results

results

Classes:

Name	Description
RecordResult	Represents a NER result for a record in a dataset.
ExporterConfig
RecordResultExporter	Responsible for creating a representation of RecordResult that is exportable to
DatasetClassifyResults	Represents classification result on the dataset.

RecordResult(index, entities) dataclass

Represents a NER result for a record in a dataset.

Parameters:

Name	Type	Description	Default
index	int	index of the record in the original dataset	required
entities	list[Prediction]	Dict representation of NERPrediction (NERPrediction.as_dict)	required

Source code in src/nemo_safe_synthesizer/pii_replacer/ner/report/results.py

def __init__(self, index: int, entities: list[Prediction]):
    self.index = index
    self.entities = entities

ExporterConfig(include_json_path=False, include_match=False) dataclass

Attributes:

Name	Type	Description
include_json_path	bool	If set to True, the exported predictions will include "json_path" field for
include_match	bool	If set to True, the exported prediction will contain text that got matched.

include_json_path = field(default=False) class-attribute instance-attribute

If set to True, the exported predictions will include "json_path" field for each entity.

include_match = field(default=False) class-attribute instance-attribute

If set to True, the exported prediction will contain text that got matched.

RecordResultExporter(config=None)

Responsible for creating a representation of RecordResult that is exportable to the outside.

Parameters:

Name	Type	Description	Default
config	Optional[ExporterConfig]	Configuration for how to export the predictions.	None

Source code in src/nemo_safe_synthesizer/pii_replacer/ner/report/results.py

def __init__(self, config: Optional[ExporterConfig] = None):
    if not config:
        config = ExporterConfig()

    self._keys_to_export = {name: name for name in self._pass_through_fields}

    if config.include_json_path:
        self._keys_to_export["json_path"] = "json_path"

    if config.include_match:
        self._keys_to_export["text"] = "match"

DatasetClassifyResults(results=None)

Represents classification result on the dataset.

For now this representation is using whatever is returned from the NER.predict() for simplicity and so we don't need to create a separate data structure (as there may be a lot of results).

Methods:

Name	Description
jsonl	Serializes results into a JSONL-formatted string.

Source code in src/nemo_safe_synthesizer/pii_replacer/ner/report/results.py

def __init__(self, results: Optional[DatasetPredictions] = None):
    if not results:
        results = []

    self.results = results

jsonl(exporter_config=None)

Serializes results into a JSONL-formatted string.

Parameters:

Name	Type	Description	Default
exporter_config	Optional[ExporterConfig]	Config to use for the result exporter.	None

JSONL representation of results. Each line contains classification

Type	Description
str	result for corresponding row in the input dataset.

Source code in src/nemo_safe_synthesizer/pii_replacer/ner/report/results.py

def jsonl(self, exporter_config: Optional[ExporterConfig] = None) -> str:
    """
    Serializes results into a JSONL-formatted string.

    Args:
        exporter_config: Config to use for the result exporter.

    Returns: JSONL representation of results. Each line contains classification
        result for corresponding row in the input dataset.
    """
    exporter = RecordResultExporter(exporter_config)
    return "\n".join(json.dumps(exporter.export(result)) for result in self.results)