sparknlp_jsl._tf_graph_builders.graph_builders.TFGraphBuilderFactory#
- class sparknlp_jsl._tf_graph_builders.graph_builders.TFGraphBuilderFactory[source]#
Bases:
object
Factory class to create the the different tensorflow graphs for ner_dl, generic_classifier, assertion_dl, relation_extraction annotators in spark-nlp healthcare
Methods
__init__
(*args, **kwargs)build
(model_name, build_params, model_location)Method that create the tf graph.
Method that return the available tf models in spark-nlp healthcare Examples -------- >>> from sparknlp_jsl.training import tf_graph >>> tf_graph.get_models()
print_model_params
(model_name)Method that return the params allowed for the tf model.This method return the params with the description for every param.
- static build(model_name, build_params, model_location, model_filename='auto')[source]#
Method that create the tf graph.
- Parameters:
- model_name: str
The name of the tf model that you want to build.Model availables ner_dl,generic_classifier,assertion_dl and relation_extraction
- build_params: dict
Configuration params to build the tf graph for the specific model.
- model_location: str
Path where the model will be saved
- model_filename: str
Name of the .rb file. If you put auto the filename will be generated.
Examples
>>> from sparknlp_jsl.training import tf_graph >>> tf_graph.build("assertion_dl",build_params={"n_classes": 10}, model_location="/tmp", model_filename="assertion_dl.pb")
- static get_models()[source]#
Method that return the available tf models in spark-nlp healthcare Examples ——– >>> from sparknlp_jsl.training import tf_graph >>> tf_graph.get_models()
- static print_model_params(model_name)[source]#
Method that return the params allowed for the tf model.This method return the params with the description for every param.
- Parameters:
- model_name: str
The name of the tf model name.Model availables ner_dl,generic_classifier,assertion_dl and relation_extraction
Examples
>>> from sparknlp_jsl.training import tf_graph >>> tf_graph.print_model_params("assertion_dl")