File size: 986 Bytes
8c6b5ee |
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 |
import torch
from dassl.utils import check_isfile
from dassl.engine import TRAINER_REGISTRY, TrainerXU
@TRAINER_REGISTRY.register()
class AdaBN(TrainerXU):
"""Adaptive Batch Normalization.
https://arxiv.org/abs/1603.04779.
"""
def __init__(self, cfg):
super().__init__(cfg)
self.done_reset_bn_stats = False
def check_cfg(self, cfg):
assert check_isfile(
cfg.MODEL.INIT_WEIGHTS
), "The weights of source model must be provided"
def before_epoch(self):
if not self.done_reset_bn_stats:
for m in self.model.modules():
classname = m.__class__.__name__
if classname.find("BatchNorm") != -1:
m.reset_running_stats()
self.done_reset_bn_stats = True
def forward_backward(self, batch_x, batch_u):
input_u = batch_u["img"].to(self.device)
with torch.no_grad():
self.model(input_u)
return None
|