PyTorch

uni
PyTorch is an open-source machine learning library developed by Facebook’s AI Research lab (FAIR).

Tensors

import torch
 
x = torch.tensor([1,2,2,2,3,4])
.zeros(3,4)
.rand
.randn
 
# moving tensor to GPU
if torch.cuda.is_available():
	x_gpu = x.to('cuda')
	# oppure
	x_gpu = x.cuda()

Autograd

x = torch.tensor([2,0], requires_grad=True)
y = //
z = x**2 + y**3
 
z.backward() # compute gradients
 
print(x.grad)
print(y.grad)

• requires_grad=True enables gradient tracking
• backward() computes gradients with respect to all tensors with requires_grad=True

DataLoaders

Data structures that launch a new thread, specialized for loading data and being able to run CPU and GPU in parallel.