Pytorch入门

PoolNet2019

1.预训练模型修改为VGG报错：RuntimeError: Error(s) in loading state_dict for PoolNet:
        Missing key(s) in state_dict: "base.vgg16.base.0.weight", "base.vgg16.base.0.bias", "base.vgg16.base.2.weight", "base.vgg16.base.2.bias", "base.vgg16.base.5.weight"....

查找到网页https://discuss.pytorch.org/t/solved-keyerror-unexpected-key-module-encoder-embedding-weight-in-state-dict/1686/15

然后将

self.net.load_state_dict(torch.load(self.config.model, map_location='cpu')）

添加字段，改成

self.net.load_state_dict(torch.load(self.config.model, map_location='cpu'),strict=False)

参数不匹配倒是解决了，但出现了新问题

RuntimeError: Error(s) in loading state_dict for PoolNet:
        size mismatch for base.infos.2.0.weight: copying a param with shape torch.Size([256, 512, 3, 3]) from checkpoint, the shape in current model is torch.Size([128, 512, 3, 3]).

emmm....是预训练模型的锅吧