训练时报错“Missing key(s) in state_dict”

StillYoung1214 commented 3 years ago

Bubble你好，我目前的操作是在darknet.py中新添加了几个类，然后直接运行了predict.py，然后就报错显示“Missing key(s) in state_dict: "backbone.cbam.channel_attention.shared_MLP.0.weight", "backbone.cbam.channel_attention.shared_MLP.2.weight", "backbone.cbam.spatial_attention.conv2d.weight", "backbone.cbam.spatial_attention.conv2d.bias".”，如下图：

请问这个问题是需要我在darknet.py修改之后重新运行train.py才能解决吗？如果有其他懂的同学能看到这条issue也烦请回答一下，谢谢。

bubbliiiing commented 3 years ago

回复了B站

StillYoung1214 commented 3 years ago

好的非常感谢 ------------------ 原始邮件 ------------------ 发件人: "bubbliiiing/yolo3-pytorch" @.>; 发送时间: 2021年9月7日(星期二) 晚上10:52 @.>; 抄送: "La @.**@.>; 主题: Re: [bubbliiiing/yolo3-pytorch] 训练时报错“Missing key(s) in state_dict” (#79)

回复了B站

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe. Triage notifications on the go with GitHub Mobile for iOS or Android.

bubbliiiing / yolo3-pytorch

训练时报错“Missing key(s) in state_dict” #79