orkgnlp.nli.templates.encoder.TemplatesRecommenderEncoder
- class TemplatesRecommenderEncoder(templates)[source]
Bases:
ORKGNLPBaseEncoder
The TemplatesRecommenderEncoder encodes the given input to the arguments needed to execute a BertForSequenceClassification model.
- Parameters
templates (
List
[Dict
[str
,str
]]) – templates used for training the service models as premises.
Methods
Encodes the
raw_input
to a model-friendly format.Releases the memory of all available attributes in a pipeline component.
- encode(raw_input, **kwargs)[source]
Encodes the
raw_input
to a model-friendly format.- Parameters
raw_input (
Any
) – The user’s input to be encoded.- Return type
Tuple
[Any
,Dict
[str
,Any
]]- Returns
The model-friendly output and kwargs.
- release_memory()
Releases the memory of all available attributes in a pipeline component.