vit-t_struct-token-sse_ade20k_512x512.py

_base_ = [
    '../_base_/struct_token.py',
    '../_base_/datasets/ade20k.py', '../_base_/default_runtime.py',
    '../_base_/schedules/schedule_160k.py'
]

model = dict(pretrained='pretrain/vit_tiny_p16_384.pth',
             backbone=dict(embed_dims=192, num_heads=3),
             decode_head=dict(type='SelfSliceExtracStructTokenHead',
                              in_channels=192,
                              channels=192))
# AdamW optimizer, no weight decay for position embedding & layer norm
# in backbone
optimizer = dict(
    _delete_=True,
    type='AdamW',
    lr=0.00002,
    betas=(0.9, 0.999),
    weight_decay=0.01,
    paramwise_cfg=dict(
        custom_keys={
            'pos_embed': dict(decay_mult=0.),
            'cls_token': dict(decay_mult=0.),
            'norm': dict(decay_mult=0.),
            'decode_head': dict(lr_mult=9.),
        }))

lr_config = dict(
    _delete_=True,
    policy='poly',
    warmup='linear',
    warmup_iters=1500,
    warmup_ratio=1e-6,
    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=2)
checkpoint_config = dict(by_epoch=False, interval=80000)
evaluation = dict(interval=8000, metric='mIoU', pre_eval=True)