IOAI ML Notes Programming Fundamentals
PyTorch Tensor Manipulation
Common tensor shape, indexing, broadcasting, and device/type operations in PyTorch.
Syllabus Map
Tensor Manipulation
import torch
# Reshape / view
X = torch.randn(8, 4)
X.view(4, 8)
X.reshape(2, 16)
# Transpose and permute
X.T
X.permute(1, 0)
# Concatenate / stack
torch.cat([X, X], dim=0)
torch.stack([X, X], dim=0)
# Slicing
X[:2]
X[:, 1:3]
# Broadcasting
X + 3
X + torch.randn(1, 4)
# Type and device
X.float()
X.long()
X.to("cpu")