# Copyright 2020-2024 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.
# ============================================================================
"""cell"""
from __future__ import absolute_import
import gc
import inspect
import os
import time
from collections import OrderedDict
import numpy
from mindspore._checkparam import args_type_check, check_hook_fn
from mindspore.common._auto_dynamic import is_auto_dynamic, convert_inputs_to_dynamic
from mindspore import log as logger
from mindspore.common.parameter import PARAMETER_NAME_DEFAULT
from mindspore.common.hook_handle import HookHandle
from mindspore.context import ParallelMode
from mindspore import context
from mindspore._c_expression import init_pipeline, update_func_graph_hyper_params, Cell_, FuncGraph, MixedPrecisionType
from mindspore import _checkparam as Validator
from mindspore.common import dtype as mstype
from mindspore.common.api import _cell_graph_executor, _pynative_executor, _get_args_for_run, cells_compile_cache, _no_grad
from mindspore.common.api import _generate_branch_control_input, _convert_python_data, _get_args_for_run_predict
from mindspore.common.api import _process_dyn_args, _generate_dyn_compile_args
from mindspore.common.parameter import Parameter, ParameterTuple
from mindspore.common.tensor import Tensor
from mindspore.ops.operations import Cast
from mindspore.ops.primitive import Primitive
from mindspore.ops.operations import _inner_ops as inner
from mindspore.parallel.shard import Shard
from mindspore._check_jit_forbidden_api import jit_forbidden_register
from mindspore.common._decorator import deprecated
from mindspore.common._register_for_recompute import recompute_registry
[docs]class Cell(Cell_):
"""
The basic building block of neural networks in MindSpore. The model or neural network layer should inherit this
base class.
Layers in `mindspore.nn` are also the subclass of Cell, such as :class:`mindspore.nn.Conv2d`,
and :class:`mindspore.nn.ReLU`, etc. Cell will be compiled into a calculation
graph in GRAPH_MODE (static graph mode) and used as the basic module of neural networks in
PYNATIVE_MODE (dynamic graph mode).
.. note::
Cell is the inference mode by default. For a class that inherits a Cell,
if the training and inference have different structures, the subclass performs the inference branch by default.
To set the training mode, refer to `mindspore.nn.Cell.set_train` .
.. warning::
In the subclass of Cell, it's not allowed to define a method named 'cast' and not allowed to define an attribute
named 'phase' or 'cells', otherwise, an error will be raised.
Args:
auto_prefix (bool, optional): Whether to automatically generate NameSpace for Cell and its child cells. It also
affects the names of parameters in the `Cell`. If set to ``True`` , the parameter name will be
automatically prefixed, otherwise not. In general, the backbone network should be set to
``True`` , otherwise the duplicate name problem will appear. The cell to train the backbone
network, such as optimizer and :class:`mindspore.nn.TrainOneStepCell`, should be set to
``False`` , otherwise the parameter name in backbone will be changed by mistake.
Default: ``True`` .
flags (dict, optional): Network configuration information, currently it is used for the binding of network
and dataset. Users can also customize network attributes by this parameter. Default: ``None`` .
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore.nn as nn
>>> from mindspore import ops
>>> class MyCell(nn.Cell):
... def __init__(self, forward_net):
... super(MyCell, self).__init__(auto_prefix=False)
... self.net = forward_net
... self.relu = ops.ReLU()
...
... def construct(self, x):
... y = self.net(x)
... return self.relu(y)
>>>
>>> inner_net = nn.Conv2d(120, 240, 4, has_bias=False, weight_init='normal')
>>> my_net = MyCell(inner_net)
>>> print(my_net.trainable_params())
... # If the 'auto_prefix' set to True or not set when call the '__init__' method of the parent class,
... # the parameter's name will be 'net.weight'.
[Parameter (name=weight, shape=(240, 120, 4, 4), dtype=Float32, requires_grad=True)]
"""
IGNORE_LIST = ['_scope', '_cell_init_args', '_auto_prefix', '_cells', '_params', '_create_time',
'_func_graph_flags', '_parameter_layout_dict', '_params_list', '_phase', '_bprop_debug',
'_forward_pre_hook', '_forward_hook', '_backward_pre_hook', '_backward_hook',
'_cell_backward_pre_hook', '_cell_backward_hook', '_is_run', '_param_prefix',
'_attr_synced', 'pynative', 'requires_grad', 'cell_type']
total_instance_count = 0
def __init__(self, auto_prefix=True, flags=None):
Cell_.__init__(self, self._cell_tag)
Cell.total_instance_count += 1
self.instance_count = Cell.total_instance_count
self._params = OrderedDict()
self._cells = OrderedDict()
self._params_list = OrderedDict()
self._primitives = OrderedDict()
self.training = False
self.requires_grad = False
self.pynative = False
self._attr_synced = False
self._param_prefix = ''
self._auto_prefix = auto_prefix
self._scope = None
self._phase = 'train'
self._parameter_layout_dict = {}
self._parallel_parameter_name_list = ()
self._parallel_parameter_merge_net_dict = {}
self._create_time = int(time.time() * 1e9)
self.arguments_key = ""
self.compile_cache = set()
self.phase_cache = dict()
cells_compile_cache[id(self)] = self.compile_cache
self.parameter_broadcast_done = False
self._id = 1
self.exist_names = set("")
self.exist_objs = set()
self._recompute_cell = None
self.mixed_precision_type = None
self.sig = inspect.signature(self.construct)
init_pipeline()
# call gc to release GE session resources used by non-used cell objects
if os.getenv('GC_COLLECT_IN_CELL') == '1':
gc.collect()
if flags:
self.add_flags(**flags)
self._bprop_debug = False
# hook
self._forward_pre_hook = OrderedDict()
self._forward_hook = OrderedDict()
self._backward_pre_hook = OrderedDict()
self._cell_backward_pre_hook = None
self._backward_hook = OrderedDict()
self._cell_backward_hook = None
self._is_recursion_hook = False
self.cell_type = None
self.cast = Cast()
self._has_config_recompute = False
self._user_parameters = []
self._dynamic_shape_inputs = None
self._compile_args = None
self.saved_dynamic_shape = None
self._jit_config_dict = dict()
self.grad_ops_label = False
self.ge_sync_data = False
self._is_check_and_refresh = False
self._amp_level = ""
self._init_flag = False
self._shard_fn = None
self.has_bprop = False
if hasattr(self, "bprop"):
self.has_bprop = True
def __getstate__(self):
base = Cell_.__getstate__(self)
return base, self.__dict__
def __setstate__(self, state):
base, dict_ = state
Cell_.__setstate__(self, base)
self.__dict__ = dict_
self._attr_synced = False
def __bool__(self):
return True
@property
def _cell_tag(self):
# `<class 'xxxxxxx'>` to `xxxxxxx`
return str(self.__class__)[8:-2]
@property
def create_time(self):
return self._create_time
@property
def cell_init_args(self):
return self._cell_init_args
@property
def param_prefix(self):
"""
Param prefix is the prefix of current cell's direct child parameter.
Examples:
>>> import mindspore as ms
>>> from mindspore import Tensor, nn
...
>>> class Net(nn.Cell):
... def __init__(self):
... super(Net, self).__init__()
... self.dense = nn.Dense(2, 2)
...
... def construct(self, x):
... x = self.dense(x)
... return x
>>> net = Net()
>>> net.update_cell_prefix()
>>> print(net.dense.param_prefix)
dense
"""
return self._param_prefix
@property
def bprop_debug(self):
"""
Get whether cell custom bprop debug is enabled.
Tutorial Examples:
- `Custom Neural Network Layers - Custom Cell Reverse
<https://mindspore.cn/docs/en/r2.4.0/model_train/custom_program/network_custom.html
#custom-cell-reverse>`_
"""
return self._bprop_debug
@bprop_debug.setter
def bprop_debug(self, value):
"""
Set whether to enable cell custom bprop debug.
Note:
When bprop is defined in cell, the bprop function will be executed
in python interpreter when bprop debug is true, and will be parsed
and add to graph when bprop debug is false.
Args:
value (bool): Specifies whether to enable bprop debug. Default: ``False``.
"""
if not isinstance(value, bool):
raise TypeError(f"For 'Cell', the property 'bprop_debug' must be bool type, but got type {type(value)}.")
self._bprop_debug = value
[docs] def update_cell_prefix(self):
"""
Update the `param_prefix` of all child cells.
After being invoked, it can get all the cell's children's name prefix by '_param_prefix'.
"""
cells_name = self.cells_and_names()
for cell_name, cell in cells_name:
cell._param_prefix = cell_name
[docs] def update_cell_type(self, cell_type):
"""
The current cell type is updated when a quantization aware training network is encountered.
After being invoked, it can set the cell type to 'cell_type'.
Args:
cell_type(str): The type of cell to be updated, cell_type can be "quant" or "second-order".
"""
self.cell_type = cell_type
@cell_init_args.setter
def cell_init_args(self, value):
if not isinstance(value, str):
raise TypeError(f"For 'Cell', the property 'cell_init_args' must be string type, "
f"but got type {type(value)}.")
self._cell_init_args = value
@property
def phase(self):
return self._phase
@phase.setter
def phase(self, value):
if not isinstance(value, str):
raise TypeError(f"For 'Cell', the property 'phase' must be string type, but got type {type(value)}.")
self._phase = value
@property
def parameter_layout_dict(self):
"""
`parameter_layout_dict` represents the tensor layout of a parameter, which is inferred by shard strategy and
distributed operator information.
"""
return self._parameter_layout_dict
@property
def cls_name(self):
return self.__class__.__name__
@parameter_layout_dict.setter
def parameter_layout_dict(self, value):
if not isinstance(value, dict):
raise TypeError(f"For 'Cell', the property 'parameter_layout_dict' must be dict type, "
f"but got type {type(value)}.")
self._parameter_layout_dict = value
@property
def parallel_parameter_name_list(self):
return self._parallel_parameter_name_list
@parallel_parameter_name_list.setter
def parallel_parameter_name_list(self, value):
if not isinstance(value, list):
raise TypeError(f"For 'Cell', the property 'parallel_parameter_name_list' must be list type, "
f"but got type {type(value)}.")
self._parallel_parameter_name_list = value
@property
def pipeline_stage(self):
"""
`pipeline_stage` represents the pipeline stage of current Cell.
"""
return self._pipeline_stage
@pipeline_stage.setter
def pipeline_stage(self, value):
"""
Set the `pipeline_stage` of a Cell.
Args:
value (int): The pipeline stage of a parameter.
Raises:
TypeError: If `value` is not int type or is a bool type.
ValueError: If `value` is not a positive integer.
"""
if not isinstance(value, int) or isinstance(value, bool):
raise TypeError("For 'Cell', the property 'pipeline_stage' "
"must be int type, but got type : {}".format(type(value)))
if value < 0:
raise ValueError("For 'Cell', the property 'pipeline_stage' "
"can not be less than 0, but got {}".format(value))
self._pipeline_stage = value
for item in self.trainable_params():
item.add_pipeline_stage(value)
@property
def pipeline_segment(self):
return self._pipeline_segment
@pipeline_segment.setter
def pipeline_segment(self, value):
if not isinstance(value, int) or isinstance(value, bool):
raise TypeError("For 'context.set_auto_parallel_context', the argument 'pipeline_stages' "
"must be int type, but got type : {}".format(type(value)))
if value < 0:
raise ValueError("For 'context.set_auto_parallel_context', the argument 'pipeline_stages' "
"can not be less than 0, but got {}".format(value))
self._pipeline_segment = value
@property
def parallel_parameter_merge_net_dict(self):
return self._parallel_parameter_merge_net_dict
@parallel_parameter_merge_net_dict.setter
def parallel_parameter_merge_net_dict(self, value):
if not isinstance(value, dict):
raise TypeError(f"For 'Cell', the property 'parallel_parameter_merge_net_dict' must be dict type, "
f"but got type {type(value)}.")
self._parallel_parameter_merge_net_dict = value
@property
def jit_config_dict(self):
return self._jit_config_dict
@property
def enable_backward_hook(self):
return self._enable_backward_hook
[docs] def get_func_graph_proto(self):
"""Return graph binary proto."""
exec_id = ".".join([self.phase, str(self.create_time), str(id(self))])
return _cell_graph_executor._get_func_graph_proto(self, exec_id, "anf_ir", True)
def __getattr__(self, name):
if '_params' in self.__dict__:
params = self.__dict__['_params']
if name in params:
return params[name]
if '_cells' in self.__dict__:
cells = self.__dict__['_cells']
if name in cells:
return cells[name]
if '_params_list' in self.__dict__:
params_list = self.__dict__['_params_list']
if name in params_list:
return params_list[name]
raise AttributeError("The '{}' object has no attribute '{}'.".format(type(self).__name__, name))
def __del__(self):
if isinstance(cells_compile_cache, dict):
# while deepcopy a cell instance, the copied cell instance can't be added to cells_compile_cache
# here using pop(id(self), None) to avoid KeyError exception
cells_compile_cache.pop(id(self), None)
if hasattr(self, "compile_cache") and self.compile_cache:
_cell_graph_executor.del_net_res(self, self.compile_cache)
Cell.total_instance_count -= 1
def __delattr__(self, name):
if name in self._params:
del self._params[name]
elif name in self._cells:
del self._cells[name]
elif '_params_list' in self.__dict__ and name in self._params_list:
del self._params_list[name]
else:
object.__delattr__(self, name)
self._attr_synced = False
def _cast_mixed_precision_inputs(self, inputs, dst_type):
"""Cast input for mixed precision"""
res = list()
for item in inputs:
if isinstance(item, tuple):
res.append(self._cast_mixed_precision_inputs(item, dst_type))
elif isinstance(item, float):
res.append(self.cast(item, dst_type))
elif hasattr(item, "dtype") and item.dtype in \
{mstype.float16, mstype.float32, mstype.float64, mstype.bfloat16} and item.dtype != dst_type:
res.append(self.cast(item, dst_type))
else:
res.append(item)
return tuple(res)
def _do_parameter_broadcast(self):
if context.get_auto_parallel_context("parallel_mode") == ParallelMode.DATA_PARALLEL:
if not self.parameter_broadcast_done:
_pynative_executor.parameter_broadcast(self, self.phase)
self.parameter_broadcast_done = True
[docs] def run_construct(self, cast_inputs, kwargs):
"""
Run the construct function.
Note:
This function will be removed in a future version. It is not recommended to call this function.
Args:
cast_inputs (tuple): The input objects of Cell.
kwargs (dict): Provide keyword arguments.
Returns:
output, the output object of Cell.
"""
logger.warning(f"The 'run_construct' function of '{self.cls_name}' will be removed in a future version. "
f"Calling this function is not recommended.")
output = self._run_construct(cast_inputs, kwargs)
return output
def _run_construct(self, *inputs, **kwargs):
"""Run the construct function"""
if self._forward_pre_hook:
inputs = self._run_forward_pre_hook(inputs)
if self._backward_hook:
output = self._backward_hook_construct(*inputs, **kwargs)
elif self._shard_fn is not None:
output = self._shard_fn(*inputs, **kwargs)
elif self._recompute_cell is not None:
output = self._recompute_cell(*inputs, **kwargs)
elif self.has_bprop and _pynative_executor.requires_grad():
output = self._call_custom_bprop(*inputs, **kwargs)
else:
output = self.construct(*inputs, **kwargs)
if self._forward_hook:
output = self._run_forward_hook(inputs, output)
if self._backward_pre_hook:
output = self._run_backward_pre_hook(output)
return output
def _check_construct_args(self, *args):
"""Check the args needed by the function construct"""
positional_args = 0
default_args = 0
has_var = False
for value in inspect.signature(self.construct).parameters.values():
if value.kind is inspect.Parameter.VAR_POSITIONAL or value.kind is inspect.Parameter.VAR_KEYWORD:
has_var = True
if value.kind is inspect.Parameter.POSITIONAL_OR_KEYWORD:
if value.default is inspect.Parameter.empty:
positional_args += 1
else:
default_args += 1
if has_var:
return
if len(args) < positional_args:
raise TypeError(f"For 'Cell', the function construct requires {positional_args} positional argument, "
f"but got {len(args)}. When using set_inputs, please make sure that all networks "
f"and loss functions are configured with set_inputs.")
if len(args) > positional_args + default_args:
construct_inputs_names = self.construct.__code__.co_varnames
if 'self' not in construct_inputs_names:
raise TypeError(f"For 'Cell', the method 'construct' must have parameter 'self'. ")
raise TypeError(f"For 'Cell', the function construct requires {positional_args} positional argument and "
f"{default_args} default argument, total {positional_args + default_args}, "
f"but got {len(args)}.")
def _hook_fn_registered(self):
'''Hook function in graph mode'''
# Check super().__init__() in graph mode.
try:
if self._forward_pre_hook or self._forward_hook or self._backward_pre_hook or self._backward_hook:
return True
except AttributeError as e:
raise AttributeError(f"The '{type(self).__name__}' object does not inherit attribute from 'cell'. "
f"Please use 'super().__init__()'.") from e
if not self._is_recursion_hook:
self._is_recursion_hook = True
for cell in self.cells():
if cell._hook_fn_registered():
return True
return False
def _get_prims_recursively(self):
all_prims = list()
for _, value in self._primitives.items():
if value:
all_prims.append(value)
for cell in self.cells():
all_prims.extend(cell._get_prims_recursively())
return all_prims
[docs] def set_data_parallel(self):
"""
For all primitive ops in this cell(including ops of cells that wrapped by this cell),
if parallel strategy is not specified, then instead of auto-searching, data parallel
strategy will be generated for those primitive ops.
Note:
Only effective while using auto_parallel_context = ParallelMode.AUTO_PARALLEL under graph mode.
Examples:
>>> import mindspore.nn as nn
>>> net = nn.Dense(3, 4)
>>> net.set_data_parallel()
"""
if context._get_mode() == context.PYNATIVE_MODE:
raise ValueError("set_data_parallel: does not support PyNative mode.")
all_prims = self._get_prims_recursively()
for prim in all_prims:
prim.add_prim_attr("strategy_gen_mode", "data_parallel")
[docs] def shard(self, in_strategy, out_strategy=None, parameter_plan=None, device="Ascend", level=0):
"""
Defining the input and output layouts of this cell and the parallel strategies of remaining ops will be
generated by sharding propagation. In PyNative mode, use this method to specify a Cell for distributed
execution in graph mode. In Graph mode, use this method to specify distribution strategy for a Cell,
strategy for others will be set by sharding propagation.
in_strategy and out_strategy define the input and output layout respectively.
in_strategy/out_strategy should be a tuple, each element of which corresponds to the desired layout of
this input/output, which can refer to the description of `mindspore.ops.Primitive.shard`.
The parallel strategies of remaining operators are derived from the strategy specified by the input and output.
Note:
If Cell.shard is called, the parallel mode in `set_auto_parallel_context` (parallel_mode) will be set to
"auto_parallel" and the search mode (search_mode) to "sharding_propagation".
If the input contain Parameter, its strategy should be set in `in_strategy`.
Args:
in_strategy (tuple): Define the layout of inputs, each element of the tuple should be a tuple. Tuple
defines the layout of the corresponding input.
out_strategy (Union[None, tuple]): Define the layout of outputs similar with in_strategy.
It is not in use right now. Default: ``None`` .
parameter_plan (Union[dict, None]): Define the layout for the specified parameters. Each element in dict
defines the layout of the parameter like "param_name: layout".
The key is a parameter name of type 'str'.
The value is a 1-D integer tuple, indicating the corresponding layout.
If the parameter name is incorrect or the corresponding parameter
has been set, the parameter setting will be ignored.
Default: ``None`` .
device (string): Select a certain device target. It is not in use right now.
Support [ ``"CPU"`` , ``"GPU"`` , ``"Ascend"`` ]. Default: ``"Ascend"`` .
level (int): Option for parallel strategy infer algorithm, namely the object function, maximize computation
over communication ratio, maximize speed performance, minimize memory usage etc. It is not in
use right now. Support [ ``"0"`` , ``"1"`` , ``"2"`` ]. Default: ``0`` .
Returns:
Function, return the cell construct function that will be executed under auto parallel process.
Examples:
>>> import mindspore.nn as nn
>>>
>>> class Block(nn.Cell):
... def __init__(self):
... self.dense1 = nn.Dense(10, 10)
... self.relu = nn.ReLU()
... self.dense2 = nn.Dense2(10, 10)
... def construct(self, x):
... x = self.relu(self.dense2(self.relu(self.dense1(x))))
... return x
>>>
>>> class example(nn.Cell):
... def __init__(self):
... self.block1 = Block()
... self.block2 = Block()
... self.block2_shard = self.block2.shard(in_strategy=((2, 1),),
... parameter_plan={'self.block2.shard.dense1.weight': (4, 1)})
... def construct(self, x):
... x = self.block1(x)
... x = self.block2_shard(x)
... return x
"""
if context.get_auto_parallel_context("parallel_mode") not in ["auto_parallel", "semi_auto_parallel"]:
raise AssertionError(f"Cell shard only supports auto parallel or semi_auto_parallel "
f"Please check the parallel mode in parallel context.")
shard_fn = Shard()
fn = shard_fn(self, in_strategy, out_strategy, parameter_plan, device, level)
self._shard_fn = fn
return fn
def _init_check(self):
for param in self.get_parameters(expand=False):
if param.has_init:
param.init_data()
self._init_flag = True
def _self_check(self):
if not self._is_check_and_refresh:
self.check_names_and_refresh_name()
self._is_check_and_refresh = True
def _predict(self, *args, **kwargs):
if not hasattr(self, "phase"):
return False, None
if (self.phase == "prefill" or self.phase == 'increment') and self.phase in self.phase_cache:
new_args = _get_args_for_run_predict(self, args, kwargs, self._compile_args)
res = _cell_graph_executor._graph_executor(tuple(new_args), self.phase_cache[self.phase])
res = _convert_python_data(res)
return True, res
return False, None
def __call__(self, *args, **kwargs):
# Run in Graph mode.
if context._get_mode() == context.GRAPH_MODE and os.getenv("MS_JIT") != '0':
if kwargs:
bound_arguments = self.sig.bind(*args, **kwargs)
bound_arguments.apply_defaults()
args = bound_arguments.args
kwargs = bound_arguments.kwargs
predict_compiled, res = self._predict(*args, **kwargs)
if predict_compiled:
return res
self._check_construct_args(*args)
if self._hook_fn_registered():
logger.warning(f"For 'Cell', it's not support hook function in graph mode. If you want to use hook "
f"function, please use context.set_context to set pynative mode.")
self._self_check()
out = self.compile_and_run(*args, **kwargs)
return out
# Run in PyNative mode.
if not (self._init_flag or self._is_check_and_refresh):
self._init_check()
self._self_check()
if not (self.requires_grad or self._dynamic_shape_inputs or self.mixed_precision_type):
if not (self._forward_pre_hook or self._forward_hook or self._backward_pre_hook or self._backward_hook or
self._shard_fn or self._recompute_cell or (self.has_bprop and _pynative_executor.requires_grad())):
return self.construct(*args, **kwargs)
return self._run_construct(*args, **kwargs)
return self._complex_call(*args, **kwargs)
def _complex_call(self, *args, **kwargs):
"""
PyNative call with requires_grad or hooks
"""
self._call_pre_process(*args, **kwargs)
if not (self._forward_pre_hook or self._forward_hook or self._backward_pre_hook or self._backward_hook or
self._shard_fn or self._recompute_cell or self.has_bprop):
output = self.construct(*args, **kwargs)
else:
output = self._run_construct(*args, **kwargs)
self._call_post_process(output, *args, **kwargs)
return output
def _call_pre_process(self, *args, **kwargs):
"""
Process cell info before call construct
"""
if self.requires_grad:
_pynative_executor.set_grad_flag(True)
_pynative_executor.new_graph(self, *args, **kwargs)
elif self._dynamic_shape_inputs is not None:
_pynative_executor.set_cell_use_dynamic_shape_process(True)
# Set mixed precision
if self.mixed_precision_type is not None:
_pynative_executor.set_mixed_precision_type(self.mixed_precision_type)
def _call_post_process(self, output, *args, **kwargs):
"""
Process cell info after call construct
"""
if self.requires_grad:
_pynative_executor.end_graph(self, output, *args, **kwargs)
elif self._dynamic_shape_inputs is not None:
_pynative_executor.set_cell_use_dynamic_shape_process(False)
# mixed precision reset
if self.mixed_precision_type is not None:
_pynative_executor.set_mixed_precision_type(MixedPrecisionType.NOTSET, False)
def _call_custom_bprop(self, *args, **kwargs):
"""
Call custom bprop for cell bprop.
"""
with _no_grad():
output = self.construct(*args, **kwargs)
_pynative_executor.call_custom_bprop(self, output, *args, **kwargs)
return output
def _add_attr(self, name, value):
if name and name[:2] != '__' and name not in Cell.IGNORE_LIST:
super(Cell, self)._add_attr(name, value)
def _sync_attr_for_compile(self):
"""Sync the attr to c++ object."""
if self._attr_synced:
return
cells = self.__dict__.get('_cells')
for key in cells:
cell = cells[key]
cell._sync_attr_for_compile()
self._add_attr(key, cell)
params = self.__dict__.get('_params')
for key in params:
if '.' in key:
continue
param = params[key]
self._add_attr(key, param)
params_list = self.__dict__.get('_params_list')
for key in params_list:
params_list_item = params_list[key]
self._add_attr(key, params_list_item)
for key in self.__dict__:
value = self.__dict__[key]
self._add_attr(key, value)
self._attr_synced = True
def _set_attr_for_parameter(self, name, value):
"""Set attr for parameter."""
cells = self.__dict__.get('_cells')
params = self.__dict__.get('_params')
if params is None:
raise AttributeError("For 'Cell', can not assign params before Cell.__init__() is called.")
if name in self.__dict__:
if self.__dict__[name] is not None:
raise TypeError(f"For 'Cell', the {name} should not be Parameter.")
del self.__dict__[name]
if cells and name in cells:
raise TypeError(f"For 'Cell', the {name} must be Cell, but got Parameter.")
self.insert_param_to_cell(name, value)
def _set_attr_for_parameter_tuple(self, name, value):
"""Set attr for parameter in ParameterTuple."""
params = self.__dict__.get('_params')
params_list = self.__dict__.get('_params_list')
if params is None:
raise AttributeError("For 'Cell', can not assign params before Cell.__init__() is called.")
exist_names = set("")
exist_objs = set()
for item in value:
if item in exist_objs:
# If there are multiple identical objects, their names only check once.
continue
exist_objs.add(item)
if item.name == PARAMETER_NAME_DEFAULT:
logger.warning("For 'Cell', the parameter definition is deprecated.\n"
"Please set a unique name for the parameter in ParameterTuple '{}'.".format(value))
item.name = item.name + "$" + str(self._id)
self._id += 1
self.insert_param_to_cell(item.name, item, check_name_contain_dot=False)
if item.name in exist_names:
raise ValueError("The value {} , its name '{}' already exists. "
"Please set a unique name for the parameter.".format(value, item.name))
exist_names.add(item.name)
if context._get_mode() == context.PYNATIVE_MODE:
if name in self.__dict__:
del self.__dict__[name]
if name in params:
del params[name]
params_list[name] = value
else:
object.__setattr__(self, name, value)
def _set_attr_for_parameter_in_list_or_tuple(self, name, value):
"""Set attr for parameter in list or tuple."""
for item in value:
if item in self.exist_objs:
# If there are multiple identical objects, their names only check once.
continue
self.exist_objs.add(item)
if item.name == PARAMETER_NAME_DEFAULT:
item.name = item.name + "$" + str(self._id)
self._id += 1
if item.name in self.exist_names:
raise ValueError("The value {} , its name '{}' already exists. "
"Please set a unique name for the parameter.".format(value, item.name))
self.exist_names.add(item.name)
object.__setattr__(self, name, value)
def _set_attr_for_cell(self, name, value):
"""Set attr for cell."""
cells = self.__dict__.get('_cells')
params = self.__dict__.get('_params')
if cells is None:
raise AttributeError("For 'Cell', can not assign cells before Cell.__init__() is called.")
if name in self.__dict__:
del self.__dict__[name]
if params and name in params:
raise TypeError(f"For 'Cell', the {name} must be Parameter, but got Cell.")
if self._auto_prefix:
value.update_parameters_name(name + '.')
cells[name] = value
if hasattr(self, '_cell_init_args'):
self.cell_init_args += str({name: value})
def _set_attr_for_params(self, name, value):
if isinstance(value, Tensor) and self._params[name] is not None:
self._params[name].set_data(value)
elif value is not None:
raise TypeError(f"For 'Cell', the type of {name} must be Parameter or ParameterTuple, "
f"but got {type(value).__name__}.")
else:
self.insert_param_to_cell(name, None)
def __setattr__(self, name, value):
cells = self.__dict__.get('_cells')
params = self.__dict__.get('_params')
if isinstance(value, Parameter):
self._set_attr_for_parameter(name, value)
elif isinstance(value, ParameterTuple):
self._set_attr_for_parameter_tuple(name, value)
elif isinstance(value, (list, tuple)) and value and _check_param_list_tuple(value):
self._set_attr_for_parameter_in_list_or_tuple(name, value)
elif isinstance(value, Cell):
self._set_attr_for_cell(name, value)
elif params and name in params:
self._set_attr_for_params(name, value)
elif cells and name in cells:
if value is not None:
raise TypeError(f"For 'Cell', the type of {name} must be cell, but got {type(value).__name__}.")
self._cells[name] = None
else:
if isinstance(value, Primitive):
value.set_prim_instance_name(name)
self._primitives[name] = value
object.__setattr__(self, name, value)
if name not in Cell.IGNORE_LIST:
self._attr_synced = False
[docs] def extend_repr(self):
"""
Expand the description of Cell.
To print customized extended information, re-implement this method in your own cells.
"""
return ''
def __str__(self):
return self.__repr__()
def __repr__(self):
extra_str = self.extend_repr()
info_str = self.__class__.__name__ + '<'
if self._cells:
sub_str = '\n'
if extra_str:
sub_str += '{}\n'.format(self.extend_repr())
for key, value in self._cells.items():
sub_str += '({}): {}\n'.format(key, repr(value))
sub_str = sub_str.replace('\n', '\n ') + '>'
info_str += sub_str
else:
info_str += extra_str + '>'
return info_str
def load_parameter_slice(self, params):
"""
Replace parameters with sliced tensors by parallel strategies.
Note:
This interface is deprecated.
"""
logger.warning("'load_parameter_slice' function is deprecated.")
def set_parallel_input_with_inputs(self, *inputs):
"""
Slice inputs tensors by parallel strategies.
Note:
This interface is deprecated.
"""
logger.warning("'set_parallel_input_with_inputs' function is deprecated.")
def _check_parameter_consistency(self, set_inputs, net_inputs):
"""Check consistency for parameter."""
for index, (set_input, net_input) in enumerate(zip(set_inputs, net_inputs)):
if isinstance(set_input, Tensor):
if not isinstance(net_input, Tensor):
raise TypeError(
f"For 'set_inputs' and tuple(list) in 'set_inputs',the type of {index + 1}th input must "
f"be Tensor, but got {type(net_input)}.")
if isinstance(set_input, Parameter) != isinstance(net_input, Parameter):
raise TypeError(
f"For 'set_inputs' and tuple(list) in 'set_inputs', the {index + 1}th input must be the same "
f"as expected, but got expected: {type(set_input)} and input: {type(net_input)}.")
elif isinstance(set_input, (tuple, list)):
if not isinstance(net_input, (tuple, list)):
raise TypeError(
f"The {index + 1}th input type of 'set_inputs' or tuple(list) in "
f"'set_inputs' must be tuple or list, but got {type(net_input)}.")
self._check_parameter_consistency(set_input, net_input)
def _get_compile_args(self, args):
"""Get compile arguments."""
# this is used only for test
set_by_auto_dynamic = False
if is_auto_dynamic():
if self._dynamic_shape_inputs is None:
set_by_auto_dynamic = True
else:
if isinstance(self._dynamic_shape_inputs, (list, tuple)) and self._dynamic_shape_inputs[0] is None:
set_by_auto_dynamic = True
if set_by_auto_dynamic:
self._dynamic_shape_inputs = convert_inputs_to_dynamic(*args)
if self._dynamic_shape_inputs is not None:
logger.debug("Compiled Graph with dynamic shape")
compile_args = _generate_dyn_compile_args(args, self._dynamic_shape_inputs)
_cell_graph_executor._graph_executor.check_argument_consistency(compile_args, args, "set_inputs")
self._check_parameter_consistency(compile_args, args)
Validator.check_symbolic_shape(compile_args, args)
self.saved_dynamic_shape = compile_args
return compile_args
return args
[docs] def compile(self, *args, **kwargs):
"""
Compile Cell as a computation graph, the input must be consistent with the input defined in construct.
Args:
args (tuple): Args of the Cell object.
kwargs (dict): Kwargs of the Cell object.
"""
self._compile_args = self._get_compile_args(args)
_cell_graph_executor.compile(self, *self._compile_args, phase=self.phase,
jit_config_dict=self._jit_config_dict, **kwargs)
[docs] def compile_and_run(self, *args, **kwargs):
"""
Compile and run Cell, the input must be consistent with the input defined in construct.
Note:
It is not recommended to call directly.
Args:
args (tuple): Args of the Cell object.
kwargs (dict): Kwargs of the Cell object.
Returns:
Object, the result of executing.
"""
self.compile(*args, **kwargs)
self.add_flags(ge_sync_data=False)
new_args = _get_args_for_run(self, args, kwargs, self._compile_args)
return _cell_graph_executor(self, *new_args, phase=self.phase)
def auto_parallel_compile_and_run(self):
"""
Whether or not to execute compile and run in 'AUTO_PARALLEL' or 'SEMI_AUTO_PARALLEL' mode.
Note:
This interface is deprecated.
"""
logger.warning("'auto_parallel_compile_and_run' function is deprecated.")
def exec_checkpoint_graph(self):
"""Executes GE saving checkpoint graph operation."""
logger.warning("'exec_checkpoint_graph' function is deprecated.")
self.add_flags(ge_sync_data=True)
_cell_graph_executor(self, phase='save')
[docs] def insert_param_to_cell(self, param_name, param, check_name_contain_dot=True):
"""
Adds a parameter to the current cell.
Inserts a parameter with given name to the cell. The method is currently used in
`mindspore.nn.Cell.__setattr__`.
Args:
param_name (str): Name of the parameter.
param (Parameter): Parameter to be inserted to the cell.
check_name_contain_dot (bool): Determines whether the name input is compatible. Default: ``True`` .
Raises:
KeyError: If the name of parameter is null or contains dot.
TypeError: If the type of parameter is not Parameter.
Examples:
>>> import mindspore as ms
>>> from mindspore import Tensor, nn, Parameter
...
>>> class Net(nn.Cell):
... def __init__(self):
... super(Net, self).__init__()
... self.relu = nn.ReLU()
...
... def construct(self, x):
... x = self.relu(x)
... return x
>>> net = Net()
>>> net.insert_param_to_cell("bias", Parameter(Tensor([1, 2, 3])))
>>> print(net.bias)
Parameter(name=bias, shape=(3,), dtype=Int64, requires_grad=True)
"""
if not param_name:
raise KeyError(f"For 'insert_param_to_cell', the argument 'param_name' should not be None.")
if check_name_contain_dot and '.' in param_name:
raise KeyError(f"For 'insert_param_to_cell', the argument 'param_name' should not contain'.' ")
if '_params' not in self.__dict__:
raise AttributeError(f"For 'insert_param_to_cell', please call Cell.__init__() firstly.")
if hasattr(self, param_name) and param_name not in self._params:
raise KeyError(f"For 'insert_param_to_cell', the {param_name} parameter already exists in the network."
f"Cannot insert another parameter with the same name.")
if not isinstance(param, Parameter) and param is not None:
raise TypeError(f"For 'insert_param_to_cell', the argument 'param' must be 'Parameter' if not None, "
f"but got {type(param)}.")
if isinstance(param, Parameter) and param.name == PARAMETER_NAME_DEFAULT:
param.name = param_name
self._params[param_name] = param
[docs] def cast_param(self, param):
"""
Cast parameter according to auto mix precision level in pynative mode.
This interface is currently used in the case of auto mix precision and usually needs not to be used explicitly.
Args:
param (Parameter): Parameters, the type of which should be cast.
Returns:
Parameter, the input parameter with type automatically cast.
"""
msg = f"'cast_param' is deprecated from version 2.0 and will be removed in a future version."
logger.warning(msg)
mixed_type = self.get_mixed_precision_type()
if mixed_type != MixedPrecisionType.NOTSET:
if mixed_type == MixedPrecisionType.FP32:
param.set_cast_dtype(mstype.float32)
elif mixed_type == MixedPrecisionType.FP16:
param.set_cast_dtype(mstype.float16)
elif hasattr(param, "set_cast_dtype"):
# retest dtype
param.set_cast_dtype()
return param
[docs] def insert_child_to_cell(self, child_name, child_cell):
"""
Adds a child cell to the current cell with a given name.
Args:
child_name (str): Name of the child cell.
child_cell (Cell): The child cell to be inserted.
Raises:
KeyError: Child Cell's name is incorrect or duplicated with the other child name.
TypeError: If type of `child_name` is not str.
TypeError: Child Cell's type is incorrect.
Examples:
>>> import mindspore as ms
>>> from mindspore import Tensor, nn
...
>>> net1 = nn.ReLU()
>>> net2 = nn.Dense(2, 2)
>>> net1.insert_child_to_cell("child", net2)
>>> print(net1)
ReLU<
(child): Dense<input_channels=2, output_channels=2, has_bias=True>
>
"""
if not isinstance(child_name, str):
raise TypeError(f"For 'insert_child_to_cell', the type of parameter 'child_name' must be str, "
f"but got {type(child_name)}.")
if not child_name or '.' in child_name:
raise KeyError(f"For 'insert_child_to_cell', the parameter 'child_name' can not be None and "
"can not contain '.' ")
if hasattr(self, child_name) and child_name not in self._cells:
raise KeyError(f"For 'insert_child_to_cell', the {child_name} child cell already exists in the network."
f"Cannot insert another child cell with the same name.")
if not isinstance(child_cell, Cell) and child_cell is not None:
raise TypeError(f"For 'insert_child_to_cell', the argument 'child_cell' must be 'Cell' if not None, "
f"but got type {type(child_cell)}.")
self._cells[child_name] = child_cell
[docs] def construct(self, *args, **kwargs):
"""
Defines the computation to be performed. This method must be overridden by all subclasses.
Note:
It is not supported currently that inputs contain both tuple and non-tuple types at same time.
Args:
args (tuple): Tuple of variable parameters.
kwargs (dict): Dictionary of variable keyword parameters.
Returns:
Tensor, returns the computed result.
"""
raise AttributeError("For 'Cell', the method 'construct' is not defined.")
[docs] def remove_redundant_parameters(self):
"""
Remove the redundant parameters.
This interface usually needs not to be used explicitly.
"""
cells = self.cells_and_names()
for _, cell in cells:
params = cell._params.items()
for param_name, param in list(params):
if param.name not in self.parallel_parameter_name_list:
cell._params.pop(param_name)
logger.info("remove the redundant parameter: %s", param.name)
continue
cell_dict = cell.__dict__
for key in cell_dict:
if isinstance(cell_dict[key], ParameterTuple):
param_tuple = cell_dict[key]
new_param_tuple = []
for param in param_tuple:
if param.name not in self.parallel_parameter_name_list:
logger.info("remove the redundant parameter: %s in ParameterTuple", param.name)
continue
new_param_tuple.append(param)
cell.__dict__[key] = ParameterTuple(new_param_tuple)
[docs] def init_parameters_data(self, auto_parallel_mode=False):
"""
Initialize all parameters and replace the original saved parameters in cell.
Note:
trainable_params() and other similar interfaces may return different parameter instance after
`init_parameters_data`, do not save these results.
Args:
auto_parallel_mode (bool): If running in auto_parallel_mode. Default: ``False`` .
Returns:
Dict[Parameter, Parameter], returns a dict of original parameter and replaced parameter.
Examples:
>>> import mindspore as ms
>>> from mindspore import Tensor, nn
...
>>> class Net(nn.Cell):
... def __init__(self):
... super(Net, self).__init__()
... self.dense = nn.Dense(2, 2)
...
... def construct(self, x):
... x = self.dense(x)
... return x
>>> net = Net()
>>> print(net.init_parameters_data())
{Parameter (name=dense.weight, shape=(2,2), dtype=Float32, requires_grad=True):
Parameter (name=dense.weight, shape=(2,2), dtype=Float32, requires_grad=True),
Parameter (name=dense.bias, shape=(2,), dtype=Float32, requires_grad=True):
Parameter (name=dense.bias, shape=(2,), dtype=Float32, requires_grad=True)}
"""
replace = dict()
def _updata(param):
if param in replace:
return replace.get(param)
new_p = param.init_data(None, set_sliced=False)
replace[param] = new_p
return new_p
# replace all original usage.
cells = self.cells_and_names()
for _, cell in cells:
params = cell._params.items()
for param_name, param in params:
if not auto_parallel_mode:
cell._params[param_name] = _updata(param)
continue
if param.name in self.parallel_parameter_name_list:
cell._params[param_name] = _updata(param)
cell_dict = cell.__dict__
for key in cell_dict:
if isinstance(cell_dict[key], ParameterTuple):
param_tuple = cell_dict[key]
new_param_tuple = []
for param in param_tuple:
if not auto_parallel_mode:
new_param_tuple.append(_updata(param))
continue
if param.name in self.parallel_parameter_name_list:
new_param_tuple.append(_updata(param))
else:
new_param_tuple.append(param)
cell.__dict__[key] = ParameterTuple(new_param_tuple)
return replace
[docs] def parameters_dict(self, recurse=True):
"""
Gets the parameters dictionary of this cell.
Args:
recurse (bool): Whether contains the parameters of subcells. Default: ``True`` .
Returns:
OrderedDict, return parameters dictionary.
Examples:
>>> import mindspore as ms
>>> from mindspore import Tensor, nn, Parameter
...
>>> class Net(nn.Cell):
... def __init__(self):
... super(Net, self).__init__()
... self.dense = nn.Dense(2, 2)
...
... def construct(self, x):
... x = self.dense(x)
... return x
>>> net = Net()
>>> print(net.parameters_dict())
OrderedDict([('dense.weight', Parameter(name=dense.weight, shape=(2, 2), dtype=Float32,
requires_grad=True)), ('dense.bias', Parameter(name=dense.bias, shape=(2,), dtype=Float32,
requires_grad=True))])
"""
param_dict = OrderedDict()
for param in self.get_parameters(expand=recurse):
param_dict[param.name] = param
return param_dict
[docs] def parameters_broadcast_dict(self, recurse=True):
"""
Gets the parameters broadcast dictionary of this cell.
Args:
recurse (bool): Whether contains the parameters of subcells. Default: ``True`` .
Returns:
OrderedDict, return parameters broadcast dictionary.
"""
param_dict = OrderedDict()
for param in self.get_parameters(expand=recurse):
if param.layerwise_parallel is False:
param_dict[param.name] = param
if not param_dict:
return None
return param_dict
[docs] def update_parameters_name(self, prefix='', recurse=True):
"""
Adds the `prefix` string to the names of parameters.
Args:
prefix (str): The prefix string. Default: ``''`` .
recurse (bool): Whether contains the parameters of subcells. Default: ``True`` .
"""
Validator.check_str_and_none_by_regular(prefix)
for name, param in self.parameters_and_names(expand=recurse):
if prefix != '':
param.is_init = False
param.name = prefix + name
def _update_local_parameters_name(self, prefix='', recurse=True):
"""
Updates the names of local parameters with given prefix string.
Adds the given prefix to the names of local parameters.
Local parameters means the parameters without user input.
Args:
prefix (str): The prefix string. Default: ''.
recurse (bool): Whether contains the parameters of subcells. Default: ``True``.
"""
Validator.check_str_by_regular(prefix)
for name, param in self.parameters_and_names(expand=recurse):
if name in self._user_parameters:
continue
if prefix != '':
param.is_init = False
param.name = prefix + name
# generate api by del decorator.
[docs] def trainable_params(self, recurse=True):
"""
Returns all trainable parameters.
Returns a list of all trainable parameters.
Args:
recurse (bool): Whether contains the trainable parameters of subcells. Default: ``True`` .
Returns:
List, the list of trainable parameters.
Tutorial Examples:
- `Model Training - Optimizer
<https://mindspore.cn/tutorials/en/r2.4.0/beginner/train.html#optimizer>`_
"""
return list(filter(lambda x: x.requires_grad, self.get_parameters(expand=recurse)))
# generate api by del decorator.
[docs] def untrainable_params(self, recurse=True):
"""
Returns all untrainable parameters.
Returns a list of all untrainable parameters.
Args:
recurse (bool): Whether contains the untrainable parameters of subcells. Default: ``True`` .
Returns:
List, the list of untrainable parameters.
"""
return list(filter(lambda x: not x.requires_grad, self.get_parameters(expand=recurse)))
# generate api by del decorator.
[docs] def get_parameters(self, expand=True):
"""
Returns an iterator over cell parameters.
Yields parameters of this cell. If `expand` is ``true`` , yield parameters of this cell and all subcells.
For more details about subcells, please see the example below.
Args:
expand (bool): If ``true`` , yields parameters of this cell and all subcells. Otherwise, only yield
parameters that are direct members of this cell. Default: ``True`` .
Returns:
Iteration, all parameters at the cell.
Examples:
>>> import mindspore as ms
>>> from mindspore import nn, ops, Tensor
>>> import numpy as np
>>> class TestNet(nn.Cell):
... def __init__(self):
... super().__init__()
... self.my_w1 = ms.Parameter(Tensor(np.ones([4, 4]), ms.float32))
... self.my_w2 = ms.Parameter(Tensor(np.ones([16]), ms.float32))
... def construct(self, x):
... x += self.my_w1
... x = ops.reshape(x, (16,)) - self.my_w2
... return x
>>> class TestNet2(nn.Cell):
... def __init__(self):
... super().__init__()
... self.my_t1 = ms.Parameter(Tensor(np.ones([4, 4]), ms.float32))
... # self.subcell is a subcell of TestNet2, when using expand=True, the parameters of TestNet will
... # also be gathered.
... self.subcell = TestNet()
... def construct(self, x):
... x += self.my_w1
... x = ops.reshape(x, (16,)) - self.my_w2
... return x
>>> net = TestNet2()
>>> print([p for p in net.get_parameters(expand=True)])
[Parameter (name=my_t1, shape=(4, 4), dtype=Float32, requires_grad=True), Parameter (name=subcell.my_w1,
shape=(4, 4), dtype=Float32, requires_grad=True), Parameter (name=subcell.my_w2, shape=(16,), dtype=Float32,
requires_grad=True)]
"""
for _, param in self.parameters_and_names(expand=expand):
yield param
# pylint: disable=missing-docstring
def check_names_and_refresh_name(self):
if not hasattr(self, "_params"):
return
all_name = [i.name for i in dict(self.parameters_and_names()).values()]
if len(set(all_name)) < len(all_name):
self.update_parameters_name()
self.check_names()
[docs] def check_names(self):
"""
Check the names of cell parameters.
"""
names = set("")
for value, param in self.parameters_and_names():
if param.name in names:
raise ValueError("The value of {} is {}, its name '{}' already exists. "
"Please set a unique name for the parameter.".format(value, param, param.name))
names.add(param.name)
[docs] def parameters_and_names(self, name_prefix='', expand=True):
"""
Returns an iterator over cell parameters.
Includes the parameter's name and itself.
Args:
name_prefix (str): Namespace. Default: ``''`` .
expand (bool): If true, yields parameters of this cell and all subcells. Otherwise, only yield parameters
that are direct members of this cell. Default: ``True`` .
Returns:
Iteration, all the names and corresponding parameters in the cell.
Examples:
>>> from mindspore import nn
>>> n = nn.Dense(3, 4)
>>> names = []
>>> for m in n.parameters_and_names():
... if m[0]:
... names.append(m[0])
Tutorial Examples:
- `Building a Network - Model Parameters
<https://mindspore.cn/tutorials/en/r2.4.0/beginner/model.html#model-parameters>`_
"""
cells = []
if expand:
cells = self.cells_and_names(name_prefix=name_prefix)
else:
cells.append((name_prefix, self))
params_set = set()
for cell_name, cell in cells:
params = cell._params.items()
for par_name, par in params:
if par is not None and par.inited_param is not None:
par = par.inited_param
if par is not None and id(par) not in params_set:
params_set.add(id(par))
par_new_name = par_name
if cell_name:
par_new_name = cell_name + '.' + par_new_name
yield par_new_name, par
[docs] def cells_and_names(self, cells=None, name_prefix=''):
"""
Returns an iterator over all cells in the network, including the cell's name and itself.
Args:
cells (str): Cells to iterate over. Default: ``None`` .
name_prefix (str): Namespace. Default: ``''`` .
Returns:
Iteration, all the child cells and corresponding names in the cell.
Examples:
>>> from mindspore import nn
>>> class Net(nn.Cell):
... def __init__(self):
... super(Net, self).__init__()
... self.conv = nn.Conv2d(3, 64, 3)
... def construct(self, x):
... out = self.conv(x)
... return out
>>> names = []
>>> n = Net()
>>> for m in n.cells_and_names():
... if m[0]:
... names.append(m[0])
"""
t_cells = cells if cells else set()
if self in t_cells:
return
t_cells.add(self)
yield name_prefix, self
for name, cell in self._cells.items():
if cell:
cells_name_prefix = name
if name_prefix:
cells_name_prefix = name_prefix + '.' + cells_name_prefix
for ele in cell.cells_and_names(t_cells, cells_name_prefix):
yield ele
[docs] def cells(self):
"""
Returns an iterator over immediate cells.
Returns:
Iteration, the immediate cells in the cell.
Examples:
>>> import mindspore as ms
>>> from mindspore import Tensor, nn
...
>>> class Net(nn.Cell):
... def __init__(self):
... super(Net, self).__init__()
... self.dense = nn.Dense(2, 2)
...
... def construct(self, x):
... x = self.dense(x)
... return x
>>> net = Net()
>>> print(net.cells())
odict_values([Dense<input_channels=2, output_channels=2, has_bias=True>])
"""
return self.name_cells().values()
def _set_scope(self, name):
"""Sets the name on the first time."""
if self._scope is None:
self._scope = name
elif self._scope == 'recompute_':
self._scope = self._scope + name
def _children_scope_recursive(self, parent_prefix='Default'):
"""Generates the scope of each layer of the network recursively."""
reserve_class_name_in_scope = context.get_context("reserve_class_name_in_scope")
for name, cell in self.name_cells().items():
class_name = ("-" + cell.__class__.__name__) if reserve_class_name_in_scope else ""
yield parent_prefix + "/" + name + class_name, cell
for name, cell in self.name_cells().items():
class_name = ("-" + cell.__class__.__name__) if reserve_class_name_in_scope else ""
for key, value in cell._children_scope_recursive(parent_prefix + "/" + name + class_name):
yield key, value
[docs] def get_scope(self):
"""
Returns the scope of a cell object in one network.
Returns:
String, scope of the cell.
"""
return self._scope
[docs] def generate_scope(self):
"""Generate the scope for each cell object in the network."""
for name, cell in self._children_scope_recursive():
cell._set_scope(name)
[docs] def name_cells(self):
"""
Returns an iterator over all immediate cells in the network.
Include name of the cell and cell itself.
Returns:
Dict, all the child cells and corresponding names in the cell.
Examples:
>>> import mindspore as ms
>>> from mindspore import Tensor, nn
...
>>> class Net(nn.Cell):
... def __init__(self):
... super(Net, self).__init__()
... self.dense = nn.Dense(2, 2)
...
... def construct(self, x):
... x = self.dense(x)
... return x
>>> net = Net()
>>> print(net.name_cells())
OrderedDict([('dense', Dense<input_channels=2, output_channels=2, has_bias=True>)])
"""
value_set = set()
cells = OrderedDict()
for name, cell in self._cells.items():
if cell is not None and cell not in value_set:
value_set.add(cell)
cells[name] = cell
return cells
def _add_mixed_precision_flag(self, **flags):
"""Add mixed precision flag to current cell"""
if "fp16" in flags and flags.get("fp16", False):
self.mixed_precision_type = MixedPrecisionType.FP16
Cell_.set_mixed_precision_type(self, MixedPrecisionType.FP16)
if "fp32" in flags and flags.get("fp32", False):
self.mixed_precision_type = MixedPrecisionType.FP32
Cell_.set_mixed_precision_type(self, MixedPrecisionType.FP32)
if "bf16" in flags and flags.get("bf16", False):
self.mixed_precision_type = MixedPrecisionType.BF16
Cell_.set_mixed_precision_type(self, MixedPrecisionType.BF16)
[docs] def apply(self, fn):
"""
Applies fn recursively to every subcell (as returned by .cells()) as well as self.
Typical use includes initializing the parameters of a model.
Args:
fn (function): function to be applied to each subcell.
Returns:
Cell, self.
Examples:
>>> import mindspore.nn as nn
>>> from mindspore.common.initializer import initializer, One
>>> net = nn.SequentialCell(nn.Dense(2, 2), nn.Dense(2, 2))
>>> def func(cell):
... if isinstance(cell, nn.Dense):
... cell.weight.set_data(initializer(One(), cell.weight.shape, cell.weight.dtype))
>>> net.apply(func)
SequentialCell<
(0): Dense<input_channels=2, output_channels=2, has_bias=True>
(1): Dense<input_channels=2, output_channels=2, has_bias=True>
>
>>> print(net[0].weight.asnumpy())
[[1. 1.]
[1. 1.]]
"""
for cell in self.cells():
cell.apply(fn)
fn(self)
return self
[docs] def add_flags(self, **flags):
"""
Add customized attributes for cell.
This method is also called when the cell class is instantiated and the class parameter 'flags' is set to True.
Args:
flags (dict): Network configuration information, currently it is used for the binding of network and
dataset. Users can also customize network attributes by this parameter.
Examples:
>>> import mindspore as ms
>>> from mindspore import Tensor, nn
...
>>> class Net(nn.Cell):
... def __init__(self):
... super(Net, self).__init__()
... self.relu = nn.ReLU()
...
... def construct(self, x):
... x = self.relu(x)
... return x
>>> net = Net()
>>> net.add_flags(sink_mode=True)
>>> print(net.sink_mode)
True
"""
if not hasattr(self, "_func_graph_flags"):
self._func_graph_flags = {}
self._func_graph_flags.update({**flags})
if context._get_mode() == context.PYNATIVE_MODE and self._func_graph_flags.get("output_no_recompute"):
raise TypeError("Recompute is not supported in PyNative mode currently, you can use "
"'context.set_context(mode=context.GRAPH_MODE)' or @jit to set graph mode.")
self.__dict__.update({**flags})
self._add_mixed_precision_flag(**flags)
return self
[docs] def add_flags_recursive(self, **flags):
"""
If a cell contains child cells, this method can recursively customize attributes of all cells.
Args:
flags (dict): Network configuration information, currently it is used for the binding of network and
dataset. Users can also customize network attributes by this parameter.
Examples:
>>> import mindspore as ms
>>> from mindspore import Tensor, nn
...
>>> class Net(nn.Cell):
... def __init__(self):
... super(Net, self).__init__()
... self.relu = nn.ReLU()
...
... def construct(self, x):
... x = self.relu(x)
... return x
>>> net = Net()
>>> net.add_flags_recursive(sink_mode=True)
>>> print(net.sink_mode)
True
"""
self.add_flags(**flags)
for cell in self.cells():
cell.add_flags_recursive(**flags)
return self
def _add_init_args(self, **args):
if hasattr(self, '_cell_init_args'):
self._cell_init_args += str({**args})
[docs] def get_flags(self):
"""
Get the self_defined attributes of the cell, which can be added by `add_flags` method.
Examples:
>>> import mindspore as ms
>>> from mindspore import Tensor, nn
...
>>> class Net(nn.Cell):
... def __init__(self):
... super(Net, self).__init__()
... self.relu = nn.ReLU()
...
... def construct(self, x):
... x = self.relu(x)
... return x
>>> net = Net()
>>> net.add_flags(sink_mode=True)
>>> print(net.get_flags())
{'sink_mode':True}
"""
if not hasattr(self, "_func_graph_flags"):
self._func_graph_flags = {}
return self._func_graph_flags
[docs] def to_float(self, dst_type):
"""
Add cast on all inputs of cell and child cells to run with certain float type.
If `dst_type` is `mindspore.dtype.float16`, all the inputs of Cell, including input, Parameter and Tensor, will
be cast to float16. Please refer to the usage in source code of :func:`mindspore.amp.build_train_network`.
Note:
Multiple calls will overwrite.
Args:
dst_type (:class:`mindspore.dtype`): Transfer cell to run with dst_type.
dst_type can be `mstype.float16` , `mstype.float32` or `mstype.bfloat16`.
Returns:
Cell, the cell itself.
Raises:
ValueError: If dst_type is not `mstype.float32` , `mstype.float16` or `mstype.bfloat16`.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore.nn as nn
>>> from mindspore import dtype as mstype
>>>
>>> net = nn.Conv2d(120, 240, 4, has_bias=False, weight_init='normal')
>>> net.to_float(mstype.float16)
Conv2d<input_channels=120, output_channels=240, kernel_size=(4, 4), stride=(1, 1), pad_mode=same,
padding=0, dilation=(1, 1), group=1, has_bias=False, weight_init=normal, bias_init=None, format=NCHW>
"""
if dst_type not in (mstype.float16, mstype.float32, mstype.bfloat16):
raise ValueError("For 'to_float', the argument 'dst_type' must be mstype.float32, mstype.float16 or "
"mstype.bfloat16, but got type: {} and value: {}.".format(type(dst_type), dst_type))
flags = {'fp16': dst_type == mstype.float16, 'fp32': dst_type == mstype.float32,
'bf16': dst_type == mstype.bfloat16}
self._add_init_args(**flags)
self.add_flags_recursive(**flags)
return self
[docs] def set_boost(self, boost_type):
"""
In order to improve the network performance, configure the network auto enable to
accelerate the algorithm in the algorithm library.
If `boost_type` is not in the algorithm library, please view the algorithm in the algorithm library through
`algorithm library <https://gitee.com/mindspore/mindspore/tree/v2.4.0/mindspore/python/mindspore/boost>`_.
Note:
Some acceleration algorithms may affect the accuracy of the network, please choose carefully.
Args:
boost_type (str): accelerate algorithm.
Returns:
Cell, the cell itself.
Raises:
ValueError: If boost_type is not in the algorithm library.
"""
if boost_type not in ("less_bn",):
raise ValueError("For 'set_boost', the argument 'boost_type' must be 'less_bn', "
"but got {}.".format(boost_type))
flags = {"less_bn": boost_type == "less_bn"}
self.add_flags_recursive(**flags)
return self
[docs] def set_grad(self, requires_grad=True):
"""
Sets the cell flag for gradient. In pynative mode, this parameter specifies whether the network requires
gradients. If ``true`` , the backward network needed to compute the gradients will be generated when the forward
network is executed.
Args:
requires_grad (bool): Specifies if the net need to grad, if it is
``true`` , the cell will construct backward network in pynative mode. Default: ``True`` .
Returns:
Cell, the cell itself.
"""
self.requires_grad = requires_grad
return self
[docs] def set_train(self, mode=True):
"""
Sets the cell to training mode.
The cell itself and all children cells will be set to training mode. Layers that have different constructions
for training and predicting, such as `BatchNorm`, will distinguish between the branches by this attribute. If
set to true, the training branch will be executed, otherwise another branch.
Note:
When execute function Model.train(), framework will call Cell.set_train(True).
When execute function Model.eval(), framework will call Cell.set_train(False).
Args:
mode (bool): Specifies whether the model is training. Default: ``True`` .
Returns:
Cell, the cell itself.
Tutorial Examples:
- `Model Training - Implementing Training and Evaluation
<https://mindspore.cn/tutorials/en/r2.4.0/beginner/train.html#training-and-evaluation>`_
"""
if mode:
self._phase = 'train'
else:
self._phase = 'predict'
self.add_flags_recursive(training=mode)
return self
[docs] def set_broadcast_flag(self, mode=True):
"""
Set parameter broadcast mode for this cell.
Args:
mode (bool): Specifies whether the mode is parameter broadcast. Default: ``True`` .
"""
self.add_flags_recursive(broadcast_flag=mode)
return self
def set_auto_parallel(self):
"""
Set the cell to auto parallel mode.
Note:
This interface is deprecated.
"""
logger.warning("'set_auto_parallel' function is deprecated.")
[docs] def set_jit_config(self, jit_config):
"""
Set jit config for cell.
Args:
jit_config (JitConfig): Jit config for compile. For details, please refer to :class:`mindspore.JitConfig`.
Examples:
>>> import mindspore as ms
>>> from mindspore import Tensor, nn
...
>>> class Net(nn.Cell):
... def __init__(self):
... super(Net, self).__init__()
... self.relu = nn.ReLU()
...
... def construct(self, x):
... x = self.relu(x)
... return x
>>> net = Net()
>>> jitconfig = ms.JitConfig()
>>> net.set_jit_config(jitconfig)
"""
if self._jit_config_dict:
logger.warning("For Cell, jit config can only be set once, ignore this setting.")
else:
self._jit_config_dict = jit_config.jit_config_dict
[docs] def flatten_weights(self, fusion_size=0):
"""
Reset data for weight parameters so that they are using contiguous memory chunks grouped by data type.
Note:
By default, parameters with same data type will using a single contiguous memory chunk. but for
some models with huge number of parameters, splitting a large memory chunk into several smaller
memory chunks has the potential for performance gains, if this is the case, we can use 'fusion_size'
to limit the maximum memory chunk size.
Args:
fusion_size (int): Maximum memory chunk size in bytes, ``0`` for unlimited. Default: ``0`` .
"""
if fusion_size < 0:
raise ValueError(f"Negative 'fusion_size' {fusion_size} is invalid.")
Tensor._flatten_tensors(self.trainable_params(), fusion_size) # pylint: disable=W0212
[docs] def register_forward_pre_hook(self, hook_fn):
"""
Register forward pre hook function for Cell object.
Note:
- The `register_forward_pre_hook(hook_fn)` does not work in graph mode or functions decorated with 'jit'.
- 'hook_fn' must be defined as the following code.
`cell` is the object of registered Cell. `inputs` is the forward
input objects passed to the Cell. The 'hook_fn' can modify the forward input objects by returning new
forward input objects.
- It should have the following signature:
hook_fn(cell, inputs) -> new input objects or none.
- In order to prevent running failed when switching to graph mode, it is not recommended to write it in the
`construct` function of Cell object. In the pynative mode, if the `register_forward_pre_hook` function is
called in the `construct` function of the Cell object, a hook function will be added at each run time of
Cell object.
Args:
hook_fn (function): Python function. Forward pre hook function.
Returns:
A handle corresponding to the `hook_fn` . The handle can be used to remove the added `hook_fn` by calling
`handle.remove()` .
Raises:
TypeError: If the `hook_fn` is not a function of python.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import numpy as np
>>> import mindspore as ms
>>> from mindspore import Tensor, nn, ops
>>> ms.set_context(mode=ms.PYNATIVE_MODE)
>>> def forward_pre_hook_fn(cell, inputs):
... print("forward inputs: ", inputs)
...
>>> class Net(nn.Cell):
... def __init__(self):
... super(Net, self).__init__()
... self.mul = nn.MatMul()
... self.handle = self.mul.register_forward_pre_hook(forward_pre_hook_fn)
...
... def construct(self, x, y):
... x = x + x
... x = self.mul(x, y)
... return x
>>> grad = ops.GradOperation(get_all=True)
>>> net = Net()
>>> output = grad(net)(Tensor(np.ones([1]).astype(np.float32)), Tensor(np.ones([1]).astype(np.float32)))
forward inputs: (Tensor(shape=[1], dtype=Float32, value= [ 2.00000000e+00]), Tensor(shape=[1],
dtype=Float32, value= [ 1.00000000e+00]))
>>> print(output)
(Tensor(shape=[1], dtype=Float32, value= [ 2.00000000e+00]), Tensor(shape=[1], dtype=Float32,
value= [ 2.00000000e+00]))
"""
if context._get_mode() == context.GRAPH_MODE:
return HookHandle()
if not check_hook_fn("register_forward_pre_hook", hook_fn):
return HookHandle()
handle = HookHandle(self._forward_pre_hook)
self._forward_pre_hook[handle.handle_id] = hook_fn
return handle
def _run_forward_pre_hook(self, inputs):
"""
Running forward pre hook function registered on Cell object.
Args:
inputs: The input objects of cell object.
Returns:
- **outputs** - New input objects or none.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
"""
forward_pre_hook_inputs = inputs
for fn in self._forward_pre_hook.values():
ret = fn(self, forward_pre_hook_inputs)
if ret is not None:
if not isinstance(ret, tuple):
forward_pre_hook_inputs = (ret,)
else:
forward_pre_hook_inputs = ret
if isinstance(inputs, tuple):
if not isinstance(forward_pre_hook_inputs, tuple):
forward_pre_hook_inputs = (forward_pre_hook_inputs,)
if len(forward_pre_hook_inputs) != len(inputs):
raise TypeError(
"The forward pre hook return value size is {} not equal to input size {}".format(
len(forward_pre_hook_inputs), len(inputs)))
return forward_pre_hook_inputs
[docs] def register_forward_hook(self, hook_fn):
"""
Set the Cell forward hook function.
Note:
- The `register_forward_hook(hook_fn)` does not work in graph mode or functions decorated with 'jit'.
- 'hook_fn' must be defined as the following code.
`cell` is the object of registered Cell. `inputs` is the forward
input objects passed to the Cell. `output` is the forward output object of the Cell. The 'hook_fn' can
modify the forward output object by returning new forward output object.
- It should have the following signature:
hook_fn(cell, inputs, output) -> new output object or none.
- In order to prevent running failed when switching to graph mode, it is not recommended to write it in the
`construct` function of Cell object. In the pynative mode, if the `register_forward_hook` function is
called in the `construct` function of the Cell object, a hook function will be added at each run time of
Cell object.
Args:
hook_fn (function): Python function. Forward hook function.
Returns:
A handle corresponding to the `hook_fn` . The handle can be used to remove the added `hook_fn` by calling
`handle.remove()` .
Raises:
TypeError: If the `hook_fn` is not a function of python.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import numpy as np
>>> import mindspore as ms
>>> from mindspore import Tensor, nn, ops
>>> ms.set_context(mode=ms.PYNATIVE_MODE)
>>> def forward_hook_fn(cell, inputs, output):
... print("forward inputs: ", inputs)
... print("forward output: ", output)
...
>>> class Net(nn.Cell):
... def __init__(self):
... super(Net, self).__init__()
... self.mul = nn.MatMul()
... self.handle = self.mul.register_forward_hook(forward_hook_fn)
...
... def construct(self, x, y):
... x = x + x
... x = self.mul(x, y)
... return x
>>> grad = ops.GradOperation(get_all=True)
>>> net = Net()
>>> output = grad(net)(Tensor(np.ones([1]).astype(np.float32)), Tensor(np.ones([1]).astype(np.float32)))
forward inputs: (Tensor(shape=[1], dtype=Float32, value= [ 2.00000000e+00]), Tensor(shape=[1],
dtype=Float32, value= [ 1.00000000e+00]))
forward output: 2.0
>>> print(output)
(Tensor(shape=[1], dtype=Float32, value= [ 2.00000000e+00]), Tensor(shape=[1], dtype=Float32,
value= [ 2.00000000e+00]))
"""
if context._get_mode() == context.GRAPH_MODE:
return HookHandle()
if not check_hook_fn("register_forward_hook", hook_fn):
return HookHandle()
handle = HookHandle(self._forward_hook)
self._forward_hook[handle.handle_id] = hook_fn
return handle
def _run_forward_hook(self, inputs, output):
"""
Running forward hook function registered on Cell object.
Args:
inputs: The input objects of Cell object.
output: The output object of Cell object.
Returns:
- **output** - New output object or none.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
"""
forward_hook_output = output
for fn in self._forward_hook.values():
ret = fn(self, inputs, forward_hook_output)
if ret is not None:
forward_hook_output = ret
if isinstance(output, tuple):
if not isinstance(forward_hook_output, tuple):
forward_hook_output = (forward_hook_output,)
if len(forward_hook_output) != len(output):
raise TypeError(
"The forward hook return value size is {} not equal to output size {}".format(
len(forward_hook_output), len(output)))
return forward_hook_output
[docs] def register_backward_pre_hook(self, hook_fn):
"""
Register the backward pre hook function.
Note:
- The `register_backward_pre_hook(hook_fn)` does not work in graph mode or functions decorated with 'jit'.
- The 'hook_fn' must be defined as the following code.
`cell` is the Cell object. `grad_output` is the gradient passed to the Cell.
- The 'hook_fn' should have the following signature:
hook_fn(cell, grad_output) -> New grad_output gradient or None.
- The 'hook_fn' is executed in the python environment. In order to prevent running failed when switching to
graph mode, it is not recommended to write it in the `construct` function of Cell object.
- In the pynative
mode, if the `register_backward_pre_hook` function is called in the `construct` function of the Cell
object, a hook function will be added at each run time of Cell object.
Args:
hook_fn (function): Python function. Backward pre hook function.
Returns:
A handle corresponding to the `hook_fn` . The handle can be used to remove the added `hook_fn` by calling
`handle.remove()` .
Raises:
TypeError: If the `hook_fn` is not a function of python.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import numpy as np
>>> import mindspore as ms
>>> from mindspore import Tensor, nn, ops
>>> ms.set_context(mode=ms.PYNATIVE_MODE)
>>> def backward_pre_hook_fn(cell, grad_output):
... print("backward input: ", grad_output)
...
>>> class Net(nn.Cell):
... def __init__(self):
... super(Net, self).__init__()
... self.relu = nn.ReLU()
... self.handle = self.relu.register_backward_pre_hook(backward_pre_hook_fn)
...
... def construct(self, x):
... x = x + x
... x = self.relu(x)
... return x
>>> grad = ops.GradOperation(get_all=True)
>>> net = Net()
>>> output = grad(net)(Tensor(np.ones([1]).astype(np.float32)))
backward input: (Tensor(shape=[1], dtype=Float32, value= [ 1.00000000e+00]),)
>>> print(output)
(Tensor(shape=[1], dtype=Float32, value= [ 2.00000000e+00]),)
"""
if context._get_mode() == context.GRAPH_MODE:
return HookHandle()
if not check_hook_fn("register_backward_pre_hook", hook_fn):
return HookHandle()
handle = HookHandle(self._backward_pre_hook)
self._backward_pre_hook[handle.handle_id] = hook_fn
if self._cell_backward_pre_hook is None:
# Generate a CellBackwardHook prim, and add function for it
self._cell_backward_pre_hook = inner.CellBackwardHook(self.cls_name + "(" + str(id(self)) + ")",
self, self._backward_pre_hook)
self._cell_backward_pre_hook.register_backward_pre_hook()
return handle
def _run_backward_pre_hook(self, outputs):
"""
Running backward pre hook function registered on Cell object.
Args:
outputs: The output objects of cell object.
Returns:
- **outputs** - New backward gradient or None.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
"""
ret = self._cell_backward_pre_hook(outputs)
if isinstance(outputs, tuple):
if not isinstance(ret, tuple):
ret = (ret,)
if len(ret) != len(outputs):
raise TypeError(
"The backward pre hook return value size is {} not equal to output size {}".format(
len(ret), len(outputs)))
return ret
[docs] def register_backward_hook(self, hook_fn):
"""
Register the backward hook function.
Note:
- The `register_backward_hook(hook_fn)` does not work in graph mode or functions decorated with 'jit'.
- The 'hook_fn' must be defined as the following code.
`cell` is the registered Cell object. `grad_input` is the gradient computed and passed to
the next Cell or primitive, which can be return a new gradient or None. `grad_output` is the gradient
passed to the Cell.
- The 'hook_fn' should have the following signature:
hook_fn(cell, grad_input, grad_output) -> New grad_input gradient or none.
- The 'hook_fn' is executed in the python environment. In order to prevent running failed when switching to
graph mode, it is not recommended to write it in the `construct` function of Cell object. In the pynative
mode, if the `register_backward_hook` function is called in the `construct` function of the Cell object,
a hook function will be added at each run time of Cell object.
Args:
hook_fn (function): Python function. Backward hook function.
Returns:
A handle corresponding to the `hook_fn` . The handle can be used to remove the added `hook_fn` by calling
`handle.remove()` .
Raises:
TypeError: If the `hook_fn` is not a function of python.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import numpy as np
>>> import mindspore as ms
>>> from mindspore import Tensor, nn, ops
>>> ms.set_context(mode=ms.PYNATIVE_MODE)
>>> def backward_hook_fn(cell, grad_input, grad_output):
... print("backward input: ", grad_output)
... print("backward output: ", grad_input)
...
>>> class Net(nn.Cell):
... def __init__(self):
... super(Net, self).__init__()
... self.relu = nn.ReLU()
... self.handle = self.relu.register_backward_hook(backward_hook_fn)
...
... def construct(self, x):
... x = x + x
... x = self.relu(x)
... return x
>>> grad = ops.GradOperation(get_all=True)
>>> net = Net()
>>> output = grad(net)(Tensor(np.ones([1]).astype(np.float32)))
backward input: (Tensor(shape=[1], dtype=Float32, value= [ 1.00000000e+00]),)
backward output: (Tensor(shape=[1], dtype=Float32, value= [ 1.00000000e+00]),)
>>> print(output)
(Tensor(shape=[1], dtype=Float32, value= [ 2.00000000e+00]),)
"""
if context._get_mode() == context.GRAPH_MODE:
return HookHandle()
if not check_hook_fn("register_backward_hook", hook_fn):
return HookHandle()
handle = HookHandle(self._backward_hook)
self._backward_hook[handle.handle_id] = hook_fn
if self._cell_backward_hook is None:
# Generate a CellBackwardHook prim, and add function for it
self._cell_backward_hook = inner.CellBackwardHook(self.cls_name + "(" + str(id(self)) + ")",
self, self._backward_hook)
self._cell_backward_hook.register_backward_hook()
return handle
def _backward_hook_construct(self, *inputs, **kwargs):
"""
Backward hook construct method to replace original construct method.
Args:
inputs: The input objects of Cell object.
kwargs (dict): Dictionary of variable keyword parameters.
Returns:
- **outputs** - The output objects of Cell object.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
"""
# cell_backward_hook has CellBackwardHook op, so keep input args as they are.
outputs = self._cell_backward_hook(*inputs)
# If the inputs have more than two args, the outputs will also have more than two args and will be wrapped into
# a tuple, so need to do unwrapping. If inputs is empty, we also need to unwrap it.
# Because when output of runop method is one, it will not wrap a tuple, we need not unwrap it.
is_need_unwrap = False
if isinstance(outputs, tuple) and len(inputs) != 1:
is_need_unwrap = True
if self._recompute_cell is not None:
if is_need_unwrap:
outputs = self._recompute_cell(*outputs, **kwargs)
else:
outputs = self._recompute_cell(outputs, **kwargs)
elif self.has_bprop:
if is_need_unwrap:
outputs = self._call_custom_bprop(*outputs, **kwargs)
else:
outputs = self._call_custom_bprop(outputs, **kwargs)
else:
if is_need_unwrap:
outputs = self.construct(*outputs, **kwargs)
else:
outputs = self.construct(outputs, **kwargs)
outputs = self._cell_backward_hook(outputs)
return outputs
[docs] def set_param_ps(self, recurse=True, init_in_server=False):
"""
Set whether the trainable parameters are updated by parameter server and whether the
trainable parameters are initialized on server.
Note:
It only works when a running task is in the parameter server mode.
It is only supported in graph mode.
Args:
recurse (bool): Whether sets the trainable parameters of subcells. Default: ``True`` .
init_in_server (bool): Whether trainable parameters updated by parameter server are
initialized on server. Default: ``False`` .
"""
params = self.trainable_params(recurse)
for param in params:
param.set_param_ps(init_in_server)
@deprecated("1.8", "set_param_fl")
def set_param_fl(self, push_to_server=False, pull_from_server=False, requires_aggr=True):
params = self.parameters_and_names()
for param in params:
param[1].set_param_fl(push_to_server, pull_from_server, requires_aggr)
[docs] def set_comm_fusion(self, fusion_type, recurse=True):
"""
Set `comm_fusion` for all the parameters in this cell. Please refer to the description of
:class:`mindspore.Parameter.comm_fusion`.
Note:
The value of attribute will be overwritten when the function is called multiply.
Args:
fusion_type (int): The value of `comm_fusion`.
recurse (bool): Whether sets the trainable parameters of subcells. Default: ``True`` .
"""
Validator.check_non_negative_int(fusion_type)
for param in self.trainable_params(recurse):
param.comm_fusion = fusion_type
return self
def _set_recompute_scope(self, mode):
prefix = 'recompute_'
if mode:
if self._scope is None:
self._scope = prefix
elif not self._scope.startswith(prefix):
self._scope = prefix + self._scope
elif self._scope is not None and self._scope.startswith(prefix):
self._scope = self._scope[len(prefix):]
def _mp_comm_recompute(self, mp_comm_recompute=True):
"""
Set the model parallel communication in cell recomputed.
"""
for _, value in self._primitives.items():
if value:
value.add_prim_attr("recompute_comm_op", mp_comm_recompute)
for cell in self.cells():
cell._mp_comm_recompute(mp_comm_recompute)
def _parallel_optimizer_comm_recompute(self, parallel_optimizer_comm_recompute=False):
"""
Set the parallel optimizer communication in cell recomputed.
"""
for param in self.trainable_params():
param.parallel_optimizer_comm_recompute = parallel_optimizer_comm_recompute
def _recompute_slice_activation(self, slice_activation=False):
"""
Slice the cell output which would remains in memory.
"""
for _, value in self._primitives.items():
if value:
value.add_prim_attr("slice_activation", slice_activation)
for cell in self.cells():
cell._recompute_slice_activation(slice_activation)
def _recompute(self, mode=True, output_recompute=False):
"""
Set the cell recomputed.
"""
Validator.check_bool(mode)
Validator.check_bool(output_recompute)
if not self._has_config_recompute:
self._has_config_recompute = True
else:
raise RuntimeError("The recompute interface can be configured only once."
" When the parent cell is configured, the child cell should not be configured")
self._set_recompute_scope(mode)
if mode and not output_recompute:
self.add_flags(output_no_recompute=True)
for cell in self.cells():
cell._recompute(mode, True)
[docs] @args_type_check(mp_comm_recompute=bool, parallel_optimizer_comm_recompute=bool)
def recompute(self, **kwargs):
"""
Set the cell recomputed. All the primitive in the cell except the outputs will be set recomputed.
If a primitive set recomputed feeds into some backward nodes for computing gradient, rather than
storing the intermediate activation computed in forward pass, we will recompute it in backward pass.
Note:
- If the computation involves something like randomization or global variable, the equivalence
is not guaranteed currently.
- If the recompute api of a primitive in this cell is also called, the recompute mode of this
primitive is subject to the recompute api of the primitive.
- The interface can be configured only once.
Therefore, when the parent cell is configured, the child cell should not be configured.
- The outputs of cell are excluded from recomputation by default, which is based on our configuration
experience to reduce memory footprint. If a cell has only one primitive and the primitive is wanted
to be set recomputed, use the recompute api of the primtive.
- When the memory remains after applying the recomputation, configuring 'mp_comm_recompute=False'
to improve performance if necessary.
- When the memory still not enough after applying the recompute, configuring
'parallel_optimizer_comm_recompute=True' to save more memory if necessary.
Cells in the same fusion group should have the same parallel_optimizer_comm_recompute configures.
Args:
mp_comm_recompute (bool): Specifies whether the model parallel communication operators
in the cell are recomputed in auto parallel or semi auto parallel mode. Default: ``True`` .
parallel_optimizer_comm_recompute (bool): Specifies whether the communication operator allgathers
introduced by optimizer shard are recomputed in auto parallel or semi auto parallel mode.
Default: ``False`` .
"""
if context.get_context("mode") == context.PYNATIVE_MODE:
self._recompute_cell = recompute_registry.get()(self.construct)
return
self._recompute()
if 'mp_comm_recompute' in kwargs.keys():
self._mp_comm_recompute(kwargs.get('mp_comm_recompute', False))
if 'parallel_optimizer_comm_recompute' in kwargs.keys():
if (kwargs.get('parallel_optimizer_comm_recompute', False) and
context.get_auto_parallel_context("pipeline_stages") > 1):
logger.warning("Currently, the communication operator allgathers introduced by optimizer shard "
"are not support recomputation in pipeline parallel.")
elif context.get_auto_parallel_context("pipeline_stages") == 1:
self._parallel_optimizer_comm_recompute(kwargs.get('parallel_optimizer_comm_recompute', False))
if 'recompute_slice_activation' in kwargs:
self._recompute_slice_activation(kwargs.get('recompute_slice_activation', False))
for key, _ in kwargs.items():
if key not in ('mp_comm_recompute', 'parallel_optimizer_comm_recompute', 'recompute_slice_activation'):
raise ValueError("For 'recompute', keyword '%s' is not recognized! "
"the key kwargs must be 'mp_comm_recompute', "
"'parallel_optimizer_comm_recompute', 'recompute_slice_activation'" % key)
[docs] @deprecated("2.3", "infer_param_pipeline_stage")
def infer_param_pipeline_stage(self):
"""
Infer pipeline stages of all parameters in the cell.
Note:
- The interface is deprecated from version 2.3 and will be removed in a future version.
Returns:
The params belong to current stage in pipeline parallel.
Raises:
RuntimeError: If there is a parameter does not belong to any stage.
"""
from mindspore.parallel._utils import _get_global_rank, _get_device_num
logger.warning(f"This interface may be deleted in the future.")
stage_num = context.get_auto_parallel_context("pipeline_stages")
device_num = _get_device_num()
rank_id = _get_global_rank()
per_stage_devices = device_num // stage_num
current_stage = rank_id // per_stage_devices
params = []
for param in self.trainable_params():
if not param._pipeline_stage_list: # pylint: disable=W0212
raise RuntimeError("For 'infer_param_pipeline_stage', the parameter {} does not belong to any stage, "
"please check whether the cell where the param locates has been set "
"'pipeline_stage'. Otherwise, the parameter should use 'add_pipeline_stage' "
"to add its stage information".format(param.name))
if current_stage in param._pipeline_stage_list:
params.append(param)
return params
[docs] def place(self, role, rank_id):
"""
Set the label for all operators in this cell.
This label tells MindSpore compiler on which process this cell should be launched.
And each process's identical label consists of input `role` and `rank_id`.
So by setting different cells with different labels, which will be launched on different processes,
users can launch a distributed training or predicting job.
Note:
- This method is effective only after
`mindspore.communication.init()` is called for dynamic cluster building.
Args:
role (str): The role of the process on which this cell will be launched.
Only 'MS_WORKER' is supported for now.
rank_id (int): The rank id of the process on which this cell will be launched.
The rank is unique in processes with the same role.
Examples:
>>> from mindspore import context
>>> import mindspore.nn as nn
>>> context.set_context(mode=context.GRAPH_MODE)
>>> fc = nn.Dense(2, 3)
>>> fc.place('MS_WORKER', 0)
"""
all_ops = self._get_prims_recursively()
for op in all_ops:
op.place(role, rank_id)
def _mixed_precision_cast(self, inputs):
mixed_type = self.get_mixed_precision_type()
if mixed_type == MixedPrecisionType.NOTSET:
return inputs
if mixed_type == MixedPrecisionType.FP16:
cast_type = mstype.float16
elif mixed_type == MixedPrecisionType.BF16:
cast_type = mstype.bfloat16
else:
cast_type = mstype.float32
cast_inputs = self._cast_mixed_precision_inputs(inputs, cast_type)
return cast_inputs
def _get_attr_from_cell(self, network):
if not isinstance(network, Cell):
return
if hasattr(network, "jit_config_dict"):
self._jit_config_dict = network.jit_config_dict
if hasattr(network, "_amp_level"):
self._amp_level = getattr(network, "_amp_level")
[docs]class GraphCell(Cell):
"""
Base class for running the graph loaded from a MindIR.
This feature is still under development. Currently `GraphCell` do not support modifying the structure of the
diagram, and can only use data that shape and type are the same as the input when exporting the MindIR.
Args:
graph (FuncGraph): A compiled graph loaded from MindIR.
params_init (dict): Parameters need to be inited in the graph.
The key is the parameter name whose type is str, and the value is a Tensor or Parameter.
If the parameter exists in the graph according to the name, update it's value.
If the parameter does not exist, ignore it. Default: ``None`` .
obf_random_seed (Union[int, None]): The random seed used for dynamic obfuscation. "dynamic obfuscation" is
used for model protection, which can refer to :func:`mindspore.obfuscate_model`. If the input `graph` is
a func_graph loaded from a mindir file obfuscated with `obf_random_seed` , then `obf_random_seed` should be
provided. `obf_random_seed` should be in (0, 9223372036854775807]. default: ``None`` .
Raises:
TypeError: If the `graph` is not a FuncGraph.
TypeError: If the `params_init` is not a dict.
TypeError: If the key of the `params_init` is not a str.
TypeError: If the value of the `params_init` is neither a Tensor nor a Parameter.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import numpy as np
>>> import mindspore as ms
>>> import mindspore.nn as nn
>>> from mindspore import Tensor
>>> from mindspore import context
>>> context.set_context(mode=context.GRAPH_MODE)
>>> net = nn.Conv2d(1, 1, kernel_size=3, weight_init="ones")
>>> input = Tensor(np.ones([1, 1, 3, 3]).astype(np.float32))
>>> ms.export(net, input, file_name="net", file_format="MINDIR")
>>> graph = ms.load("net.mindir")
>>> net = nn.GraphCell(graph)
>>> output = net(input)
>>> print(output)
[[[[4. 6. 4.]
[6. 9. 6.]
[4. 6. 4.]]]]
"""
def __init__(self, graph, params_init=None, obf_random_seed=None):
super(GraphCell, self).__init__(auto_prefix=True)
if not isinstance(graph, FuncGraph):
raise TypeError(f"For 'GraphCell', the argument 'graph' must be a FuncGraph loaded from MindIR, "
f"but got type {type(graph)}.")
self.graph = graph
self.obf_random_seed = obf_random_seed
if obf_random_seed is not None:
if not isinstance(obf_random_seed, int):
raise TypeError("'obf_random_seed' must be int, but got {}.".format(type(obf_random_seed)))
int_64_max = 9223372036854775807
if obf_random_seed <= 0 or obf_random_seed > int_64_max:
raise ValueError(
"'obf_random_seed' must be larger than 0, and less or equal than int64 ({}),"
"but got {}.".format(int_64_max, obf_random_seed))
self._branch_control_input = _generate_branch_control_input(self.obf_random_seed)
params_init = {} if params_init is None else params_init
if not isinstance(params_init, dict):
raise TypeError(f"For 'GraphCell', the argument 'params_init' must be a dict, but got {type(params_init)}.")
for name, value in params_init.items():
if not isinstance(name, str) or not isinstance(value, Tensor):
raise TypeError("For 'GraphCell', the key of the 'params_init' must be str, "
"and the value must be Tensor or Parameter, "
f"but got the key type: {type(name)}, and the value type: {type(value)}")
params_dict = update_func_graph_hyper_params(self.graph, params_init)
for name, param in params_dict.items():
self._params[name] = param
def construct(self, *inputs):
return self.graph(*inputs)
def __call__(self, *args, **kwargs):
self.phase = "graph_load_from_mindir"
self._add_attr("graph_load_from_mindir", self.graph)
if not self.obf_random_seed:
return self.compile_and_run(*args, **kwargs)
append_input = Tensor((numpy.ones((1,)) * self._branch_control_input).astype(numpy.int32))
return self.compile_and_run(*args, append_input, **kwargs)
def _check_param_list_tuple(value):
"""
Check the type of input in list or tuple is Parameter.
:param value: list or tuple.
:return: The types of all inputs are parameter.
"""
for item in value:
if not isinstance(item, Parameter):
return False
return True