PyTorch Developer Podcast
A podcast by Edward Yang, Team PyTorch

Categories:
83 Episodes
-
Code generation
Published: 6/4/2021 -
Why is autograd so complicated
Published: 6/3/2021 -
__torch_function__
Published: 6/2/2021 -
TensorIterator
Published: 6/1/2021 -
native_functions.yaml
Published: 5/28/2021 -
Serialization
Published: 5/27/2021 -
Continuous integration
Published: 5/26/2021 -
Stacked diffs and ghstack
Published: 5/25/2021 -
Shared memory
Published: 5/24/2021 -
Automatic mixed precision
Published: 5/21/2021 -
Conjugate views
Published: 5/20/2021 -
History and constraints of Tensor
Published: 5/19/2021 -
How new operators are authored
Published: 5/18/2021 -
The life and death of Variable
Published: 5/17/2021 -
Backend extensibility
Published: 5/14/2021 -
The road to structured kernels
Published: 5/13/2021 -
Functionalization
Published: 5/12/2021 -
Just enough CUDA to be dangerous
Published: 5/11/2021 -
Inference mode
Published: 5/10/2021 -
Vectorization
Published: 5/7/2021
The PyTorch Developer Podcast is a place for the PyTorch dev team to do bite sized (10-20 min) topics about all sorts of internal development topics in PyTorch.