KeyError: ‘layer1.1.bn1.bias‘

在跑AICITY2020_DMT_HST的代码时遇到的问题，记录一下

参考文章

问题描述

加载resnet101模型的时候，参数多了

解决方法

将不用的参数过滤掉

#原代码
# def load_param(self, model_path):
    #     param_dict = torch.load(model_path)
    #     if 'state_dict' in param_dict:
    #         param_dict = param_dict['state_dict']
    #     for i in param_dict:
    #         if 'fc' in i:
    #             continue
    #         embed()
    #         self.state_dict()[i.replace('module.','')].copy_(param_dict[i])
 
#改的代码   
   def load_param(self, model_path):
        param_dict = torch.load(model_path)
        param_dict = {k: v for k, v in param_dict.items() if 'bn1.running_mean' not in k}
        if 'state_dict' in param_dict:
            param_dict = param_dict['state_dict']
        for i in param_dict:
            # embed()
            if 'layer1.0.bn1.running_mean' in i:
                continue
            if 'fc' in i:
                continue
            if 'bn1.running_var' in i:
                continue
            if 'bn1.weight' in i:
                continue
            if 'bn1.bias' in i:
                continue
            if 'layer1.1.bn1.running_var' in i:
                continue
            if 'layer1.1.bn1.weight'  in i:
                continue
            self.state_dict()[i.replace('module.', '')].copy_(param_dict[i])