경사 하강법

```python import torch import torch.nn.functional as F

create the target tensor

target = torch.FloatTensor([[.1, .2, .3], [.4, .5, .6], [.7, .8, .9]])

create a random tensor with the same shape as the target tensor

x = torch.rand_like(target)

set the requires_grad attribute of x to True, so that gradients will be calculated for x

x.requires_grad = True

print the initial value of x

print(x)

calculate the mean squared error loss between x and the target tensor

loss = F.mse_loss(x, target) loss

set the threshold for the loss

threshold = 1e-5

set the learning rate for the optimization algorithm

learning_rate = 1.

Initialize iteration counter

iter_cnt = 0

loop until the loss is less than the threshold

while loss > threshold: iter_cnt += 1 # calculate the gradients of the loss with respect to x loss.backward()

# update the values of x by subtracting the gradients scaled by the learning rate
x = x - learning_rate * x.grad

# disconnect the tensor from the computation graph and set the requires_grad attribute of x to True again
x.detach_()
x.requires_grad_(True)

# calculate the new loss after the update
loss = F.mse_loss(x, target)

# print the current loss and values of x