[英]How can I extract the weight and bias of Linear layers in PyTorch?
In model.state_dict()
, model.parameters()
and model.named_parameters()
weights and biases of nn.Linear()
modules are contained separately, eq fc1.weight
and fc1.bias
.在model.state_dict()
model.parameters()
和model.named_parameters()
的重量和的偏差nn.Linear()
模块分开容纳,当量fc1.weight
和fc1.bias
。 Is there a simple pythonic way to get both of them?有没有一种简单的pythonic方法来获得它们?
Expected example looks similar to this:预期示例与此类似:
layer = model['fc1']
print(layer.weight)
print(layer.bias)
From the full model, no.从完整模型来看,没有。 There isn't.没有。 But you can get the state_dict()
of that particular Module
and then you'd have a single dict
with the weight
and bias
:但是您可以获得该特定Module
的state_dict()
,然后您将拥有一个带有weight
和bias
dict
:
import torch
m = torch.nn.Linear(3, 5) # arbitrary values
l = m.state_dict()
print(l['weight'])
print(l['bias'])
The equivalent in your code would be:您的代码中的等效项是:
layer = model.fc1.state_dict()
print(layer['weight'])
print(layer['bias'])
To extract the Values from a Layer.从图层中提取值。
layer = model['fc1']
print(layer.weight.data[0])
print(layer.bias.data[0])
instead of 0 index you can use which neuron values to be extracted.您可以使用要提取的神经元值而不是 0 索引。
>> nn.Linear(2,3).weight.data
tensor([[-0.4304, 0.4926],
[ 0.0541, 0.2832],
[-0.4530, -0.3752]])
You can recover the named parameters for each linear layer in your model like so:您可以为模型中的每个线性层恢复命名参数,如下所示:
from torch import nn
for layer in model.children():
if isinstance(layer, nn.Linear):
print(layer.state_dict()['weight'])
print(layer.state_dict()['bias'])
声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.