simplexfmrartifact 0.0.13

BentoML artifact framework for simpletransformers

simplexfmrartifact

BentoML artifact framework for simpletransformers.

Installation:

pip install simplexfmrartifact

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()

metadata = {
    "classpackage": "simpletransformers.classification", 
    "classname": "ClassificationModel", 
    "opts": {
        "use_cuda": true, 
        "num_labels": 2, 
        "args": {
            "use_multiprocessing": false, 
            "silent": true, 
            "eval_batch_size": 10, 
            "fp16": false
        }
    }
}

svc.pack(model_name, model_path, metadata)

Alternatively, during training:

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