# 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.
# ============================================================================
"""FusedAdamWeightDecay, a customized Adam for offloading."""
import numpy as np
from mindspore import context, nn
from mindspore.common import dtype as mstype
from mindspore.ops import operations as P
from mindspore.ops import composite as C
from mindspore.ops import functional as F
from mindspore.common.initializer import initializer
from mindspore.common.tensor import Tensor
from mindspore.common.parameter import Parameter, ParameterTuple
# MindSpore 2.0 has changed the APIs of _checkparam, the following try except is for compatibility
try:
from mindspore._checkparam import Validator as validator
from mindspore._checkparam import Rel
except ImportError:
import mindspore._checkparam as validator
import mindspore._checkparam as Rel
from mindspore.nn.optim.optimizer import Optimizer
from mindformers.version_control import fix_optim_global_step_sig
from mindformers.tools.register import MindFormerRegister, MindFormerModuleType
__all__ = ['FusedAdamWeightDecay', 'FP32StateAdamWeightDecay']
_adam_opt = C.MultitypeFuncGraph("adam_opt")
_scaler_one = Tensor(1, mstype.int32)
op_mul = P.Mul()
@_adam_opt.register("Function", "Tensor", "Tensor", "Tensor", "Tensor", "Tensor", "Tensor", "Tensor", "Tensor",
"Tensor", "Tensor", "Bool", "Bool")
def _fused_update_with_global_norm(opt, global_norm, beta1, beta2, eps, lr, weight_decay,
param, m, v, gradient, decay_flags, optim_filter):
"""
Update parameters by FusedAdamWeightDecay.
"""
success = True
if optim_filter:
if decay_flags:
next_param = opt(param, m, v, lr, beta1, beta2, eps, weight_decay,
P.Cast()(gradient, mstype.float16), global_norm)
else:
next_param = opt(param, m, v, lr, beta1, beta2, eps, 0.0,
P.Cast()(gradient, mstype.float16), global_norm)
return F.depend(success, next_param)
return success
@_adam_opt.register("Function", "Tensor", "Tensor", "Tensor", "Tensor", "Tensor", "Tensor", "Tensor",
"Tensor", "Tensor", "Bool", "Bool")
def _fused_update(opt, beta1, beta2, eps, lr, weight_decay,
param, m, v, gradient, decay_flags, optim_filter):
"""
Update parameters by FusedAdamWeightDecay.
"""
success = True
op_cast = P.Cast()
if optim_filter:
if decay_flags:
opt(param, m, v, lr, beta1, beta2, eps, weight_decay, op_cast(gradient, F.dtype(param)))
else:
opt(param, m, v, lr, beta1, beta2, eps, 0.0, op_cast(gradient, F.dtype(param)))
return success
@_adam_opt.register("Tensor", "Tensor", "Tensor", "Tensor", "Tensor", "Tensor", "Tensor", "Tensor", "Tensor", "Tensor",
"Tensor", "Bool", "Bool")
def _update_run_op(beta1_power, beta2_power, beta1, beta2, eps, lr, weight_decay, param, \
m, v, gradient, decay_flag, optim_filter):
"""
Update parameters.
Args:
beta1 (Tensor): The exponential decay rate for the 1st moment estimations. Should be in range (0.0, 1.0).
beta2 (Tensor): The exponential decay rate for the 2nd moment estimations. Should be in range (0.0, 1.0).
eps (Tensor): Term added to the denominator to improve numerical stability. Should be greater than 0.
lr (Tensor): Learning rate.
weight_decay (Number): Weight decay. Should be equal to or greater than 0.
param (Tensor): Parameters.
m (Tensor): m value of parameters.
v (Tensor): v value of parameters.
gradient (Tensor): Gradient of parameters.
decay_flag (bool): Applies weight decay or not.
optim_filter (bool): Applies parameter update or not.
Returns:
Tensor, the new value of v after updating.
"""
if optim_filter:
# op_mul = P.Mul(), defined output
op_square = P.Square()
op_sqrt = P.Sqrt()
op_cast = P.Cast()
op_reshape = P.Reshape()
op_shape = P.Shape()
param_fp32 = op_cast(param, mstype.float32)
m_fp32 = op_cast(m, mstype.float32)
v_fp32 = op_cast(v, mstype.float32)
gradient_fp32 = op_cast(gradient, mstype.float32)
next_m = op_mul(beta1, m_fp32) + op_mul(op_cast(F.tuple_to_array((1.0,)), mstype.float32)
- beta1, gradient_fp32)
next_v = op_mul(beta2, v_fp32) + op_mul(op_cast(F.tuple_to_array((1.0,)), mstype.float32)
- beta2, op_square(gradient_fp32))
regulate_m = next_m / (_scaler_one - beta1_power)
regulate_v = next_v / (_scaler_one - beta2_power)
update = regulate_m / (eps + op_sqrt(regulate_v))
if decay_flag:
update = op_mul(weight_decay, param_fp32) + update
update_with_lr = op_mul(lr, update)
next_param = param_fp32 - op_reshape(update_with_lr, op_shape(param_fp32))
next_param = F.depend(next_param, F.assign(param, op_cast(next_param, F.dtype(param))))
next_param = F.depend(next_param, F.assign(m, op_cast(next_m, F.dtype(m))))
next_param = F.depend(next_param, F.assign(v, op_cast(next_v, F.dtype(v))))
return op_cast(next_param, F.dtype(param))
return gradient
def _check_param_value(beta1, beta2, eps, prim_name):
"""Check the type of inputs."""
validator.check_value_type("beta1", beta1, [float], prim_name)
validator.check_value_type("beta2", beta2, [float], prim_name)
validator.check_value_type("eps", eps, [float], prim_name)
validator.check_float_range(beta1, 0.0, 1.0, Rel.INC_NEITHER, "beta1", prim_name)
validator.check_float_range(beta2, 0.0, 1.0, Rel.INC_NEITHER, "beta2", prim_name)
validator.check_positive_float(eps, "eps", prim_name)
[文档]@MindFormerRegister.register(MindFormerModuleType.OPTIMIZER)
class FusedAdamWeightDecay(Optimizer):
"""
Implements the Adam algorithm to fix the weight decay. It is a complete operator, not a combination of other ops.
Note:
When separating parameter groups, the weight decay in each group will be applied on the parameters if the
weight decay is positive. When not separating parameter groups, the `weight_decay` in the API will be applied
on the parameters without 'beta' or 'gamma' in their names if `weight_decay` is positive.
To improve parameter groups performance, the customized order of parameters can be supported.
Args:
params (Union[list[Parameter], list[dict]]): When the `params` is a list of `Parameter` which will be updated,
the element in `params` must be class `Parameter`. When the `params` is a list of `dict`, the "params",
"lr", "weight_decay" and "order_params" are the keys can be parsed.
- params: Required. The value must be a list of `Parameter`.
- lr: Optional. If "lr" is in the keys, the value of the corresponding learning rate will be used.
If not, the `learning_rate` in the API will be used.
- weight_decay: Optional. If "weight_decay" is in the keys, the value of the corresponding weight decay
will be used. If not, the `weight_decay` in the API will be used.
- order_params: Optional. If "order_params" is in the keys, the value must be the order of parameters and
the order will be followed in the optimizer. There are no other keys in the `dict` and the parameters
which in the 'order_params' must be in one of group parameters.
learning_rate (Union[float, Tensor, Iterable, LearningRateSchedule]): A value or a graph for the learning rate.
When the learning_rate is an Iterable or a Tensor in a 1D dimension, use the dynamic learning rate, then
the i-th step will take the i-th value as the learning rate. When the learning_rate is LearningRateSchedule,
use dynamic learning rate, the i-th learning rate will be calculated during the process of training
according to the formula of LearningRateSchedule. When the learning_rate is a float or a Tensor in a zero
dimension, use fixed learning rate. Other cases are not supported. The float learning rate must be
equal to or greater than 0. If the type of `learning_rate` is int, it will be converted to float.
Default: 1e-3.
beta1 (float): The exponential decay rate for the 1st moment estimations. Default: 0.9.
Should be in range (0.0, 1.0).
beta2 (float): The exponential decay rate for the 2nd moment estimations. Default: 0.999.
Should be in range (0.0, 1.0).
eps (float): Term added to the denominator to improve numerical stability. Default: 1e-6.
Should be greater than 0.
weight_decay (float): Weight decay (L2 penalty). It must be equal to or greater than 0. Default: 0.0.
Inputs:
- **gradients** (tuple[Tensor]) - The gradients of `params`, the shape is the same as `params`.
Outputs:
tuple[bool], all elements are True.
Supported Platforms:
``CPU``
Examples:
>>> net = Net()
>>> #1) All parameters use the same learning rate and weight decay
>>> optim = FusedAdamWeightDecay(params=net.trainable_params())
>>>
>>> #2) Use parameter groups and set different values
>>> conv_params = list(filter(lambda x: 'conv' in x.name, net.trainable_params()))
>>> no_conv_params = list(filter(lambda x: 'conv' not in x.name, net.trainable_params()))
>>> group_params = [{'params': conv_params, 'weight_decay': 0.01},
... {'params': no_conv_params, 'lr': 0.01},
... {'order_params': net.trainable_params()}]
>>> optim = FusedAdamWeightDecay(group_params, learning_rate=0.1, weight_decay=0.0)
>>> # The conv_params's parameters will use default learning rate of 0.1 and weight decay of 0.01.
>>> # The no_conv_params's parameters will use learning rate of 0.01 and default weight decay of 0.0.
>>> # The final parameters order in which the optimizer will be followed is the value of 'order_params'.
>>>
>>> loss = nn.SoftmaxCrossEntropyWithLogits()
>>> model = Model(net, loss_fn=loss, optimizer=optim)
"""
def __init__(self, params, learning_rate=1e-3, beta1=0.9, beta2=0.999, eps=1e-6, weight_decay=0.0,
offload=False):
super(FusedAdamWeightDecay, self).__init__(learning_rate, params, weight_decay)
_check_param_value(beta1, beta2, eps, self.cls_name)
self.beta1 = Tensor(np.array([beta1]).astype(np.float32))
self.beta2 = Tensor(np.array([beta2]).astype(np.float32))
self.eps = Tensor(np.array([eps]).astype(np.float32))
self.moments1 = self.clone_state(prefix="adam_m", init='zeros')
self.moments2 = self.clone_state(prefix="adam_v", init='zeros')
self.opt = P.AdamWeightDecay()
if offload:
self.opt.add_prim_attr("primitive_target", "CPU")
self.version_flag = fix_optim_global_step_sig()
def construct(self, gradients):
"""construct with gradients"""
lr = self.get_lr()
if self.version_flag:
self.assignadd(self.global_step, self.global_step_increase_tensor)
if self.is_group:
if self.is_group_lr:
optim_result = self.map_reverse(F.partial(_adam_opt, self.opt,
self.beta1, self.beta2, self.eps),
lr, self.weight_decay, self._parameters, self.moments1, self.moments2,
gradients, self.decay_flags, self.optim_filter)
else:
optim_result = self.map_reverse(F.partial(_adam_opt, self.opt,
self.beta1, self.beta2, self.eps, lr),
self.weight_decay, self._parameters, self.moments1, self.moments2,
gradients, self.decay_flags, self.optim_filter)
else:
optim_result = self.map_reverse(F.partial(_adam_opt, self.opt,
self.beta1, self.beta2, self.eps, lr,
self.weight_decay), self._parameters, self.moments1,
self.moments2,
gradients, self.decay_flags, self.optim_filter)
if self.use_parallel:
self.broadcast_params(optim_result)
return optim_result
[文档] def clone_state(self, prefix, init, forced_dtype=mstype.float32, is_follow=False):
r"""
Clone the parameters
parameter_tuple: ParameterTuple. The parameters of the network
prefix: str. The prefix name of the parameters
init: str. The initialization method
forced_dtype: mstype. The except the dtype to be cloned. If is_follow is True, forced_dtype will be ignored.
Default: mstype.float32
is_follow: bool. Is clone the parameters with the original dtype. If is_follow is True, the forced_dtype
argument will be ignored. Default: False.
"""
parameter_tuple = self.parameters
new = []
for old_param in parameter_tuple:
param_init = init
if init is None:
param_init = old_param.init
cur_dtype = forced_dtype
if is_follow:
cur_dtype = old_param.dtype
new_state = Parameter(initializer(param_init, shape=old_param.shape, dtype=cur_dtype))
new_state.param_info = old_param.param_info.clone()
if hasattr(old_param.param_info, "cloned_obj"):
old_param.param_info.cloned_obj.append(new_state)
else:
old_param.param_info.cloned_obj = [new_state]
new_state.is_init = False
new_state.is_param_ps = old_param.is_param_ps
new_state.init_in_server = old_param.init_in_server
new_state.cache_enable = old_param.cache_enable
new_state.requires_aggr = old_param.requires_aggr
if old_param.cache_shape:
new_state.cache_shape = old_param.cache_shape
new_state.name = prefix + '.' + new_state.name
new.append(new_state)
return ParameterTuple(new)
[文档]@MindFormerRegister.register(MindFormerModuleType.OPTIMIZER)
class FP32StateAdamWeightDecay(nn.AdamWeightDecay):
r"""
This class is almost same with the mindspore's AdamWeightDecay implements, the
only difference is the optimizer's state will be always initialized with float32,
where the original AdamWeightDecay will initialize the optimizer's state with float16,
if the parameters are initialized with fp16.
This setting will avoid overflow in training big model using fp16.
Args:
params (Union[list[Parameter], list[dict]]): Must be list of `Parameter` or list of `dict`. When the
`params` is a list of `dict`, the string "params", "lr", "weight_decay", and "order_params"
are the keys can be parsed.
- params: Required. Parameters in current group. The value must be a list of `Parameter`.
- lr: Optional. If "lr" in the keys, the value of corresponding learning rate will be used.
If not, the `learning_rate` in optimizer will be used. Fixed and dynamic learning rate are supported.
- weight_decay: Optional. If "weight_decay" in the keys, the value of corresponding weight decay
will be used. If not, the `weight_decay` in the optimizer will be used. It should be noted that weight
decay can be a constant value or a Cell. It is a Cell only when dynamic weight decay is applied. Dynamic
weight decay is similar to dynamic learning rate, users need to customize a weight decay schedule only
with global step as input, and during training, the optimizer calls the instance of WeightDecaySchedule
to get the weight decay value of current step.
- order_params: Optional. When parameters is grouped, this usually is used to maintain the order of
parameters that appeared in the network to improve performance. The value should be parameters whose
order will be followed in optimizer.
If `order_params` in the keys, other keys will be ignored and the element of 'order_params' must be in
one group of `params`.
learning_rate (Union[float, int, Tensor, Iterable, LearningRateSchedule]): Default: 1e-3.
- float: The fixed learning rate value. Must be equal to or greater than 0.
- int: The fixed learning rate value. Must be equal to or greater than 0. It will be converted to float.
- Tensor: Its value should be a scalar or a 1-D vector. For scalar, fixed learning rate will be applied.
For vector, learning rate is dynamic, then the i-th step will take the i-th value as the learning rate.
- Iterable: Learning rate is dynamic. The i-th step will take the i-th value as the learning rate.
- LearningRateSchedule: Learning rate is dynamic. During training, the optimizer calls the instance of
LearningRateSchedule with step as the input to get the learning rate of current step.
beta1 (float): The exponential decay rate for the 1st moment estimations. Default: 0.9.
Should be in range (0.0, 1.0).
beta2 (float): The exponential decay rate for the 2nd moment estimations. Default: 0.999.
Should be in range (0.0, 1.0).
eps (float): Term added to the denominator to improve numerical stability. Default: 1e-6.
Should be greater than 0.
weight_decay (Union[float, int, Cell]): Weight decay (L2 penalty). Default: 0.0.
- float: The fixed weight decay value. Must be equal to or greater than 0.
- int: The fixed weight decay value. Must be equal to or greater than 0. It will be converted to float.
- Cell: Weight decay is dynamic. During training, the optimizer calls the instance of
the Cell with step as the input to get the weight decay value of current step.
Inputs:
- **gradients** (tuple[Tensor]) - The gradients of `params`, the shape is the same as `params`.
Outputs:
tuple[bool], all elements are True.
Raises:
TypeError: If `learning_rate` is not one of int, float, Tensor, Iterable, LearningRateSchedule.
TypeError: If element of `parameters` is neither Parameter nor dict.
TypeError: If `beta1`, `beta2` or `eps` is not a float.
TypeError: If `weight_decay` is neither float nor int.
ValueError: If `eps` is less than or equal to 0.
ValueError: If `beta1`, `beta2` is not in range (0.0, 1.0).
ValueError: If `weight_decay` is less than 0.
Supported Platforms:
``Ascend`` ``GPU`` ``CPU``
Examples:
>>> import mindspore as ms
>>> from mindformers.core.optim import FP32StateAdamWeightDecay
>>>
>>> net = Net()
>>> #1) All parameters use the same learning rate and weight decay
>>> optim = FP32StateAdamWeightDecay(params=net.trainable_params())
>>>
>>> #2) Use parameter groups and set different values
>>> conv_params = list(filter(lambda x: 'conv' in x.name, net.trainable_params()))
>>> no_conv_params = list(filter(lambda x: 'conv' not in x.name, net.trainable_params()))
>>> group_params = [{'params': conv_params, 'weight_decay': 0.01},
... {'params': no_conv_params, 'lr': 0.01},
... {'order_params': net.trainable_params()}]
>>> optim = FP32StateAdamWeightDecay(group_params, learning_rate=0.1, weight_decay=0.0)
>>> # The conv_params's parameters will use default learning rate of 0.1 and weight decay of 0.01.
>>> # The no_conv_params's parameters will use learning rate of 0.01 and default weight decay of 0.0.
>>> # The final parameters order in which the optimizer will be followed is the value of 'order_params'.
>>>
>>> loss = nn.SoftmaxCrossEntropyWithLogits()
>>> model = ms.Model(net, loss_fn=loss, optimizer=optim)
"""
def __init__(self, params, learning_rate=1e-3, beta1=0.9, beta2=0.999, eps=1e-6, weight_decay=0.0):
super(nn.AdamWeightDecay, self).__init__(learning_rate, params, weight_decay)
_check_param_value(beta1, beta2, eps, self.cls_name)
self.beta1 = Tensor(np.array([beta1]).astype(np.float32))
self.beta2 = Tensor(np.array([beta2]).astype(np.float32))
self.eps = Tensor(np.array([eps]).astype(np.float32))
self.moments1 = self.clone_state(prefix='adam_m', init='zeros')
self.moments2 = self.clone_state(prefix='adam_v', init='zeros')
self.fused_opt = P.AdamWeightDecay()
if context.get_context("device_target") == "CPU":
self.use_fused_opt = True
else:
self.use_fused_opt = False
[文档] def clone_state(self, prefix, init):
r"""
parameter_tuple: ParameterTuple. The parameters of the network
prefix: str. The prefix name of the parameters
init: str. The initialization method
"""
parameter_tuple = self.parameters
new = []
for old_param in parameter_tuple:
new_state = Parameter(initializer(init, shape=old_param.shape, dtype=mstype.float32))
new_state.param_info = old_param.param_info.clone()
if hasattr(old_param.param_info, "cloned_obj"):
old_param.param_info.cloned_obj.append(new_state)
else:
old_param.param_info.cloned_obj = [new_state]
new_state.is_init = False
new_state.set_data(initializer(init, shape=old_param.shape, dtype=mstype.float32))
new_state.name = prefix + '.' + new_state.name
new.append(new_state)
return ParameterTuple(new)