forked from mrq/DL-Art-School
73 lines
2.7 KiB
Python
73 lines
2.7 KiB
Python
|
import torch
|
||
|
|
from torch.optim import Optimizer
|
||
|
|
|
||
|
|
|
||
|
|
class SGDNoBiasMomentum(Optimizer):
|
||
|
|
r"""
|
||
|
|
Copy of pytorch implementation of SGD with a modification which turns off momentum for params marked
|
||
|
|
with `is_bn` or `is_bias`.
|
||
|
|
"""
|
||
|
|
|
||
|
|
def __init__(self, params, lr, momentum=0, dampening=0,
|
||
|
|
weight_decay=0, nesterov=False):
|
||
|
|
if lr < 0.0:
|
||
|
|
raise ValueError("Invalid learning rate: {}".format(lr))
|
||
|
|
if momentum < 0.0:
|
||
|
|
raise ValueError("Invalid momentum value: {}".format(momentum))
|
||
|
|
if weight_decay < 0.0:
|
||
|
|
raise ValueError("Invalid weight_decay value: {}".format(weight_decay))
|
||
|
|
|
||
|
|
defaults = dict(lr=lr, momentum=momentum, dampening=dampening,
|
||
|
|
weight_decay=weight_decay, nesterov=nesterov)
|
||
|
|
if nesterov and (momentum <= 0 or dampening != 0):
|
||
|
|
raise ValueError("Nesterov momentum requires a momentum and zero dampening")
|
||
|
|
super().__init__(params, defaults)
|
||
|
|
|
||
|
|
def __setstate__(self, state):
|
||
|
|
super().__setstate__(state)
|
||
|
|
for group in self.param_groups:
|
||
|
|
group.setdefault('nesterov', False)
|
||
|
|
|
||
|
|
@torch.no_grad()
|
||
|
|
def step(self, closure=None):
|
||
|
|
"""Performs a single optimization step.
|
||
|
|
|
||
|
|
Arguments:
|
||
|
|
closure (callable, optional): A closure that reevaluates the model
|
||
|
|
and returns the loss.
|
||
|
|
"""
|
||
|
|
loss = None
|
||
|
|
if closure is not None:
|
||
|
|
with torch.enable_grad():
|
||
|
|
loss = closure()
|
||
|
|
|
||
|
|
for group in self.param_groups:
|
||
|
|
weight_decay = group['weight_decay']
|
||
|
|
momentum = group['momentum']
|
||
|
|
dampening = group['dampening']
|
||
|
|
nesterov = group['nesterov']
|
||
|
|
|
||
|
|
for p in group['params']:
|
||
|
|
if p.grad is None:
|
||
|
|
continue
|
||
|
|
d_p = p.grad
|
||
|
|
if weight_decay != 0:
|
||
|
|
d_p = d_p.add(p, alpha=weight_decay)
|
||
|
|
# **this is the only modification over standard torch.optim.SGD:
|
||
|
|
is_bn_or_bias = (hasattr(p, 'is_bn') and p.is_bn) or (hasattr(p, 'is_bias') and p.is_bias)
|
||
|
|
if not is_bn_or_bias and momentum != 0:
|
||
|
|
param_state = self.state[p]
|
||
|
|
if 'momentum_buffer' not in param_state:
|
||
|
|
buf = param_state['momentum_buffer'] = torch.clone(d_p).detach()
|
||
|
|
else:
|
||
|
|
buf = param_state['momentum_buffer']
|
||
|
|
buf.mul_(momentum).add_(d_p, alpha=1 - dampening)
|
||
|
|
if nesterov:
|
||
|
|
d_p = d_p.add(buf, alpha=momentum)
|
||
|
|
else:
|
||
|
|
d_p = buf
|
||
|
|
|
||
|
|
p.add_(d_p, alpha=-group['lr'])
|
||
|
|
|
||
|
|
return loss
|