Unpickling export

load_data[source]

load_data(path:Path=Path('.'), fn='data')

Opens pkl file containing exported Transform information

load_model[source]

load_model(path:Path='.', fn='model', cpu=True)

Pipelines

These generate our transform "pipelines" (as we're not using fastcore's Pipeline) to pass our data through

get_tfm[source]

get_tfm(key, tfms)

Makes a transform from key. Class or function must be in global memory (imported)

generate_pipeline[source]

generate_pipeline(tfms, order=True)

Generate pipe of transforms from dict and (potentially) sort them

make_pipelines[source]

make_pipelines(tfms)

Make item and batch transform pipelines