convmixer-1024-20.py

# Model settings
model = dict(
    type='ImageClassifier',
    backbone=dict(type='ConvMixer', arch='1024/20'),
    neck=dict(type='GlobalAveragePooling'),
    head=dict(
        type='LinearClsHead',
        num_classes=1000,
        in_channels=1024,
        loss=dict(type='CrossEntropyLoss', loss_weight=1.0),
    ))