# Copyright 2023 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.
# ============================================================================
"""rprop"""
from __future__ import absolute_import
from mindspore.ops import functional as F, composite as C, operations as P
from mindspore.common import Tensor, Parameter
import mindspore.common.dtype as mstype
from mindspore import _checkparam as validator
from mindspore.experimental.optim.optimizer import Optimizer, check_not_less_than_without_equal
from mindspore import ops
from mindspore import jit
_rprop_opt = C.MultitypeFuncGraph("rprop_opt")
op_sign = P.Sign()
op_fill = P.FillV2()
op_assign = P.Assign()
op_assignadd = P.AssignAdd()
op_cast = P.Cast()
op_select = P.Select()
op_oneslike = P.OnesLike()
@_rprop_opt.register("Tensor", "Tensor", "Number", "Number", "Tensor", "Tensor", "Tensor", "Tensor", "Tensor", "Tensor")
def _tensor_run_opt(etaminus, etaplus, step_size_min, step_size_max, step, lr, param, prev, step_size, gradient):
"""Apply rprop optimizer to the weight parameter."""
if step == 1:
step_size_value = op_oneslike(step_size) * lr
else:
step_size_value = step_size.value()
sign = op_sign(gradient * prev)
sign[sign.gt(0)] = etaplus
sign[sign.lt(0)] = etaminus
sign[sign.eq(0)] = 1
step_size_clip = ops.clip_by_value(step_size_value * sign, step_size_min, step_size_max)
op_assign(step_size, step_size_clip)
gradient_update = op_select(sign == etaminus, op_fill(sign.shape, op_cast(0., mstype.float32)), gradient)
op_assign(prev, gradient_update)
next_param = param - op_sign(gradient_update) * step_size_clip
op_assign(param, next_param)
return True
[文档]class Rprop(Optimizer):
r"""
Implements Rprop algorithm.
.. warning::
This is an experimental optimizer API that is subject to change.
This module must be used with lr scheduler module in `LRScheduler Class
<https://www.mindspore.cn/docs/en/r2.3.0rc2/api_python/mindspore.experimental.html#lrscheduler-class>`_ .
Args:
params (Union[list(Parameter), list(dict)]): list of parameters to optimize or dicts defining
parameter groups.
lr (Union[int, float, Tensor], optional): learning rate. Default: ``1e-2``.
etas (Tuple[float, float], optional): pair of (etaminus, etaplus), that
are multiplicative increase and decrease factors. Default:``(0.5, 1.2)``
step_sizes (Tuple[float, float], optional): a pair of minimal and
maximal allowed step sizes. Default:``(1e-6, 50)``
Keyword Args:
maximize (bool, optional): maximize the params based on the objective, instead of minimizing.
Default: ``False``.
Inputs:
- **gradients** (tuple[Tensor]) - The gradients of `params`.
Raises:
ValueError: If the learning rate is not int, float or Tensor.
ValueError: If the learning rate is less than 0.
ValueError: If the `etas[1]` is less than or equal to 1.0.
ValueError: If the `etas[0]` not in the range of 0-1.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore
>>> from mindspore import nn
>>> from mindspore.experimental import optim
>>> # Define the network structure of LeNet5. Refer to
>>> # https://gitee.com/mindspore/docs/blob/r2.3.0rc2/docs/mindspore/code/lenet.py
>>> net = LeNet5()
>>> loss_fn = nn.SoftmaxCrossEntropyWithLogits(sparse=True)
>>> optimizer = optim.Rprop(net.trainable_params(), lr=0.1)
>>> def forward_fn(data, label):
... logits = net(data)
... loss = loss_fn(logits, label)
... return loss, logits
>>> grad_fn = mindspore.value_and_grad(forward_fn, None, optimizer.parameters, has_aux=True)
>>> def train_step(data, label):
... (loss, _), grads = grad_fn(data, label)
... optimizer(grads)
... return loss
"""
def __init__(self, params, lr=1e-2, etas=(0.5, 1.2), step_sizes=(1e-6, 50), *, maximize=False):
check_not_less_than_without_equal(lr, "lr", self.cls_name)
check_not_less_than_without_equal(etas[1], "etas[1]", self.cls_name, 1.)
validator.check_float_range(etas[0], 0., 1., validator.INC_NEITHER, "etas[0]", self.cls_name)
defaults = dict(
lr=lr,
etas=etas,
step_sizes=step_sizes,
maximize=maximize,
)
super(Rprop, self).__init__(params, defaults)
self.prev = self.parameters.clone(prefix="prev", init='zeros')
self.step_size = self.parameters.clone(prefix="step_size", init='zeros')
self.step_t = Parameter(Tensor(0, mstype.int32), "step_t")
self.increase_tensor = Tensor(1, mstype.int32)
self.op_cast = P.Cast()
@jit
def implementation(self, etaminus, etaplus, group_id, lr, gradients, maximize, step_size_min, step_size_max):
"""Extract the common computing part for acceleration"""
etaminus, etaplus = op_cast(etaminus, mstype.float32), op_cast(etaplus, mstype.float32)
start_id = self.group_start_id[group_id]
end_id = self.group_start_id[group_id + 1]
params = self.parameters[start_id: end_id]
grads = tuple([grad if not maximize else F.neg(grad) for grad in gradients[start_id: end_id]])
prev = self.prev[start_id: end_id]
step_size = self.step_size[start_id: end_id]
self.hyper_map(F.partial(_rprop_opt, etaminus, etaplus, step_size_min, step_size_max, self.step_t, lr),
params, prev, step_size, grads)
return True
def construct(self, gradients):
op_assignadd(self.step_t, self.increase_tensor)
for group_id, group in enumerate(self.param_groups):
lr = self.lrs[group_id]
if isinstance(group.get("lr"), float):
lr = self.op_cast(group.get("lr"), mstype.float32)
maximize = group.get("maximize")
etaminus, etaplus = group["etas"]
step_size_min, step_size_max = group["step_sizes"]
self.implementation(etaminus, etaplus, group_id, lr, gradients, maximize, step_size_min, step_size_max)
return True