Missing key(s) in state_dict

最新推荐文章于 2024-04-16 00:15:47 发布

AbyssssssssssS

最新推荐文章于 2024-04-16 00:15:47 发布

阅读量1k

点赞数

分类专栏：计算机散养文章标签： python

本文链接：https://blog.csdn.net/abysssssssssss/article/details/128604662

版权

计算机散养专栏收录该内容

17 篇文章 0 订阅

订阅专栏

RuntimeError: Error(s) in loading state_dict for DataParallel:
Missing key(s) in state_dict: "module.model.Frequency_Edge_Module1.UAM.bn.weight", "module.model.Frequency_Edge_Module1.UAM.bn.bias", "module.model.Frequency_Edge_Module1.UAM.bn.running_mean", "module.model.Frequency_Edge_Module1.UAM.bn.running_var",

使用多gpu训练，保存的模型key前面会加module.

这样去掉

# 多gpu前面会加module
new_state = {}
state_dict = torch.load(model_path, map_location=torch.device('cpu'))
for key, value in state_dict.items():
    new_state[key.replace('module.', '')] = value
self.model.load_state_dict(new_state)