File size: 2,239 Bytes
9e14838 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 | # -*- coding:utf-8 -*-
import torch
from torch._six import inf
class NativeScalerWithGradNormCount:
state_dict_key = "amp_scaler"
def __init__(self):
self._scaler = torch.cuda.amp.GradScaler()
def __call__(
self,
cfg,
loss,
optimizer,
clip_grad=None,
parameters=None,
create_graph=False,
update_grad=True,
step=0,
):
self._scaler.scale(loss).backward(create_graph=create_graph)
if update_grad:
if clip_grad is not None:
assert parameters is not None
self._scaler.unscale_(
optimizer
) # unscale the gradients of optimizer's assigned params in-place
norm = torch.nn.utils.clip_grad_norm_(parameters, clip_grad)
else:
self._scaler.unscale_(optimizer)
norm = get_grad_norm_(parameters)
if cfg.TRAIN.optimizer != "SAM":
self._scaler.step(optimizer)
else:
if step == 0:
optimizer.first_step(zero_grad=True)
else:
self._scaler = optimizer.second_step(
zero_grad=False, scaler=self._scaler
)
self._scaler.update()
else:
norm = None
return norm
def state_dict(self):
return self._scaler.state_dict()
def load_state_dict(self, state_dict):
self._scaler.load_state_dict(state_dict)
def get_grad_norm_(parameters, norm_type: float = 2.0) -> torch.Tensor:
if isinstance(parameters, torch.Tensor):
parameters = [parameters]
parameters = [p for p in parameters if p.grad is not None]
norm_type = float(norm_type)
if len(parameters) == 0:
return torch.tensor(0.0)
device = parameters[0].grad.device
if norm_type == inf:
total_norm = max(p.grad.detach().abs().max().to(device) for p in parameters)
else:
total_norm = torch.norm(
torch.stack(
[torch.norm(p.grad.detach(), norm_type).to(device) for p in parameters]
),
norm_type,
)
return total_norm
|