【PIR API adaptor No.146】inner (#58931)

Author: cocoshe

python/paddle/tensor/math.py

@@ -2478,7 +2478,7 @@ def inner(x, y, name=None):
 
 
     """
-    if x.size == 1 or y.size == 1:
+    if in_dynamic_mode() and (x.size == 1 or y.size == 1):
         return multiply(x, y)
     else:
         xshape = x.shape
@@ -2488,8 +2488,10 @@ def inner(x, y, name=None):
         nx = x.reshape((-1, xshape[-1]))
         ny = y.reshape((-1, yshape[-1]))
 
-        if in_dynamic_mode():
-            return _C_ops.matmul(nx, ny.T, False, False).reshape(dstshape)
+        if in_dynamic_or_pir_mode():
+            return _C_ops.matmul(
+                nx, paddle.transpose(ny, [1, 0]), False, False
+            ).reshape(dstshape)
         else:
 
             def __check_input(x, y):
@@ -2513,7 +2515,6 @@ def __check_input(x, y):
                         )
 
             __check_input(nx, ny)
-
             helper = LayerHelper('inner', **locals())
             out = helper.create_variable_for_type_inference(dtype=nx.dtype)
             helper.append_op(

test/legacy_test/test_inner.py

@@ -17,12 +17,14 @@
 import numpy as np
 
 import paddle
-from paddle.static import Program, program_guard
+from paddle.pir_utils import test_with_pir_api
 
 
 class TestMultiplyApi(unittest.TestCase):
     def _run_static_graph_case(self, x_data, y_data):
-        with program_guard(Program(), Program()):
+        with paddle.static.program_guard(
+            paddle.static.Program(), paddle.static.Program()
+        ):
             paddle.enable_static()
             x = paddle.static.data(
                 name='x', shape=x_data.shape, dtype=x_data.dtype
@@ -53,45 +55,52 @@ def _run_dynamic_graph_case(self, x_data, y_data):
         res = paddle.inner(x, y)
         return res.numpy()
 
-    def test_multiply(self):
-        np.random.seed(7)
-
+    @test_with_pir_api
+    def test_multiply_static_case1(self):
         # test static computation graph: 3-d array
         x_data = np.random.rand(2, 10, 10).astype(np.float64)
         y_data = np.random.rand(2, 5, 10).astype(np.float64)
         res = self._run_static_graph_case(x_data, y_data)
         np.testing.assert_allclose(res, np.inner(x_data, y_data), rtol=1e-05)
 
+    @test_with_pir_api
+    def test_multiply_static_case2(self):
         # test static computation graph: 2-d array
         x_data = np.random.rand(200, 5).astype(np.float64)
         y_data = np.random.rand(50, 5).astype(np.float64)
         res = self._run_static_graph_case(x_data, y_data)
         np.testing.assert_allclose(res, np.inner(x_data, y_data), rtol=1e-05)
 
+    @test_with_pir_api
+    def test_multiply_static_case3(self):
         # test static computation graph: 1-d array
         x_data = np.random.rand(50).astype(np.float64)
         y_data = np.random.rand(50).astype(np.float64)
         res = self._run_static_graph_case(x_data, y_data)
         np.testing.assert_allclose(res, np.inner(x_data, y_data), rtol=1e-05)
 
+    def test_multiply_dynamic_case1(self):
         # test dynamic computation graph: 3-d array
         x_data = np.random.rand(5, 10, 10).astype(np.float64)
         y_data = np.random.rand(2, 10).astype(np.float64)
         res = self._run_dynamic_graph_case(x_data, y_data)
         np.testing.assert_allclose(res, np.inner(x_data, y_data), rtol=1e-05)
 
+    def test_multiply_dynamic_case2(self):
         # test dynamic computation graph: 2-d array
         x_data = np.random.rand(20, 50).astype(np.float64)
         y_data = np.random.rand(50).astype(np.float64)
         res = self._run_dynamic_graph_case(x_data, y_data)
         np.testing.assert_allclose(res, np.inner(x_data, y_data), rtol=1e-05)
 
+    def test_multiply_dynamic_case3(self):
         # test dynamic computation graph: Scalar
         x_data = np.random.rand(20, 10).astype(np.float32)
         y_data = np.random.rand(1).astype(np.float32).item()
         res = self._run_dynamic_graph_case(x_data, y_data)
         np.testing.assert_allclose(res, np.inner(x_data, y_data), rtol=1e-05)
 
+    def test_multiply_dynamic_case4(self):
         # test dynamic computation graph: 2-d array Complex
         x_data = np.random.rand(20, 50).astype(
             np.float64
@@ -102,6 +111,7 @@ def test_multiply(self):
         res = self._run_dynamic_graph_case(x_data, y_data)
         np.testing.assert_allclose(res, np.inner(x_data, y_data), rtol=1e-05)
 
+    def test_multiply_dynamic_case5(self):
         # test dynamic computation graph: 3-d array Complex
         x_data = np.random.rand(5, 10, 10).astype(
             np.float64
@@ -114,41 +124,49 @@ def test_multiply(self):
 
 
 class TestMultiplyError(unittest.TestCase):
-    def test_errors(self):
+    def test_errors_static_case1(self):
         # test static computation graph: dtype can not be int8
         paddle.enable_static()
-        with program_guard(Program(), Program()):
+        with paddle.static.program_guard(
+            paddle.static.Program(), paddle.static.Program()
+        ):
             x = paddle.static.data(name='x', shape=[100], dtype=np.int8)
             y = paddle.static.data(name='y', shape=[100], dtype=np.int8)
             self.assertRaises(TypeError, paddle.inner, x, y)
 
+    def test_errors_static_case2(self):
         # test static computation graph: inputs must be broadcastable
-        with program_guard(Program(), Program()):
+        paddle.enable_static()
+        with paddle.static.program_guard(
+            paddle.static.Program(), paddle.static.Program()
+        ):
             x = paddle.static.data(name='x', shape=[20, 50], dtype=np.float64)
             y = paddle.static.data(name='y', shape=[20], dtype=np.float64)
             self.assertRaises(ValueError, paddle.inner, x, y)
 
-        np.random.seed(7)
-
+    def test_errors_dynamic_case1(self):
         # test dynamic computation graph: inputs must be broadcastable
         x_data = np.random.rand(20, 5)
         y_data = np.random.rand(10, 2)
         x = paddle.to_tensor(x_data)
         y = paddle.to_tensor(y_data)
         self.assertRaises(ValueError, paddle.inner, x, y)
 
+    def test_errors_dynamic_case2(self):
         # test dynamic computation graph: dtype must be Tensor type
         x_data = np.random.randn(200).astype(np.float64)
         y_data = np.random.randn(200).astype(np.float64)
         y = paddle.to_tensor(y_data)
         self.assertRaises(TypeError, paddle.inner, x_data, y)
 
+    def test_errors_dynamic_case3(self):
         # test dynamic computation graph: dtype must be Tensor type
         x_data = np.random.randn(200).astype(np.float64)
         y_data = np.random.randn(200).astype(np.float64)
         x = paddle.to_tensor(x_data)
         self.assertRaises(TypeError, paddle.inner, x, y_data)
 
+    def test_errors_dynamic_case4(self):
         # test dynamic computation graph: dtype must be Tensor type
         x_data = np.random.randn(200).astype(np.float32)
         y_data = np.random.randn(200).astype(np.float32)