saymrwulf/pytorch
1
0
Fork 0
mirror of https://github.com/saymrwulf/pytorch.git synced 2026-05-14 20:57:59 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

[optim] take kw-only argument for functional optim APIs (#56185)

Browse source 
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/56185

ghstack-source-id: 126670123

Reviewed By: albanD

Differential Revision: D27802169

fbshipit-source-id: f5e1cb2046dcdeecf5f6b0f70892828bf0adb22f
This commit is contained in:
Wanchao Liang 2021-04-15 20:05:14 -07:00 committed by Facebook GitHub Bot
parent bd3c63aeeb
commit 4611387608
17 changed files with 88 additions and 80 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

8
torch/distributed/optim/functional_adadelta.py
View file

@ -76,7 +76,7 @@ class _FunctionalAdadelta(object):
grads,
square_avgs,
acc_deltas,
lr,
rho,
eps,
weight_decay)
lr=lr,
rho=rho,
eps=eps,
weight_decay=weight_decay)

8
torch/distributed/optim/functional_adagrad.py
View file

@ -84,7 +84,7 @@ class _FunctionalAdagrad(object):
grads,
state_sums,
state_steps,
self.defaults['lr'],
self.defaults['weight_decay'],
self.defaults['lr_decay'],
self.defaults['eps'])
lr=self.defaults['lr'],
weight_decay=self.defaults['weight_decay'],
lr_decay=self.defaults['lr_decay'],
eps=self.defaults['eps'])

12
torch/distributed/optim/functional_adam.py
View file

@ -105,9 +105,9 @@ class _FunctionalAdam(object):
exp_avg_sqs,
max_exp_avg_sqs,
state_steps,
self.amsgrad,
self.defaults['beta1'],
self.defaults['beta2'],
self.defaults['lr'],
self.defaults['weight_decay'],
self.defaults['eps'])
amsgrad=self.amsgrad,
beta1=self.defaults['beta1'],
beta2=self.defaults['beta2'],
lr=self.defaults['lr'],
weight_decay=self.defaults['weight_decay'],
eps=self.defaults['eps'])

10
torch/distributed/optim/functional_adamax.py
View file

@ -95,8 +95,8 @@ class _FunctionalAdamax(object):
exp_avgs,
exp_infs,
state_steps,
self.defaults['eps'],
self.defaults['beta1'],
self.defaults['beta2'],
self.defaults['lr'],
self.defaults['weight_decay'])
eps=self.defaults['eps'],
beta1=self.defaults['beta1'],
beta2=self.defaults['beta2'],
lr=self.defaults['lr'],
weight_decay=self.defaults['weight_decay'])

12
torch/distributed/optim/functional_adamw.py
View file

@ -105,9 +105,9 @@ class _FunctionalAdamW(object):
exp_avg_sqs,
max_exp_avg_sqs,
state_steps,
self.amsgrad,
self.defaults['beta1'],
self.defaults['beta2'],
self.defaults['lr'],
self.defaults['weight_decay'],
self.defaults['eps'])
amsgrad=self.amsgrad,
beta1=self.defaults['beta1'],
beta2=self.defaults['beta2'],
lr=self.defaults['lr'],
weight_decay=self.defaults['weight_decay'],
eps=self.defaults['eps'])

12
torch/distributed/optim/functional_rmsprop.py
View file

@ -91,9 +91,9 @@ class _FunctionalRMSprop(object):
square_avgs,
grad_avgs,
momentum_buffer_list,
lr,
alpha,
eps,
weight_decay,
momentum,
self.centered)
lr=lr,
alpha=alpha,
eps=eps,
weight_decay=weight_decay,
momentum=momentum,
centered=self.centered)

8
torch/distributed/optim/functional_rprop.py
View file

@ -75,7 +75,7 @@ class _FunctionalRprop(object):
grads,
prevs,
step_sizes,
step_size_min,
step_size_max,
etaminus,
etaplus)
step_size_min=step_size_min,
step_size_max=step_size_max,
etaminus=etaminus,
etaplus=etaplus)

10
torch/distributed/optim/functional_sgd.py
View file

@ -73,11 +73,11 @@ class _FunctionalSGD(object):
F.sgd(params,
grads,
momentum_buffer_list,
weight_decay,
momentum,
lr,
dampening,
self.nesterov)
weight_decay=weight_decay,
momentum=momentum,
lr=lr,
dampening=dampening,
nesterov=self.nesterov)
# update momentum_buffers in state
for i, p in enumerate(params):

8
torch/optim/_functional.py
View file

@ -17,6 +17,7 @@ def adagrad(params: List[Tensor],
grads: List[Tensor],
state_sums: List[Tensor],
state_steps: List[int],
*,
lr: float,
weight_decay: float,
lr_decay: float,
@ -56,6 +57,7 @@ def adam(params: List[Tensor],
exp_avg_sqs: List[Tensor],
max_exp_avg_sqs: List[Tensor],
state_steps: List[int],
*,
amsgrad: bool,
beta1: float,
beta2: float,
@ -102,6 +104,7 @@ def adamw(params: List[Tensor],
exp_avg_sqs: List[Tensor],
max_exp_avg_sqs: List[Tensor],
state_steps: List[int],
*,
amsgrad: bool,
beta1: float,
beta2: float,
@ -143,6 +146,7 @@ def adamw(params: List[Tensor],
def sgd(params: List[Tensor],
d_p_list: List[Tensor],
momentum_buffer_list: List[Optional[Tensor]],
*,
weight_decay: float,
momentum: float,
lr: float,
@ -180,6 +184,7 @@ def adadelta(params: List[Tensor],
grads: List[Tensor],
square_avgs: List[Tensor],
acc_deltas: List[Tensor],
*,
lr: float,
rho: float,
eps: float,
@ -205,6 +210,7 @@ def rmsprop(params: List[Tensor],
square_avgs: List[Tensor],
grad_avgs: List[Tensor],
momentum_buffer_list: List[Tensor],
*,
lr: float,
alpha: float,
eps: float,
@ -244,6 +250,7 @@ def rprop(params: List[Tensor],
grads: List[Tensor],
prevs: List[Tensor],
step_sizes: List[Tensor],
*,
step_size_min: float,
step_size_max: float,
etaminus: float,
@ -282,6 +289,7 @@ def adamax(params: List[Tensor],
exp_avgs: List[Tensor],
exp_infs: List[Tensor],
state_steps: List[int],
*,
eps: float,
beta1: float,
beta2: float,

8
torch/optim/adadelta.py
View file

@ -81,9 +81,9 @@ class Adadelta(Optimizer):
grads,
square_avgs,
acc_deltas,
lr,
rho,
eps,
weight_decay)
lr=lr,
rho=rho,
eps=eps,
weight_decay=weight_decay)
return loss

8
torch/optim/adagrad.py
View file

@ -84,9 +84,9 @@ class Adagrad(Optimizer):
grads,
state_sums,
state_steps,
group['lr'],
group['weight_decay'],
group['lr_decay'],
group['eps'])
lr=group['lr'],
weight_decay=group['weight_decay'],
lr_decay=group['lr_decay'],
eps=group['eps'])
return loss

12
torch/optim/adam.py
View file

@ -110,10 +110,10 @@ class Adam(Optimizer):
exp_avg_sqs,
max_exp_avg_sqs,
state_steps,
group['amsgrad'],
beta1,
beta2,
group['lr'],
group['weight_decay'],
group['eps'])
amsgrad=group['amsgrad'],
beta1=beta1,
beta2=beta2,
lr=group['lr'],
weight_decay=group['weight_decay'],
eps=group['eps'])
return loss

10
torch/optim/adamax.py
View file

@ -89,10 +89,10 @@ class Adamax(Optimizer):
exp_avgs,
exp_infs,
state_steps,
eps,
beta1,
beta2,
lr,
weight_decay)
eps=eps,
beta1=beta1,
beta2=beta2,
lr=lr,
weight_decay=weight_decay)
return loss

12
torch/optim/adamw.py
View file

@ -113,11 +113,11 @@ class AdamW(Optimizer):
exp_avg_sqs,
max_exp_avg_sqs,
state_steps,
amsgrad,
beta1,
beta2,
group['lr'],
group['weight_decay'],
group['eps'])
amsgrad=amsgrad,
beta1=beta1,
beta2=beta2,
lr=group['lr'],
weight_decay=group['weight_decay'],
eps=group['eps'])
return loss

12
torch/optim/rmsprop.py
View file

@ -108,11 +108,11 @@ class RMSprop(Optimizer):
square_avgs,
grad_avgs,
momentum_buffer_list,
group['lr'],
group['alpha'],
group['eps'],
group['weight_decay'],
group['momentum'],
group['centered'])
lr=group['lr'],
alpha=group['alpha'],
eps=group['eps'],
weight_decay=group['weight_decay'],
momentum=group['momentum'],
centered=group['centered'])
return loss

8
torch/optim/rprop.py
View file

@ -74,9 +74,9 @@ class Rprop(Optimizer):
grads,
prevs,
step_sizes,
step_size_min,
step_size_max,
etaminus,
etaplus)
step_size_min=step_size_min,
step_size_max=step_size_max,
etaminus=etaminus,
etaplus=etaplus)
return loss

10
torch/optim/sgd.py
View file

@ -110,11 +110,11 @@ class SGD(Optimizer):
F.sgd(params_with_grad,
d_p_list,
momentum_buffer_list,
weight_decay,
momentum,
lr,
dampening,
nesterov)
weight_decay=weight_decay,
momentum=momentum,
lr=lr,
dampening=dampening,
nesterov=nesterov)
# update momentum_buffers in state
for p, momentum_buffer in zip(params_with_grad, momentum_buffer_list):