# Copyright (c) 2017-2019 Uber Technologies, Inc.
# SPDX-License-Identifier: Apache-2.0
import math
import torch
from torch.optim.optimizer import Optimizer
[docs]class ClippedAdam(Optimizer):
"""
:param params: iterable of parameters to optimize or dicts defining parameter groups
:param lr: learning rate (default: 1e-3)
:param Tuple betas: coefficients used for computing
running averages of gradient and its square (default: (0.9, 0.999))
:param eps: term added to the denominator to improve
numerical stability (default: 1e-8)
:param weight_decay: weight decay (L2 penalty) (default: 0)
:param clip_norm: magnitude of norm to which gradients are clipped (default: 10.0)
:param lrd: rate at which learning rate decays (default: 1.0)
Small modification to the Adam algorithm implemented in torch.optim.Adam
to include gradient clipping and learning rate decay.
Reference
`A Method for Stochastic Optimization`, Diederik P. Kingma, Jimmy Ba
https://arxiv.org/abs/1412.6980
"""
def __init__(self, params, lr=1e-3, betas=(0.9, 0.999), eps=1e-8,
weight_decay=0, clip_norm=10.0, lrd=1.0):
defaults = dict(lr=lr, betas=betas, eps=eps,
weight_decay=weight_decay,
clip_norm=clip_norm, lrd=lrd)
super(ClippedAdam, self).__init__(params, defaults)
[docs] def step(self, closure=None):
"""
:param closure: An optional closure that reevaluates the model and returns the loss.
Performs a single optimization step.
"""
loss = None
if closure is not None:
loss = closure()
for group in self.param_groups:
group['lr'] *= group['lrd']
for p in group['params']:
if p.grad is None:
continue
grad = p.grad.data
grad.clamp_(-group['clip_norm'], group['clip_norm'])
state = self.state[p]
# State initialization
if len(state) == 0:
state['step'] = 0
# Exponential moving average of gradient values
state['exp_avg'] = torch.zeros_like(grad)
# Exponential moving average of squared gradient values
state['exp_avg_sq'] = torch.zeros_like(grad)
exp_avg, exp_avg_sq = state['exp_avg'], state['exp_avg_sq']
beta1, beta2 = group['betas']
state['step'] += 1
if group['weight_decay'] != 0:
grad = grad.add(group['weight_decay'], p.data)
# Decay the first and second moment running average coefficient
exp_avg.mul_(beta1).add_(1 - beta1, grad)
exp_avg_sq.mul_(beta2).addcmul_(1 - beta2, grad, grad)
denom = exp_avg_sq.sqrt().add_(group['eps'])
bias_correction1 = 1 - beta1 ** state['step']
bias_correction2 = 1 - beta2 ** state['step']
step_size = group['lr'] * math.sqrt(bias_correction2) / bias_correction1
p.data.addcdiv_(-step_size, exp_avg, denom)
return loss