我們只能指定計算圖的leaf節點的requires_grad變數來決定改變量是否記錄梯度，而不能指定它們運算產生的節點的requires_grad，它們是否要梯度取決於它們的輸入節點，它們的輸入節點只要有乙個requires_grad是true，那麼它的requires_grad也是true.
x = torch.randn(2,
100)
x.requires_grad =
false
w = torch.randn(10,
100)
w2 = torch.randn(3,
10)w.requires_grad =
true
2w.requires_grad =
true
y = x @ w.t(
)z = y @ w2.t(
)print
(y.requires_grad, z.requires_grad)
z.sum()
.backward(
)
import torch
from torch import nn
defhook
(grad)
:print
(grad)
x = torch.randn(2,
100)
x.requires_grad =
false
w = torch.randn(10,
100)
w2 = torch.randn(3,
10)w.requires_grad =
true
w2.requires_grad =
true
y = x @ w.t(
)z = y @ w2.t(
)y.register_hook(hook)
z.sum()
.backward(
)# invoke get_grad('y') here
import torch
class
gradcollector
(object):
def__init__
(self)
: self.grads =
def__call__
(self, name:
str)
:def
hook
(grad)
: self.grads[name]
= grad
return hook
x = torch.randn(2,
100)
x.requires_grad =
false
w = torch.randn(10,
100)
w2 = torch.randn(3,
10)w.requires_grad =
true
w2.requires_grad =
true
y = x @ w.t(
)z = y @ w2.t(
)grad_collector = gradcollector(
)y.register_hook(grad_collector(
"y")
)z.register_hook(grad_collector(
'z'))z.
sum(
).backward(
)print
(grad_collector.grads[
'y']
)print
(grad_collector.grads[
'z']
)
Pytorch Tensor和tensor的區別
在pytorch中，tensor和tensor都能用於生成新的張量 a torch.tensor 1 2 a tensor 1 2.a torch.tensor 1 2 a tensor 1 2 首先，我們需要明確一下，torch.tensor 是python類，更明確地說，是預設張量型別torch...
pytorch tensor 篩選排除
篩選排除還沒找到答案 取數運算 正好遇到乙個需求。我有m行k列的乙個表a，和乙個長為m的索引列表b。b中儲存著，取每行第幾列的元素。這種情況下，你用普通的索引是會失效的。import torch a torch.longtensor 1,2,3 4,5,6 b torch.longtensor 0,...
Pytorch tensor的感知機
單層感知機的主要步驟 單層感知機梯度的推導 要進行優化的是w，對w進行梯度下降 a torch.randn 1,10 a是乙個 1,10 的向量 w torch.randn 1,10,requires grad true w是乙個可導的 1,10 的向量 1.2.經過乙個sigmoid啟用函式 o ...