fix dygraph trace bug, test=develop (#21193)

python/paddle/fluid/dygraph/jit.py

@@ -376,7 +376,7 @@ class TracedLayer(object):
             if partial_vars is None:
                 return all_vars
 
-            return [all_vars[idx] for idx in feed]
+            return [all_vars[idx] for idx in partial_vars]
 
         with scope_guard(self._scope):
             feeded_var_names = get_feed_fetch(self._feed_names, feed)

python/paddle/fluid/tests/unittests/test_imperative_transformer_sorted_gradient.py

@@ -988,8 +988,9 @@ class TestDygraphTransformerSortGradient(unittest.TestCase):
                 if i % 2 == 0:
                     outs, traced_layer = TracedLayer.trace(
                         transformer, [enc_inputs, dec_inputs, label, weights])
-                    outs_static = traced_layer(enc_inputs + dec_inputs +
-                                               [label, weights])
+
+                    ins_static = enc_inputs + dec_inputs + [label, weights]
+                    outs_static = traced_layer(ins_static)
                     helper.assertEachVar(outs, outs_static)
                     if program is not None:
                         self.assertTrue(
@@ -997,7 +998,9 @@ class TestDygraphTransformerSortGradient(unittest.TestCase):
 
                     program = traced_layer.program
                     traced_layer.save_inference_model(
-                        './infer_imperative_transformer')
+                        './infer_imperative_transformer',
+                        feed=range(len(ins_static)),
+                        fetch=range(len(outs_static)))
                 else:
                     outs = transformer(enc_inputs, dec_inputs, label, weights)