repmlp-base_224.py

# model settings
model = dict(
    type='ImageClassifier',
    backbone=dict(
        type='RepMLPNet',
        arch='B',
        img_size=224,
        out_indices=(3, ),
        reparam_conv_kernels=(1, 3),
        deploy=False),
    neck=dict(type='GlobalAveragePooling'),
    head=dict(
        type='LinearClsHead',
        num_classes=1000,
        in_channels=768,
        loss=dict(type='CrossEntropyLoss', loss_weight=1.0),
        topk=(1, 5),
    ))