f8adaa 100644
--- a/modules/deeplearning/cnn_cld_frac.py
+++ b/modules/deeplearning/cnn_cld_frac.py
@@ -763,11 +763,11 @@ class SRCNN:
         preds = np.concatenate(self.test_preds)
         print(labels.shape, preds.shape)
 
-        if label_param != 'cloud_probability':
-            labels_denorm = denormalize(labels, label_param, mean_std_dct)
-            preds_denorm = denormalize(preds, label_param, mean_std_dct)
+        # if label_param != 'cloud_probability':
+        #     labels_denorm = denormalize(labels, label_param, mean_std_dct)
+        #     preds_denorm = denormalize(preds, label_param, mean_std_dct)
 
-        return labels_denorm, preds_denorm
+        return labels, preds
 
     def do_evaluate(self, data, ckpt_dir):
 
@@ -815,9 +815,10 @@ class SRCNN:
 
 def run_restore_static(directory, ckpt_dir, out_file=None):
     nn = SRCNN()
-    labels_denorm, preds_denorm = nn.run_restore(directory, ckpt_dir)
+    labels, preds = nn.run_restore(directory, ckpt_dir)
     if out_file is not None:
-        np.save(out_file, [np.squeeze(labels_denorm), preds_denorm.argmax(axis=3)])
+        np.save(out_file,
+                [np.squeeze(labels), preds.argmax(axis=3), preds[:, :, :, 0], preds[:, :, :, 1], preds[:, :, :, 2]])
 
 
 def run_evaluate_static(in_file, out_file, ckpt_dir):