transfer nearest_interp op to phi, change name from nearest_interp_v2 to nearest_interp (#45148)

paddle/phi/api/yaml/legacy_api.yaml

@@ -1852,6 +1852,17 @@
     func : multiply
   backward : multiply_grad
 
+- api : nearest_interp
+  args : (Tensor x, Tensor out_size, Tensor[] size_tensor, Tensor scale_tensor, str data_layout, int out_d, int out_h, int out_w, float[] scale, str interp_method, bool align_corners, int align_mode)
+  output : Tensor(output)
+  infer_meta :
+    func : InterpolateInferMeta
+  optional: out_size, size_tensor, scale_tensor
+  kernel :
+    func : nearest_interp
+    data_type : x
+  backward : nearest_interp_grad
+
 - api : nll_loss
   args : (Tensor input, Tensor label, Tensor weight, int64_t ignore_index, str reduction)
   output : Tensor(out), Tensor(total_weight)

paddle/phi/api/yaml/legacy_backward.yaml

@@ -1619,6 +1619,18 @@
     func : multiply_triple_grad
   optional : fwd_grad_grad_x, fwd_grad_grad_y, grad_grad_out_grad
 
+- backward_api : nearest_interp_grad
+  forward : nearest_interp (Tensor x, Tensor out_size, Tensor[] size_tensor, Tensor scale_tensor, str data_layout, int out_d, int out_h, int out_w, float[] scale, str interp_method, bool align_corners, int align_mode) -> Tensor(output)
+  args : (Tensor x, Tensor out_size, Tensor[] size_tensor, Tensor scale_tensor, Tensor output_grad, str data_layout, int out_d, int out_h, int out_w, float[] scale, str interp_method, bool align_corners, int align_mode)
+  output : Tensor(x_grad)
+  infer_meta :
+    func : UnchangedInferMeta
+    param: [x]
+  optional: out_size, size_tensor, scale_tensor
+  kernel :
+    func : nearest_interp_grad
+    data_type : output_grad
+
 - backward_api : nll_loss_grad
   forward : nll_loss (Tensor input, Tensor label, Tensor weight, int64_t ignore_index, str reduction) -> Tensor(out), Tensor(total_weight)
   args : (Tensor input, Tensor label, Tensor weight, Tensor total_weight, Tensor out_grad, int64_t ignore_index, str reduction)

paddle/phi/core/compat/op_utils.h

@@ -73,7 +73,9 @@ const std::unordered_set<std::string> deprecated_op_names(
      "top_k",
      "top_k_grad",
      "linear_interp",
-     "linear_interp_grad"});
+     "linear_interp_grad",
+     "nearest_interp",
+     "nearest_interp_grad"});
 
 class DefaultKernelSignatureMap {
  public:

paddle/phi/kernels/cpu/interpolate_grad_kernel.cc

@@ -1045,7 +1045,7 @@ PD_REGISTER_KERNEL(bilinear_interp_v2_grad,
   kernel->InputAt(2).SetBackend(phi::Backend::ALL_BACKEND);
   kernel->InputAt(3).SetBackend(phi::Backend::ALL_BACKEND);
 }
-PD_REGISTER_KERNEL(nearest_interp_v2_grad,
+PD_REGISTER_KERNEL(nearest_interp_grad,
                    CPU,
                    ALL_LAYOUT,
                    phi::NearestInterpGradKernel,

paddle/phi/kernels/cpu/interpolate_kernel.cc

@@ -1197,7 +1197,7 @@ PD_REGISTER_KERNEL(bilinear_interp_v2,
   kernel->InputAt(2).SetBackend(phi::Backend::ALL_BACKEND);
   kernel->InputAt(3).SetBackend(phi::Backend::ALL_BACKEND);
 }
-PD_REGISTER_KERNEL(nearest_interp_v2,
+PD_REGISTER_KERNEL(nearest_interp,
                    CPU,
                    ALL_LAYOUT,
                    phi::NearestInterpKernel,

paddle/phi/kernels/gpu/interpolate_grad_kernel.cu

@@ -1578,7 +1578,7 @@ PD_REGISTER_KERNEL(bilinear_interp_v2_grad,
   kernel->InputAt(2).SetBackend(phi::Backend::ALL_BACKEND);
   kernel->InputAt(3).SetBackend(phi::Backend::ALL_BACKEND);
 }
-PD_REGISTER_KERNEL(nearest_interp_v2_grad,
+PD_REGISTER_KERNEL(nearest_interp_grad,
                    GPU,
                    ALL_LAYOUT,
                    phi::NearestInterpGradKernel,

paddle/phi/kernels/gpu/interpolate_kernel.cu

@@ -1450,7 +1450,7 @@ PD_REGISTER_KERNEL(bilinear_interp_v2,
   kernel->InputAt(2).SetBackend(phi::Backend::ALL_BACKEND);
   kernel->InputAt(3).SetBackend(phi::Backend::ALL_BACKEND);
 }
-PD_REGISTER_KERNEL(nearest_interp_v2,
+PD_REGISTER_KERNEL(nearest_interp,
                    GPU,
                    ALL_LAYOUT,
                    phi::NearestInterpKernel,

paddle/phi/ops/compat/interpolate_sig.cc

@@ -33,7 +33,7 @@ KernelSignature BilinearInterpOpArgumentMapping(
 
 KernelSignature NearestInterpOpArgumentMapping(
     const ArgumentMappingContext& ctx) {
-  return KernelSignature("nearest_interp_v2",
+  return KernelSignature("nearest_interp",
                          {"X", "OutSize", "SizeTensor", "Scale"},
                          {"data_layout",
                           "out_d",
@@ -107,7 +107,7 @@ KernelSignature BilinearInterpGradOpArgumentMapping(
 
 KernelSignature NearestInterpGradOpArgumentMapping(
     const ArgumentMappingContext& ctx) {
-  return KernelSignature("nearest_interp_v2_grad",
+  return KernelSignature("nearest_interp_grad",
                          {"X", "OutSize", "SizeTensor", "Scale", "Out@GRAD"},
                          {"data_layout",
                           "out_d",
@@ -167,7 +167,10 @@ KernelSignature BicubicInterpGradOpArgumentMapping(
 }  // namespace phi
 
 PD_REGISTER_BASE_KERNEL_NAME(linear_interp_v2, linear_interp);
+PD_REGISTER_BASE_KERNEL_NAME(nearest_interp_v2, nearest_interp);
+
 PD_REGISTER_BASE_KERNEL_NAME(linear_interp_v2_grad, linear_interp_grad);
+PD_REGISTER_BASE_KERNEL_NAME(nearest_interp_v2_grad, nearest_interp_grad);
 
 PD_REGISTER_ARG_MAPPING_FN(bilinear_interp_v2,
                            phi::BilinearInterpOpArgumentMapping);

python/paddle/fluid/tests/unittests/test_nearest_interp_v2_op.py

@@ -22,10 +22,39 @@ import paddle.fluid as fluid
 import paddle.nn as nn
 import paddle
 from paddle.nn.functional import interpolate
+from paddle._C_ops import final_state_nearest_interp
 
 paddle.enable_static()
 
 
+def nearest_interp_test(x,
+                        OutSize=None,
+                        SizeTensor=None,
+                        Scale=None,
+                        data_layout='NCHW',
+                        out_d=-1,
+                        out_h=-1,
+                        out_w=-1,
+                        scale=[],
+                        interp_method='linear',
+                        align_corners=False,
+                        align_mode=1):
+    if isinstance(scale, float) or isinstance(scale, int):
+        scale_list = []
+        for _ in range(len(x.shape) - 2):
+            scale_list.append(scale)
+        scale = list(map(float, scale_list))
+    elif isinstance(scale, list) or isinstance(scale, tuple):
+        scale = list(map(float, scale))
+    if SizeTensor is not None:
+        if not isinstance(SizeTensor, list) and not isinstance(
+                SizeTensor, tuple):
+            SizeTensor = [SizeTensor]
+    return final_state_nearest_interp(x, OutSize, SizeTensor, Scale,
+                                      data_layout, out_d, out_h, out_w, scale,
+                                      interp_method, align_corners, align_mode)
+
+
 def nearest_neighbor_interp_np(X,
                                out_h,
                                out_w,
@@ -160,6 +189,7 @@ def nearest_neighbor_interp3d_np(X,
 class TestNearestInterpOp(OpTest):
 
     def setUp(self):
+        self.python_api = nearest_interp_test
         self.out_size = None
         self.actual_shape = None
         self.data_layout = 'NCHW'
@@ -254,17 +284,17 @@ class TestNearestInterpOp(OpTest):
         self.outputs = {'Out': output_np}
 
     def test_check_output(self):
-        self.check_output()
+        self.check_output(check_eager=True)
 
     def test_check_grad(self):
-        self.check_grad(['X'], 'Out', in_place=True)
+        self.check_grad(['X'], 'Out', in_place=True, check_eager=True)
 
     def init_test_case(self):
         self.interp_method = 'nearest'
         self.input_shape = [2, 3, 4, 5]
         self.out_h = 2
         self.out_w = 2
-        self.scale = 0.
+        self.scale = []
         self.out_size = np.array([3, 3]).astype("int32")
         self.align_corners = True
 
@@ -277,7 +307,7 @@ class TestNearestNeighborInterpCase1(TestNearestInterpOp):
         self.out_d = 1
         self.out_h = 1
         self.out_w = 1
-        self.scale = 0.
+        self.scale = []
         self.align_corners = True
 
 
@@ -288,7 +318,7 @@ class TestNearestNeighborInterpCase2(TestNearestInterpOp):
         self.input_shape = [3, 3, 9, 6]
         self.out_h = 12
         self.out_w = 12
-        self.scale = 0.
+        self.scale = []
         self.align_corners = True
 
 
@@ -299,7 +329,7 @@ class TestNearestNeighborInterpCase3(TestNearestInterpOp):
         self.input_shape = [1, 1, 32, 64]
         self.out_h = 64
         self.out_w = 32
-        self.scale = 0.
+        self.scale = []
         self.align_corners = True
 
 
@@ -310,7 +340,7 @@ class TestNearestNeighborInterpCase4(TestNearestInterpOp):
         self.input_shape = [4, 1, 7, 8]
         self.out_h = 1
         self.out_w = 1
-        self.scale = 0.
+        self.scale = []
         self.out_size = np.array([2, 2]).astype("int32")
         self.align_corners = True
 
@@ -322,7 +352,7 @@ class TestNearestNeighborInterpCase5(TestNearestInterpOp):
         self.input_shape = [3, 3, 9, 6]
         self.out_h = 12
         self.out_w = 12
-        self.scale = 0.
+        self.scale = []
         self.out_size = np.array([11, 11]).astype("int32")
         self.align_corners = True
 
@@ -334,7 +364,7 @@ class TestNearestNeighborInterpCase6(TestNearestInterpOp):
         self.input_shape = [1, 1, 32, 64]
         self.out_h = 64
         self.out_w = 32
-        self.scale = 0.
+        self.scale = []
         self.out_size = np.array([65, 129]).astype("int32")
         self.align_corners = True
 
@@ -346,7 +376,7 @@ class TestNearestNeighborInterpSame(TestNearestInterpOp):
         self.input_shape = [2, 3, 32, 64]
         self.out_h = 32
         self.out_w = 64
-        self.scale = 0.
+        self.scale = []
         self.align_corners = True
 
 
@@ -357,7 +387,7 @@ class TestNearestNeighborInterpActualShape(TestNearestInterpOp):
         self.input_shape = [3, 2, 32, 16]
         self.out_h = 64
         self.out_w = 32
-        self.scale = 0.
+        self.scale = []
         self.out_size = np.array([66, 40]).astype("int32")
         self.align_corners = True
 
@@ -369,7 +399,7 @@ class TestNearestNeighborInterpDataLayout(TestNearestInterpOp):
         self.input_shape = [2, 4, 4, 5]
         self.out_h = 2
         self.out_w = 2
-        self.scale = 0.
+        self.scale = []
         self.out_size = np.array([3, 8]).astype("int32")
         self.align_corners = True
         self.data_layout = "NHWC"
@@ -378,6 +408,7 @@ class TestNearestNeighborInterpDataLayout(TestNearestInterpOp):
 class TestNearestInterpOpUint8(OpTest):
 
     def setUp(self):
+        self.python_api = nearest_interp_test
         self.out_size = None
         self.actual_shape = None
         self.init_test_case()
@@ -422,14 +453,16 @@ class TestNearestInterpOpUint8(OpTest):
         self.outputs = {'Out': output_np}
 
     def test_check_output(self):
-        self.check_output_with_place(place=core.CPUPlace(), atol=1)
+        self.check_output_with_place(place=core.CPUPlace(),
+                                     atol=1,
+                                     check_eager=True)
 
     def init_test_case(self):
         self.interp_method = 'nearest'
         self.input_shape = [1, 3, 9, 6]
         self.out_h = 10
         self.out_w = 9
-        self.scale = 0.
+        self.scale = []
         self.align_corners = True
 
 
@@ -440,7 +473,7 @@ class TestNearestNeighborInterpCase1Uint8(TestNearestInterpOpUint8):
         self.input_shape = [2, 3, 32, 64]
         self.out_h = 80
         self.out_w = 40
-        self.scale = 0.
+        self.scale = []
         self.align_corners = True
 
 
@@ -451,7 +484,7 @@ class TestNearestNeighborInterpCase2Uint8(TestNearestInterpOpUint8):
         self.input_shape = [4, 1, 7, 8]
         self.out_h = 5
         self.out_w = 13
-        self.scale = 0.
+        self.scale = []
         self.out_size = np.array([6, 15]).astype("int32")
         self.align_corners = True
 
@@ -514,6 +547,7 @@ class TestNearestNeighbor3DInterp(TestNearestInterpOp):
 class TestNearestInterpOp_attr_tensor(OpTest):
 
     def setUp(self):
+        self.python_api = nearest_interp_test
         self.out_size = None
         self.actual_shape = None
         self.init_test_case()
@@ -569,17 +603,17 @@ class TestNearestInterpOp_attr_tensor(OpTest):
         self.outputs = {'Out': output_np}
 
     def test_check_output(self):
-        self.check_output()
+        self.check_output(check_eager=True)
 
     def test_check_grad(self):
-        self.check_grad(['X'], 'Out', in_place=True)
+        self.check_grad(['X'], 'Out', in_place=True, check_eager=True)
 
     def init_test_case(self):
         self.interp_method = 'nearest'
         self.input_shape = [2, 5, 4, 4]
         self.out_h = 3
         self.out_w = 3
-        self.scale = 0.
+        self.scale = []
         self.out_size = [3, 3]
         self.align_corners = True
 
@@ -592,7 +626,7 @@ class TestNearestInterp_attr_tensor_Case1(TestNearestInterpOp_attr_tensor):
         self.input_shape = [3, 3, 9, 6]
         self.out_h = 12
         self.out_w = 12
-        self.scale = 0.
+        self.scale = []
         self.out_size = [8, 12]
         self.align_corners = True
 
@@ -605,7 +639,7 @@ class TestNearestInterp_attr_tensor_Case2(TestNearestInterpOp_attr_tensor):
         self.input_shape = [3, 2, 32, 16]
         self.out_h = 64
         self.out_w = 32
-        self.scale = 0.
+        self.scale = []
         self.out_size = np.array([66, 40]).astype("int32")
         self.align_corners = True
         self.shape_by_1Dtensor = True

python/paddle/nn/functional/common.py

@@ -607,6 +607,16 @@ def interpolate(x,
         elif resample_type == "trilinear":
             out = _C_ops.trilinear_interp_v2(x, *dy_attr)
         elif resample_type == "nearest":
+            if in_dygraph_mode():
+                out = _C_ops.final_state_nearest_interp(
+                    x, inputs['OutSize'] if 'OutSize' in inputs else None,
+                    inputs['SizeTensor'] if 'SizeTensor' in inputs else None,
+                    inputs['Scale'] if 'Scale' in inputs else None,
+                    attrs['data_layout'], attrs['out_d'], attrs['out_h'],
+                    attrs['out_w'], attrs['scale'] if 'scale' in attrs else [],
+                    attrs['interp_method'], attrs['align_corners'],
+                    attrs['align_mode'])
+            else:
                 out = _C_ops.nearest_interp_v2(x, *dy_attr)
         elif resample_type == "bicubic":
             out = _C_ops.bicubic_interp_v2(x, *dy_attr)