inplace_utils.py 2.4 KB
Newer Older
1
# Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
2 3 4 5 6 7 8 9 10 11 12 13 14 15
#
# 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.

import warnings
16 17

import paddle  # noqa: F401
18
from paddle.fluid.wrapped_decorator import wrap_decorator
19
from paddle.framework import in_dynamic_mode
20 21


W
wanghuancoder 已提交
22
# NOTE(pangyoki): The Inplace APIs with underline(`_`) is only valid for the method of calling `_C_ops`
23
# in dygraph mode. If static graph mode is used, the inplace mechanism will not be used, and the static method
24
# of the original API will be called.
25 26
# NOTE(GGBond8488): Simply run the original version of the API under the static graph mode has a low
# probability that the result is inconsistent with the dynamic graph.
27 28
def _inplace_apis_in_dygraph_only_(func):
    def __impl__(*args, **kwargs):
29
        if not in_dynamic_mode():
30 31
            origin_api_name = func.__name__[:-1]
            warnings.warn(
32
                "In static graph mode, {}() is the same as {}() and does not perform inplace operation.".format(
33 34 35
                    func.__name__, origin_api_name
                )
            )
W
wanghuancoder 已提交
36 37 38 39 40 41 42 43 44
            from ..fluid.dygraph.base import in_declarative_mode

            if in_declarative_mode():
                for arg in args:
                    if hasattr(arg, "is_view_var") and arg.is_view_var:
                        raise ValueError(
                            f'Sorry about what\'s happend. In to_static mode, {func.__name__}\'s output variable {arg.name} is a viewed Tensor in dygraph. This will result in inconsistent calculation behavior between dynamic and static graphs. You mast find the location of the strided API be called, and call {arg.name} = {arg.name}.assign().'
                        )

45
            origin_func = f"{func.__module__}.{origin_api_name}"
46 47 48 49 50 51 52
            return eval(origin_func)(*args, **kwargs)
        return func(*args, **kwargs)

    return __impl__


inplace_apis_in_dygraph_only = wrap_decorator(_inplace_apis_in_dygraph_only_)