transfer the static.accurcay to v2 op (#35494)

* transfer the static.accurcay to v2 api

* remove the unused code

python/paddle/fluid/layers/metric_op.py

@@ -84,7 +84,9 @@ def accuracy(input, label, k=1, correct=None, total=None):
         if total is None:
             total = _varbase_creator(dtype="int32")
 
-        topk_out, topk_indices = nn.topk(input, k=k)
+        _k = k.numpy().item(0) if isinstance(k, Variable) else k
+        topk_out, topk_indices = _C_ops.top_k_v2(input, 'k', _k, 'sorted',
+                                                 False)
         _acc, _, _ = _C_ops.accuracy(topk_out, topk_indices, label, correct,
                                      total)
         return _acc
@@ -92,7 +94,20 @@ def accuracy(input, label, k=1, correct=None, total=None):
     helper = LayerHelper("accuracy", **locals())
     check_variable_and_dtype(input, 'input', ['float16', 'float32', 'float64'],
                              'accuracy')
-    topk_out, topk_indices = nn.topk(input, k=k)
+    topk_out = helper.create_variable_for_type_inference(dtype=input.dtype)
+    topk_indices = helper.create_variable_for_type_inference(dtype="int64")
+    inputs = {"X": [input]}
+    if isinstance(k, Variable):
+        inputs['K'] = [k]
+    else:
+        attrs = {'k': k}
+    attrs['sorted'] = False
+    helper.append_op(
+        type="top_k_v2",
+        inputs=inputs,
+        attrs=attrs,
+        outputs={"Out": [topk_out],
+                 "Indices": [topk_indices]})
     acc_out = helper.create_variable_for_type_inference(dtype="float32")
     if correct is None:
         correct = helper.create_variable_for_type_inference(dtype="int32")

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

@@ -78,6 +78,42 @@ class TestAccuracyOpError(unittest.TestCase):
             paddle.metric.accuracy(input=x3, label=label)
 
 
+class TestAccuracyAPI1(unittest.TestCase):
+    def setUp(self):
+        self.predictions = paddle.static.data(
+            shape=[2, 5], name="predictions", dtype="float32")
+        self.label = paddle.static.data(
+            shape=[2, 1], name="labels", dtype="int64")
+        self.result = paddle.static.accuracy(
+            input=self.predictions, label=self.label, k=1)
+        self.input_predictions = np.array(
+            [[0.2, 0.1, 0.4, 0.1, 0.1], [0.2, 0.3, 0.1, 0.15, 0.25]],
+            dtype="float32")
+        self.input_labels = np.array([[2], [0]], dtype="int64")
+        self.expect_value = np.array([0.5], dtype='float32')
+
+    def test_api(self):
+        exe = paddle.static.Executor()
+        result, = exe.run(feed={
+            "predictions": self.input_predictions,
+            'labels': self.input_labels
+        },
+                          fetch_list=[self.result.name])
+        self.assertEqual((result == self.expect_value).all(), True)
+
+
+class TestAccuracyAPI2(unittest.TestCase):
+    def test_api(self):
+        with fluid.dygraph.guard():
+            predictions = paddle.to_tensor(
+                [[0.2, 0.1, 0.4, 0.1, 0.1], [0.2, 0.3, 0.1, 0.15, 0.25]],
+                dtype='float32')
+            label = paddle.to_tensor([[2], [0]], dtype="int64")
+            result = paddle.static.accuracy(input=predictions, label=label, k=1)
+            expect_value = np.array([0.5], dtype='float32')
+            self.assertEqual((result.numpy() == expect_value).all(), True)
+
+
 class TestAccuracyAPI(unittest.TestCase):
     def test_api(self):
         with fluid.dygraph.guard():