-
Notifications
You must be signed in to change notification settings - Fork 44
/
knet_s3_upernet_swin-t_80k_adamw_ade20k.py
45 lines (42 loc) · 1.23 KB
/
knet_s3_upernet_swin-t_80k_adamw_ade20k.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
_base_ = 'knet_s3_upernet_r50-d8_80k_adamw_ade20k.py'
# model settings
norm_cfg = dict(type='SyncBN', requires_grad=True)
num_stages = 3
conv_kernel_size = 1
model = dict(
type='EncoderDecoder',
pretrained='./pretrain/swin/swin_tiny_patch4_window7_224.pth',
backbone=dict(
_delete_=True,
type='SwinTransformer',
embed_dim=96,
depths=[2, 2, 6, 2],
num_heads=[3, 6, 12, 24],
window_size=7,
mlp_ratio=4.,
qkv_bias=True,
qk_scale=None,
drop_rate=0.,
attn_drop_rate=0.,
drop_path_rate=0.3,
ape=False,
patch_norm=True,
out_indices=(0, 1, 2, 3),
use_checkpoint=False),
decode_head=dict(
kernel_generate_head=dict(
in_channels=[96, 192, 384, 768])),
auxiliary_head=dict(
in_channels=384))
# modify learning rate following the official implementation of Swin Transformer
optimizer = dict(
type='AdamW',
lr=0.00006,
betas=(0.9, 0.999),
weight_decay=0.0005,
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.)
}))