Saving and Loading
All PADL Transforms can be saved via save()
:
from padl import save
save(my_pipeline, 'mypipeline.padl')
This creates a folder
mypipeline.padl/
├── transform.py
└── versions.txt
containing a python file defining the Transform and a file containing a list of all package dependencies and their versions.
When saving Pipelines which include PyTorch Module
s as Transforms, checkpoint files with all parameters are stored, too.
Saved Transforms can be loaded with load()
:
from padl import load
my_pipeline = load('mypipeline.padl')
Learn in the next section how to print and slice Transforms in interactive sessions.