Is a pretrained model weights required for synapse dataset training?

[SilenceMonk]

nnFormer/nnformer/training/network_training/nnFormerTrainerV2_nnformer_synapse.py

Line 54 in d48aaeb

self.load_pretrain_weight=True

nnFormer/nnformer/training/network_training/nnFormerTrainerV2_nnformer_synapse.py

Line 177 in d48aaeb

checkpoint = torch.load('/home/xychen/jsguo/weight/gelunorm_former_skip_global_shift.model', map_location='cpu')

Since the defaut is True, I'm assuming it is necessary to load (ImageNet?) pretrained weights. However, I can't find the pretrained weights here, or do you mean this link?