[Cherry-Pick]Fix expand_sig infershape BUG under static graph mode and...

[Cherry-Pick]Fix expand_sig infershape BUG under static graph mode and NeedTransformPlace behavior if set skip_transform in yaml (#41973)

* [Phi]Fix expand_sig infershape BUG under static graph mode (#41936)

* [Phi]Fix expand_sig infershape BUG under static graph mode

* [Phi]Fix expand_sig infershape BUG under static graph mode

* [Phi]Fix unittest

* [Phi]Fix unittest

* [Eager]Fix NeedTransformPlace behavior if set skip_transform in yaml (#41920)

* [Eager]Fix NeedTransformPlace behavior if set skip_transform in yaml

* add unittest for full_like

* fix unittest

paddle/phi/api/lib/data_transform.cc

@@ -37,9 +37,15 @@ inline bool NeedTransformDataType(const DataType& input,
 inline bool NeedTransformPlace(const paddle::platform::Place& input,
                                const Backend& target,
                                const TransformFlag& transform_flag) {
-  bool ret =
-      input.GetType() == AllocationType::GPUPINNED ||
-      (transform_flag.need_trans_backend() && target != Backend::ALL_BACKEND &&
+  // NOTE(dev): The default value of TransformFlag is True, if it is set with
+  // False
+  // somewhere such as api.yaml or backward.yaml that means we should skip data
+  // transform. Because "stop_transform_" has highest priority.
+  if (!transform_flag.need_trans_backend()) {
+    return false;
+  }
+  bool ret = input.GetType() == AllocationType::GPUPINNED ||
+             (target != Backend::ALL_BACKEND &&
               phi::TransToPhiBackend(input) !=
                   (target != Backend::GPUDNN ? target : Backend::GPU));
   return ret;

paddle/phi/ops/compat/expand_sig.cc

@@ -17,6 +17,11 @@
 namespace phi {
 
 KernelSignature ExpandOpArgumentMapping(const ArgumentMappingContext& ctx) {
+  const auto& shape = paddle::any_cast<std::vector<int>>(ctx.Attr("shape"));
+  // Infer output shape by Attr("shape") in CompileTime if it is specified.
+  if (!ctx.IsRuntime() && !shape.empty()) {
+    return KernelSignature("expand", {"X"}, {"shape"}, {"Out"});
+  }
   if (ctx.HasInput("Shape")) {
     return KernelSignature("expand", {"X"}, {"Shape"}, {"Out"});
   } else if (ctx.InputSize("expand_shapes_tensor") > 0) {
@@ -27,6 +32,12 @@ KernelSignature ExpandOpArgumentMapping(const ArgumentMappingContext& ctx) {
 }
 
 KernelSignature ExpandGradOpArgumentMapping(const ArgumentMappingContext& ctx) {
+  const auto& shape = paddle::any_cast<std::vector<int>>(ctx.Attr("shape"));
+  // Infer output shape by Attr("shape") in CompileTime if it is specified.
+  if (!ctx.IsRuntime() && !shape.empty()) {
+    return KernelSignature(
+        "expand_grad", {"X", "Out@GRAD"}, {"shape"}, {"X@GRAD"});
+  }
   if (ctx.HasInput("Shape")) {
     return KernelSignature("expand_grad",
                            {"X", GradVarName("Out")},

python/paddle/fluid/tests/unittests/test_expand_v2_op.py

@@ -231,6 +231,18 @@ class TestExpandV2API(unittest.TestCase):
         assert np.array_equal(res_3, np.tile(input, (1, 1)))
 
 
+class TestExpandInferShape(unittest.TestCase):
+    def test_shape_with_var(self):
+        with program_guard(Program(), Program()):
+            x = paddle.static.data(shape=[-1, 1, 3], name='x')
+            fake_var = paddle.randn([2, 3])
+            target_shape = [
+                -1, paddle.shape(fake_var)[0], paddle.shape(fake_var)[1]
+            ]
+            out = paddle.expand(x, shape=target_shape)
+            self.assertListEqual(list(out.shape), [-1, -1, -1])
+
+
 if __name__ == "__main__":
     paddle.enable_static()
     unittest.main()

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

@@ -22,6 +22,7 @@ import unittest
 import numpy as np
 from op_test import OpTest
 from paddle.fluid.framework import convert_np_dtype_to_dtype_
+from paddle.fluid.framework import _test_eager_guard
 
 
 class TestFullOp(unittest.TestCase):
@@ -133,5 +134,19 @@ class TestFullLikeOp3(TestFullLikeOp1):
         self.dtype = np.int64
 
 
+@unittest.skipIf(not core.is_compiled_with_cuda(),
+                 "core is not compiled with CUDA")
+class TestFullLikeOp4(unittest.TestCase):
+    def test_skip_data_transform(self):
+        paddle.disable_static()
+        with _test_eager_guard():
+            x = paddle.to_tensor(
+                [1., 2., 3., 4.], place=paddle.CUDAPinnedPlace())
+            out = paddle.full_like(x, 1.)
+            self.assertTrue(
+                (out.numpy() == np.ones([4]).astype(np.float32)).all(), True)
+        paddle.enable_static()
+
+
 if __name__ == "__main__":
     unittest.main()

python/paddle/utils/code_gen/api.yaml

@@ -781,6 +781,8 @@
     param : [x, value, dtype]
     data_type : dtype > x
     backend : place > x
+  data_transform :
+    skip_transform : x
 
 - api : gather
   args : (Tensor x, Tensor index, Scalar axis=0)