怎么在Pytorch中只导入部分模型参数？针对这个问题，这篇文章详细介绍了相对应的分析和解答，希望可以帮助更多想解决这个问题的小伙伴找到更简单易行的方法。
1.PyTorch是相当简洁且高效快速的框架；2.设计追求最少的封装；3.设计符合人类思维，它让用户尽可能地专注于实现自己的想法；4.与google的Tensorflow类似，FAIR的支持足以确保PyTorch获得持续的开发更新；5.PyTorch作者亲自维护的论坛 供用户交流和求教问题6.入门简单
import torch as t from torch.nn import Module from torch import nn from torch.nn import functional as F class Net(Module): def __init__(self): super(Net,self).__init__() self.conv1 = nn.Conv2d(3,32,3,1) self.conv2 = nn.Conv2d(32,3,3,1) self.w = nn.Parameter(t.randn(3,10)) for p in self.children(): nn.init.xavier_normal_(p.weight.data) nn.init.constant_(p.bias.data, 0) def forward(self, x): out = self.conv1(x) out = self.conv2(x) out = F.avg_pool2d(out,(out.shape[2],out.shape[3])) out = F.linear(out,weight=self.w) return out
然后我们保存这个网络的初始值。
model = Net() t.save(model.state_dict(),'xxx.pth')
现在我们将Net修改一下，多加几个卷积层，但并不加入到forward中，仅仅出于少些几行的目的。
import torch as t from torch.nn import Module from torch import nn from torch.nn import functional as F class Net(Module): def __init__(self): super(Net, self).__init__() self.conv1 = nn.Conv2d(3, 32, 3, 1) self.conv2 = nn.Conv2d(32, 3, 3, 1) self.conv3 = nn.Conv2d(3,64,3,1) self.conv4 = nn.Conv2d(64,32,3,1) for p in self.children(): nn.init.xavier_normal_(p.weight.data) nn.init.constant_(p.bias.data, 0) self.w = nn.Parameter(t.randn(3, 10)) def forward(self, x): out = self.conv1(x) out = self.conv2(x) out = F.avg_pool2d(out, (out.shape[2], out.shape[3])) out = F.linear(out, weight=self.w) return out
我们现在试着导入之前保存的模型参数。
path = 'xxx.pth' model = Net() model.load_state_dict(t.load(path)) ''' RuntimeError: Error(s) in loading state_dict for Net: Missing key(s) in state_dict: "conv3.weight", "conv3.bias", "conv4.weight", "conv4.bias". '''
出现了没有在模型文件中找到error中的关键字的错误。
现在我们这样导入模型
path = 'xxx.pth'
model = Net()
save_model = t.load(path)
model_dict = model.state_dict()
state_dict = {k:v for k,v in save_model.items() if k in model_dict.keys()}
print(state_dict.keys()) # dict_keys(['w', 'conv1.weight', 'conv1.bias', 'conv2.weight', 'conv2.bias'])
model_dict.update(state_dict)
model.load_state_dict(model_dict)看看上面的代码，很容易弄明白。其中model_dict.update的作用是更新代码中搭建的模型参数字典。为啥更新我其实并不清楚，但这一步骤是必须的，否则还会报错。
为了弄清楚为什么要更新model_dict，我们不妨分别输出state_dict和model_dict的关键值看一看。
for k in state_dict.keys(): print(k) ''' w conv1.weight conv1.bias conv2.weight conv2.bias ''' for k in model_dict.keys(): print(k) ''' w conv1.weight conv1.bias conv2.weight conv2.bias conv3.weight conv3.bias conv4.weight conv4.bias '''
关于怎么在Pytorch中只导入部分模型参数问题的解答就分享到这里了，希望以上内容可以对大家有一定的帮助，如果你还有很多疑惑没有解开，可以关注创新互联行业资讯频道了解更多相关知识。