weixin_43467572的博客

前言记录训练ViT遇到的问题。可学习参数from torchsummary import summaryvit_model = ViT( image_size = 256, patch_size = 32, num_classes = 1000, dim = 1024, depth = 6, heads = 16, mlp_dim = 2048, dropout = 0.1, emb_dropout = 0.1)devi