[pytorch] Add PyTorch backend integration #261
Labels
critical
High priority
framework: pytorch
Related to PyTorch backend
module: datasets
Related to doctr.datasets
module: models
Related to doctr.models
module: transforms
Related to doctr.transforms
topic: build
Related to dependencies and build
Milestone
Integrating PyTorch as a possible backend is a natural evolution of DocTR. Two questions need to be answered though, I added some suggestions.
How do we add the PyTorch counterpart of an object / class / function?
For module
my_module.py
,my_module.tf.py
and put the rest inmy_module.pt.py
. If there are non framework-related features, they can be kept inmy_module
. The latter will reference the tf or pytorch implementation depending on the ibrary buildWhere do we start?
Here are my thoughts on all DocTR modules:
datasets
: theloader.py
module and the call of each datasetdocuments
: nothing framework-specificmodels
: every module will be impactedtransforms
: the call methods will be impacted if we cannot used plain torchvision transformsutils
: nothing framework-specificHow do we ensure that each build is running correctly?
setup.py
to build the specified versionNext steps
The text was updated successfully, but these errors were encountered: