-
Notifications
You must be signed in to change notification settings - Fork 11
/
swin_l_upper_w_jitter_uvo_finetune_training.py
60 lines (56 loc) · 1.77 KB
/
swin_l_upper_w_jitter_uvo_finetune_training.py
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
_base_ = [
'../_base_/models/upernet_swin.py', '../_base_/datasets/uvo_finetune.py',
'../_base_/default_runtime.py', '../_base_/schedules/schedule_160k.py'
]
model = dict(
pretrained='PATH/TO/YOUR/swin_large_patch4_window12_384_22k.pth',
backbone=dict(
pretrain_img_size=384,
embed_dims=192,
depths=[2, 2, 18, 2],
num_heads=[6, 12, 24, 48],
drop_path_rate=0.2,
window_size=12),
decode_head=dict(
in_channels=[192, 384, 768, 1536],
num_classes=2,
loss_decode=dict(
type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0)),
auxiliary_head=dict(
in_channels=768,
num_classes=2,
loss_decode=dict(
type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0))
)
# AdamW optimizer, no weight decay for position embedding & layer norm
# in backbone
optimizer = dict(
_delete_=True,
type='AdamW',
lr=0.00006 / 10,
betas=(0.9, 0.999),
weight_decay=0.01,
paramwise_cfg=dict(
custom_keys={
'absolute_pos_embed': dict(decay_mult=0.),
'relative_position_bias_table': dict(decay_mult=0.),
'norm': dict(decay_mult=0.)
}))
lr_config = dict(
_delete_=True,
policy='poly',
warmup='linear',
warmup_iters=1500,
warmup_ratio=5e-7,
power=1.0,
min_lr=0.0,
by_epoch=False)
# By default, models are trained on 8 GPUs with 2 images per GPU
data = dict(
samples_per_gpu=4,
workers_per_gpu=4,
)
load_from = '/tmp-network/user/ydu/mmsegmentation/work_dirs/biggest_model_clean_w_jitter/iter_300000.pth'
runner = dict(type='IterBasedRunner', max_iters=100000)
checkpoint_config = dict(by_epoch=False, interval=5000)
evaluation = dict(interval=5000, metric='mIoU', pre_eval=True)