mmsegmentation/configs/twins/twins_svt-s_uperhead_8xb2-160k_ade20k-512x512.py

_base_ = [
    '../_base_/models/twins_pcpvt-s_upernet.py',
    '../_base_/datasets/ade20k.py', '../_base_/default_runtime.py',
    '../_base_/schedules/schedule_160k.py'
]
crop_size = (512, 512)
data_preprocessor = dict(size=crop_size)
checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/twins/alt_gvt_small_20220308-7e1c3695.pth'  # noqa

model = dict(
    data_preprocessor=data_preprocessor,
    backbone=dict(
        type='SVT',
        init_cfg=dict(type='Pretrained', checkpoint=checkpoint),
        embed_dims=[64, 128, 256, 512],
        num_heads=[2, 4, 8, 16],
        mlp_ratios=[4, 4, 4, 4],
        depths=[2, 2, 10, 4],
        windiow_sizes=[7, 7, 7, 7],
        norm_after_stage=True),
    decode_head=dict(in_channels=[64, 128, 256, 512]),
    auxiliary_head=dict(in_channels=256))

optim_wrapper = dict(
    _delete_=True,
    type='OptimWrapper',
    optimizer=dict(
        type='AdamW', lr=0.00006, betas=(0.9, 0.999), weight_decay=0.01),
    paramwise_cfg=dict(custom_keys={
        'pos_block': dict(decay_mult=0.),
        'norm': dict(decay_mult=0.)
    }))

param_scheduler = [
    dict(
        type='LinearLR', start_factor=1e-6, by_epoch=False, begin=0, end=1500),
    dict(
        type='PolyLR',
        eta_min=0.0,
        power=1.0,
        begin=1500,
        end=160000,
        by_epoch=False,
    )
]

train_dataloader = dict(batch_size=2, num_workers=2)
val_dataloader = dict(batch_size=1, num_workers=4)
test_dataloader = val_dataloader