/usr/local/lib64/python3.6/site-packages/torch/optim/__pycache__
Edit: /usr/local/lib64/python3.6/site-packages/torch/optim/__pycache__/adamw.cpython-36.pyc (6258B)
3
Eg @ s4 d dl Z ddlmZ ddlmZ G dd deZdS ) N ) _functional) Optimizerc s@ e Zd ZdZd fdd Z fd
d Zej dd
dZ Z S )AdamWa* Implements AdamW algorithm.
.. math::
\begin{aligned}
&\rule{110mm}{0.4pt} \\
&\textbf{input} : \gamma \text{(lr)}, \: \beta_1, \beta_2
\text{(betas)}, \: \theta_0 \text{(params)}, \: f(\theta) \text{(objective)},
\: \epsilon \text{ (epsilon)} \\
&\hspace{13mm} \lambda \text{(weight decay)}, \: amsgrad \\
&\textbf{initialize} : m_0 \leftarrow 0 \text{ (first moment)}, v_0 \leftarrow 0
\text{ ( second moment)}, \: \widehat{v_0}^{max}\leftarrow 0 \\[-1.ex]
&\rule{110mm}{0.4pt} \\
&\textbf{for} \: t=1 \: \textbf{to} \: \ldots \: \textbf{do} \\
&\hspace{5mm}g_t \leftarrow \nabla_{\theta} f_t (\theta_{t-1}) \\
&\hspace{5mm} \theta_t \leftarrow \theta_{t-1} - \gamma \lambda \theta_{t-1} \\
&\hspace{5mm}m_t \leftarrow \beta_1 m_{t-1} + (1 - \beta_1) g_t \\
&\hspace{5mm}v_t \leftarrow \beta_2 v_{t-1} + (1-\beta_2) g^2_t \\
&\hspace{5mm}\widehat{m_t} \leftarrow m_t/\big(1-\beta_1^t \big) \\
&\hspace{5mm}\widehat{v_t} \leftarrow v_t/\big(1-\beta_2^t \big) \\
&\hspace{5mm}\textbf{if} \: amsgrad \\
&\hspace{10mm}\widehat{v_t}^{max} \leftarrow \mathrm{max}(\widehat{v_t}^{max},
\widehat{v_t}) \\
&\hspace{10mm}\theta_t \leftarrow \theta_{t-1} - \gamma \widehat{m_t}/
\big(\sqrt{\widehat{v_t}^{max}} + \epsilon \big) \\
&\hspace{5mm}\textbf{else} \\
&\hspace{10mm}\theta_t \leftarrow \theta_{t-1} - \gamma \widehat{m_t}/
\big(\sqrt{\widehat{v_t}} + \epsilon \big) \\
&\rule{110mm}{0.4pt} \\[-1.ex]
&\bf{return} \: \theta_t \\[-1.ex]
&\rule{110mm}{0.4pt} \\[-1.ex]
\end{aligned}
For further details regarding the algorithm we refer to `Decoupled Weight Decay Regularization`_.
Args:
params (iterable): iterable of parameters to optimize or dicts defining
parameter groups
lr (float, optional): learning rate (default: 1e-3)
betas (Tuple[float, float], optional): coefficients used for computing
running averages of gradient and its square (default: (0.9, 0.999))
eps (float, optional): term added to the denominator to improve
numerical stability (default: 1e-8)
weight_decay (float, optional): weight decay coefficient (default: 1e-2)
amsgrad (boolean, optional): whether to use the AMSGrad variant of this
algorithm from the paper `On the Convergence of Adam and Beyond`_
(default: False)
.. _Decoupled Weight Decay Regularization:
https://arxiv.org/abs/1711.05101
.. _On the Convergence of Adam and Beyond:
https://openreview.net/forum?id=ryQu7f-RZ
MbP??+?:0yE>{Gz?Fc s d|kst dj|d|ks,t dj|d|d koBdk n sZt dj|d d|d kopdk n st dj|d d|kst d j|t|||||d
}tt| j|| d S ) Ng zInvalid learning rate: {}zInvalid epsilon value: {}r g ?z%Invalid beta parameter at index 0: {}r z%Invalid beta parameter at index 1: {}z Invalid weight_decay value: {})lrbetaseps weight_decayamsgrad)
ValueErrorformatdictsuperr __init__)selfparamsr r r
r r defaults) __class__ =/usr/local/lib64/python3.6/site-packages/torch/optim/adamw.pyr < s
zAdamW.__init__c s0 t t| j| x| jD ]}|jdd qW d S )Nr F)r r __setstate__ param_groups
setdefault)r stategroup)r r r r L s zAdamW.__setstate__Nc C s d}|dk r&t j | }W dQ R X xb| jD ]V}g }g }g }g }g }g } g }
|d } |d \} }
x|d D ]}|jdkrqp|j| |jjrtd|j|j | j| }t|dkrd|d< t j |t j
d|d < t j |t j
d|d
< | rt j |t j
d|d < |j|d |j|d
| r8| j|d |d d 7 < |
j|d qpW t j ||||| |
| | |
|d
|d |d d q0W |S )zPerforms a single optimization step.
Args:
closure (callable, optional): A closure that reevaluates the model
and returns the loss.
Nr r r z'AdamW does not support sparse gradientsr step)Z
memory_formatZexp_avgZ
exp_avg_sqZmax_exp_avg_sqr r r r
)r beta1beta2r r r
)
torchZ enable_gradr ZgradappendZ is_sparse RuntimeErrorr lenZ
zeros_likeZpreserve_formatFZadamw)r ZclosureZlossr Zparams_with_gradZgradsZexp_avgsZ exp_avg_sqsZ
state_sumsZmax_exp_avg_sqsZ state_stepsr r! r" pr r r r r Q s^
z
AdamW.stepr r )r r) r r
F)N)
__name__
__module__ __qualname____doc__r r r# Zno_gradr
__classcell__r r )r r r s 4 r )r# r r' Z optimizerr r r r r r
s