_base_ = ['twins-svt-base_8xb128_in1k.py'] # model settings model = dict(backbone=dict(arch='large'), head=dict(in_channels=1024)) # dataset settings train_dataloader = dict(batch_size=64)
