autograd.py

import torch
import torch.nn.functional as F

x = torch.tensor([ 1., 2. ])

y = torch.tensor([ 4., 5. ])

z = torch.tensor([ 7., 3. ])

x.requires_grad

(x + y).requires_grad

z.requires_grad = True

(x + z).requires_grad

x = torch.tensor([1., 10.])

x.requires_grad = True

x = torch.tensor([1, 10])

w1 = torch.rand(20, 10).requires_grad_()
b1 = torch.rand(20).requires_grad_()
w2 = torch.rand(5, 20).requires_grad_()
b2 = torch.rand(5).requires_grad_()
x = torch.rand(10)
h = torch.tanh(w1 @ x + b1)
y = torch.tanh(w2 @ h + b2)
target = torch.rand(5)
loss = (y - target).pow(2).mean()


x= torch.tensor([[ 0.8008, -0.2586, 0.5019, -0.2002, -0.7416],
[ 0.0557, 0.6046, 0.0864, -0.5929, 1.2606]])
F.relu(x)

f = nn.Linear(in_features = 10, out_features = 4)
for n, p in f.named_parameters(): print(n, p.size())

x = torch.empty(523, 10).normal_()
y = f(x)
y.size()

f = torch.nn.MSELoss()

x = torch.tensor([[ 3. ]])
y = torch.tensor([[ 0. ]])
f(x, y)