simplexfmrartifact 0.0.4

BentoML artifact framework for simpletransformers

simplexfmrartifact

BentoML artifact framework for simpletransformers.

Installation:

pip install simplexfmrartifact==0.0.4

Usage example (decorate service):

from simplexfmrartifact.simpletransformers import SimpleTransformersModelArtifact

@artifacts([SimpleTransformersModelArtifact('tm_train3_roberta_l_weigh')])
class MyBentoService(BentoService):

Usage example (package model):

svc = MyBentoService()

opts = {
    'classpackage': 'simpletransformers.classification',
    'classname': 'MultiLabelClassificationModel',
    'num_labels': 33,
    'args': {
        'use_multiprocessing': False,
        'silent': True
    }
}

svc.pack('tm_train3_roberta_l_weigh', model_path, opts)

Alternatively, during training:

svc.pack('tm_train3_roberta_l_weigh', {'model': my_trained_model})