Deep Learning Section 027
PyTorch Tensors
The one object every PyTorch program is made of, and the handful of shape tricks that stop most errors before they happen.
10 of 10 lessons published Three reading levels on every lesson
Lessons in order
Work top to bottom. Each lesson assumes the one above it.
- Installing PyTorch with the right CUDA build
- Creating tensors and choosing a dtype
- Shapes and broadcasting
- reshape, view, permute and contiguity
- Reductions and the dim argument
- Indexing and boolean masks
- gather, scatter and index_select
- einsum for tensor operations
- Moving tensors between CPU and GPU
- Seeds and reproducible runs