forked from huawei/mindspore2022
136 lines
5.3 KiB
Python
136 lines
5.3 KiB
Python
# Copyright 2020-2021 Huawei Technologies Co., Ltd
|
|
#
|
|
# Licensed under the Apache License, Version 2.0 (the "License");
|
|
# you may not use this file except in compliance with the License.
|
|
# You may obtain a copy of the License at
|
|
#
|
|
# http://www.apache.org/licenses/LICENSE-2.0
|
|
#
|
|
# Unless required by applicable law or agreed to in writing, software
|
|
# distributed under the License is distributed on an "AS IS" BASIS,
|
|
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
# See the License for the specific language governing permissions and
|
|
# limitations under the License.
|
|
# ============================================================================
|
|
|
|
"""utils for operator"""
|
|
|
|
from mindspore.common.tensor import Tensor
|
|
from ..._checkparam import Validator as validator
|
|
from ..._checkparam import Rel
|
|
from ...common import dtype as mstype
|
|
from ..primitive import constexpr
|
|
|
|
|
|
def get_broadcast_shape(x_shape, y_shape, prim_name, shape_type="", arg_name1="x", arg_name2="y"):
|
|
"""
|
|
Doing broadcast between tensor x and tensor y.
|
|
|
|
Args:
|
|
x_shape (list): The shape of tensor x.
|
|
y_shape (list): The shape of tensor y.
|
|
prim_name (str): Primitive name.
|
|
shape_type (str): The type of shape, optional values are "", "min_shape" and "max_shape".
|
|
arg_name1 (str): The arg name of x_shape.
|
|
arg_name2 (str): The arg name of y_shape.
|
|
|
|
Returns:
|
|
List, the shape that broadcast between tensor x and tensor y.
|
|
|
|
Raises:
|
|
ValueError: If tensor x and tensor y are not equal and couldn't broadcast.
|
|
|
|
Examples:
|
|
>>> x_shape = [1, 2, 3]
|
|
>>> y_shape = [1, 2]
|
|
>>> broadcast_shape = get_broadcast_shape(x_shape, y_shape)
|
|
"""
|
|
if x_shape == y_shape:
|
|
return x_shape
|
|
x_len = len(x_shape)
|
|
y_len = len(y_shape)
|
|
length = x_len if x_len < y_len else y_len
|
|
broadcast_shape_back = []
|
|
|
|
for i in range(-length, 0):
|
|
if x_shape[i] == 1:
|
|
broadcast_shape_back.append(y_shape[i])
|
|
elif y_shape[i] == 1:
|
|
broadcast_shape_back.append(x_shape[i])
|
|
elif x_shape[i] == y_shape[i]:
|
|
broadcast_shape_back.append(x_shape[i])
|
|
elif x_shape[i] == -1 or y_shape[i] == -1:
|
|
broadcast_shape_back.append(-1)
|
|
else:
|
|
if shape_type == "min_shape":
|
|
broadcast_shape_back.append(max(x_shape[i], y_shape[i]))
|
|
elif shape_type == "max_shape":
|
|
broadcast_shape_back.append(min(x_shape[i], y_shape[i]))
|
|
else:
|
|
raise ValueError(f"For '{prim_name}', '{arg_name1}'.shape and '{arg_name2}'.shape are supposed "
|
|
f"to broadcast, where broadcast means that '{arg_name1}'.shape[i] = 1 or -1 "
|
|
f"or '{arg_name2}'.shape[i] = 1 or -1 "
|
|
f"or '{arg_name1}'.shape[i] = '{arg_name2}'.shape[i]', "
|
|
f"but now '{arg_name1}'.shape and '{arg_name2}'.shape can not broadcast, "
|
|
f"got 'i': {i}, '{arg_name1}'.shape: {x_shape}, '{arg_name2}'.shape: {y_shape}.")
|
|
|
|
broadcast_shape_front = y_shape[0: y_len - length] if length == x_len else x_shape[0: x_len - length]
|
|
broadcast_shape = list(broadcast_shape_front) + broadcast_shape_back
|
|
return broadcast_shape
|
|
|
|
|
|
def get_concat_offset(x_shp, x_type, axis, prim_name):
|
|
"""for concat and concatoffset check args and compute offset"""
|
|
validator.check_value_type("shape", x_shp, [tuple, list], prim_name)
|
|
validator.check_positive_int(len(x_shp), "input_x rank", prim_name)
|
|
validator.check_subclass("shape0", x_type[0], mstype.tensor, prim_name)
|
|
validator.check_positive_int(len(x_shp[0]), "len of x_shp[0]", prim_name)
|
|
rank_base = len(x_shp[0])
|
|
validator.check_int_range(axis, -rank_base - 1, rank_base, Rel.INC_BOTH, 'axis', prim_name)
|
|
if axis < 0:
|
|
axis = axis + rank_base
|
|
all_shp = x_shp[0][axis]
|
|
offset = [0]
|
|
for i in range(1, len(x_shp)):
|
|
v = x_shp[i]
|
|
validator.check('len of x_shp[%d]' % i, len(v), 'len of x_shp[0]', len(x_shp[0]), Rel.EQ, prim_name)
|
|
validator.check('x_type[%d]' % i, x_type[i], 'x_type[0]', x_type[0], Rel.EQ, prim_name)
|
|
for j in range(rank_base):
|
|
if j != axis and v[j] != x_shp[0][j]:
|
|
raise ValueError(f"The shape of the two input elements of the Concat operator do not match:"
|
|
f"shape[0] = {x_shp[0]} and shape[1] = {x_shp[1]}.")
|
|
offset.append(all_shp)
|
|
if all_shp == -1 or v[axis] == -1:
|
|
all_shp = -1
|
|
else:
|
|
all_shp += v[axis]
|
|
return offset, all_shp, axis
|
|
|
|
|
|
@constexpr
|
|
def range_op(start, limit, delta, dtype):
|
|
"""helper function to get tensor in specified range."""
|
|
output_tensor = Tensor(list(range(start, limit, delta)), dtype)
|
|
return output_tensor
|
|
|
|
|
|
@constexpr
|
|
def get_1d_shape(in_shape):
|
|
"""helper function to get 1d shape."""
|
|
out_shape = 1
|
|
for i in in_shape:
|
|
out_shape *= i
|
|
return (out_shape,)
|
|
|
|
|
|
@constexpr
|
|
def generate_shape_index(out_shape, indices_shape, axis):
|
|
out_rank = len(out_shape)
|
|
ind_rank = len(indices_shape)
|
|
if axis < 0:
|
|
axis += out_rank - ind_rank + 1
|
|
perm_part1 = tuple(range(axis, axis + ind_rank))
|
|
index = tuple(range(out_rank))
|
|
perm = perm_part1 + index[:axis] + index[axis + ind_rank:]
|
|
return perm
|