!1729 fix bprop of ExtractImagePatches

Merge pull request !1729 from zhangbuxue/fix_bprop_of_ExtractImagePatches

mindspore/nn/layer/basic.py

@@ -482,7 +482,7 @@ class Unfold(Cell):
 
     Inputs:
         - **input_x** (Tensor) - A 4-D tensor whose shape is [in_batch, in_depth, in_row, in_col] and
-          data type is int8, float16, uint8.
+          data type is number.
 
     Outputs:
         Tensor, a 4-D tensor whose data type is same as 'input_x',

mindspore/ops/_grad/grad_nn_ops.py

@@ -14,13 +14,12 @@
 # ============================================================================
 
 """Define the grad rules of neural network related operations."""
-from mindspore.common import dtype as mstype
+from .grad_base import bprop_getters
 from .. import functional as F
 from .. import operations as P
+from ..composite.multitype_ops.zeros_like_impl import zeros_like
 from ..operations import _grad_ops as G
 from ..operations import _inner_ops as inner
-from ..composite.multitype_ops.zeros_like_impl import zeros_like
-from .grad_base import bprop_getters
 from ... import context
 
 
@@ -73,7 +72,7 @@ def get_bprop_extract_image_patches(self):
     slice_op = P.Slice()
     transpose = P.Transpose()
     matmul = P.MatMul()
-    cast = P.Cast()
+
     _, ksizes_row, ksizes_col, _ = self.ksizes
 
     def bprop(x, out, dout):
@@ -82,16 +81,13 @@ def get_bprop_extract_image_patches(self):
         x_indices_num = x_row * x_col + 1
         x_idx = F.tuple_to_array(range(1, x_indices_num))
         x_idx = reshape(x_idx, (1, x_row, x_col, 1))
-        x_idx = cast(x_idx, mstype.float16)
         x_idx_patch = extract_image_patches(x_idx)
-        x_idx_patch = transpose(x_idx_patch, (0, 3, 1, 2))
-        x_idx_patch = cast(x_idx_patch, mstype.int32)
 
         out_shape = get_shape(out)
         _, out_row, out_col, _ = out_shape
         out_indices_num = out_row * out_col * ksizes_row * ksizes_col
         out_idx = F.tuple_to_array(range(out_indices_num))
-        out_idx = reshape(out_idx, (1, ksizes_row * ksizes_col, out_row, out_col))
+        out_idx = reshape(out_idx, (1, out_row, out_col, ksizes_row * ksizes_col))
 
         idx_tensor = concat((expand_dims(x_idx_patch, -1), expand_dims(out_idx, -1)))
         idx_tensor = reshape(idx_tensor, (-1, 2))

mindspore/ops/operations/_inner_ops.py

@@ -41,7 +41,7 @@ class ExtractImagePatches(PrimitiveWithInfer):
 
     Inputs:
         - **input_x** (Tensor) - A 4-D tensor whose shape is [in_batch, in_row, in_col, in_depth] and
-          data type is int8, float16, uint8.
+          data type is number.
 
     Outputs:
         Tensor, a 4-D tensor whose data type is same as 'input_x',
@@ -94,5 +94,5 @@ class ExtractImagePatches(PrimitiveWithInfer):
 
     def infer_dtype(self, input_x):
         """infer dtype"""
-        validator.check_tensor_type_same({"input_x": input_x}, (mstype.int8, mstype.float16, mstype.float32), self.name)
+        validator.check_tensor_type_same({"input_x": input_x}, mstype.number_type, self.name)
         return input_x