gt4sd.algorithms.prediction.topics_zero_shot.core module

Algortihms for topic modelling using zero-shot learning via MLNI models.

Summary

Classes:

TopicsPredictor	Configuration to generate topics.
TopicsZeroShot	Topics prediction algorithm.

Reference

class TopicsZeroShot(configuration, target)

Bases: GeneratorAlgorithm[S, T]

Topics prediction algorithm.

__init__(configuration, target)

Instantiate TopicsZeroShot ready to predict topics.

Parameters

• configuration (AlgorithmConfiguration[~S, ~T]) – domain and application specification defining parameters, types and validations.

• target (Optional[~T, None]) – a target for which to generate items.

Example

An example for predicting topics for a given text:

config = TopicsPredictor()
algorithm = TopicsZeroShot(configuration=config, target="This is a text I want to understand better")
items = list(algorithm.sample(1))
print(items)

get_generator(configuration, target)

Get the function to perform the prediction via TopicsZeroShot’s generator.

Parameters

• configuration (AlgorithmConfiguration[~S, ~T]) – helps to set up specific application of TopicsZeroShot.

• target (Optional[~T, None]) – context or condition for the generation.

Return type

Callable[[~T], Iterable[Any]]

Returns

callable with target predicting topics sorted by relevance.

__abstractmethods__ = frozenset({})

__annotations__ = {'generate': 'Untargeted', 'generator': 'Union[Untargeted, Targeted[T]]', 'max_runtime': 'int', 'max_samples': 'int', 'target': 'Optional[T]'}

__doc__ = 'Topics prediction algorithm.'

__module__ = 'gt4sd.algorithms.prediction.topics_zero_shot.core'

__orig_bases__ = (gt4sd.algorithms.core.GeneratorAlgorithm[~S, ~T],)

__parameters__ = (~S, ~T)

_abc_impl = <_abc._abc_data object>

class TopicsPredictor(*args, **kwargs)

Bases: TopicsPredictor, Generic[T]

Configuration to generate topics.

algorithm_type: ClassVar[str] = 'prediction'

General type of generative algorithm.

domain: ClassVar[str] = 'nlp'

General application domain. Hints at input/output types.

algorithm_version: str = 'dbpedia'

To differentiate between different versions of an application.

There is no imposed naming convention.

model_name: str = 'facebook/bart-large-mnli'

get_target_description()

Get description of the target for generation.

Return type

Dict[str, str]

Returns

target description.

get_conditional_generator(resources_path)

Instantiate the actual generator implementation.

Parameters

resources_path (str) – local path to model files.

Return type

ZeroShotClassifier

Returns

instance with generate_batch method for targeted generation.

__annotations__ = {'algorithm_application': 'ClassVar[str]', 'algorithm_name': 'ClassVar[str]', 'algorithm_type': typing.ClassVar[str], 'algorithm_version': <class 'str'>, 'domain': typing.ClassVar[str], 'model_name': <class 'str'>}

__dataclass_fields__ = {'algorithm_application': Field(name='algorithm_application',type=typing.ClassVar[str],default='TopicsPredictor',default_factory=<dataclasses._MISSING_TYPE object>,init=True,repr=True,hash=None,compare=True,metadata=mappingproxy({}),kw_only=<dataclasses._MISSING_TYPE object>,_field_type=_FIELD_CLASSVAR), 'algorithm_name': Field(name='algorithm_name',type=typing.ClassVar[str],default='TopicsZeroShot',default_factory=<dataclasses._MISSING_TYPE object>,init=True,repr=True,hash=None,compare=True,metadata=mappingproxy({}),kw_only=<dataclasses._MISSING_TYPE object>,_field_type=_FIELD_CLASSVAR), 'algorithm_type': Field(name='algorithm_type',type=typing.ClassVar[str],default='prediction',default_factory=<dataclasses._MISSING_TYPE object>,init=True,repr=True,hash=None,compare=True,metadata=mappingproxy({}),kw_only=<dataclasses._MISSING_TYPE object>,_field_type=_FIELD_CLASSVAR), 'algorithm_version': Field(name='algorithm_version',type=<class 'str'>,default='dbpedia',default_factory=<dataclasses._MISSING_TYPE object>,init=True,repr=True,hash=None,compare=True,metadata=mappingproxy({}),kw_only=False,_field_type=_FIELD), 'domain': Field(name='domain',type=typing.ClassVar[str],default='nlp',default_factory=<dataclasses._MISSING_TYPE object>,init=True,repr=True,hash=None,compare=True,metadata=mappingproxy({}),kw_only=<dataclasses._MISSING_TYPE object>,_field_type=_FIELD_CLASSVAR), 'model_name': Field(name='model_name',type=<class 'str'>,default='facebook/bart-large-mnli',default_factory=<dataclasses._MISSING_TYPE object>,init=True,repr=True,hash=None,compare=True,metadata=mappingproxy({'description': 'MLNI model name to use. If the  model is not found in the cache, a download from HuggingFace will be attempted.'}),kw_only=False,_field_type=_FIELD)}

__dataclass_params__ = _DataclassParams(init=True,repr=True,eq=True,order=False,unsafe_hash=False,frozen=False)

__doc__ = 'Configuration to generate topics.'

__eq__(other)

Return self==value.

__hash__ = None

__init__(*args, **kwargs)

__match_args__ = ('algorithm_version', 'model_name')

__module__ = 'gt4sd.algorithms.prediction.topics_zero_shot.core'

__orig_bases__ = (<class 'types.TopicsPredictor'>, typing.Generic[~T])

__parameters__ = (~T,)

__pydantic_complete__ = True

__pydantic_config__ = {}

__pydantic_core_schema__ = {'cls': <class 'gt4sd.algorithms.prediction.topics_zero_shot.core.TopicsPredictor'>, 'config': {'title': 'TopicsPredictor'}, 'fields': ['algorithm_version', 'model_name'], 'frozen': False, 'metadata': {'pydantic_js_annotation_functions': [], 'pydantic_js_functions': [functools.partial(<function modify_model_json_schema>, cls=<class 'gt4sd.algorithms.prediction.topics_zero_shot.core.TopicsPredictor'>, title=None)]}, 'post_init': False, 'ref': 'types.TopicsPredictor:94662829412384', 'schema': {'collect_init_only': False, 'computed_fields': [], 'dataclass_name': 'TopicsPredictor', 'fields': [{'type': 'dataclass-field', 'name': 'algorithm_version', 'schema': {'type': 'default', 'schema': {'type': 'str'}, 'default': 'dbpedia'}, 'kw_only': False, 'init': True, 'metadata': {'pydantic_js_functions': [], 'pydantic_js_annotation_functions': [<function get_json_schema_update_func.<locals>.json_schema_update_func>]}}, {'type': 'dataclass-field', 'name': 'model_name', 'schema': {'type': 'default', 'schema': {'type': 'str'}, 'default': 'facebook/bart-large-mnli'}, 'kw_only': False, 'init': True, 'metadata': {'pydantic_js_functions': [], 'pydantic_js_annotation_functions': [<function get_json_schema_update_func.<locals>.json_schema_update_func>]}}], 'type': 'dataclass-args'}, 'slots': True, 'type': 'dataclass'}

__pydantic_decorators__ = DecoratorInfos(validators={}, field_validators={}, root_validators={}, field_serializers={}, model_serializers={}, model_validators={}, computed_fields={})

__pydantic_fields__ = {'algorithm_version': FieldInfo(annotation=str, required=False, default='dbpedia', init=True, init_var=False, kw_only=False), 'model_name': FieldInfo(annotation=str, required=False, default='facebook/bart-large-mnli', description='MLNI model name to use. If the  model is not found in the cache, a download from HuggingFace will be attempted.', init=True, init_var=False, kw_only=False)}

__pydantic_serializer__ = SchemaSerializer(serializer=Dataclass(     DataclassSerializer {         class: Py(             0x0000561868521020,         ),         serializer: Fields(             GeneralFieldsSerializer {                 fields: {                     "model_name": SerField {                         key_py: Py(                             0x00007f1dc38f1170,                         ),                         alias: None,                         alias_py: None,                         serializer: Some(                             WithDefault(                                 WithDefaultSerializer {                                     default: Default(                                         Py(                                             0x00007f1dc3937640,                                         ),                                     ),                                     serializer: Str(                                         StrSerializer,                                     ),                                 },                             ),                         ),                         required: true,                     },                     "algorithm_version": SerField {                         key_py: Py(                             0x00007f1dc3951ac0,                         ),                         alias: None,                         alias_py: None,                         serializer: Some(                             WithDefault(                                 WithDefaultSerializer {                                     default: Default(                                         Py(                                             0x00007f1dc39393b0,                                         ),                                     ),                                     serializer: Str(                                         StrSerializer,                                     ),                                 },                             ),                         ),                         required: true,                     },                 },                 computed_fields: Some(                     ComputedFields(                         [],                     ),                 ),                 mode: SimpleDict,                 extra_serializer: None,                 filter: SchemaFilter {                     include: None,                     exclude: None,                 },                 required_fields: 2,             },         ),         fields: [             Py(                 0x00007f1ea52ed250,             ),             Py(                 0x00007f1ea6162b70,             ),         ],         name: "TopicsPredictor",     }, ), definitions=[])

__pydantic_validator__ = SchemaValidator(title="TopicsPredictor", validator=Dataclass(     DataclassValidator {         strict: false,         validator: DataclassArgs(             DataclassArgsValidator {                 fields: [                     Field {                         kw_only: false,                         name: "algorithm_version",                         py_name: Py(                             0x00007f1ea52ed250,                         ),                         init: true,                         init_only: false,                         lookup_key: Simple {                             key: "algorithm_version",                             py_key: Py(                                 0x00007f1dc39343a0,                             ),                             path: LookupPath(                                 [                                     S(                                         "algorithm_version",                                         Py(                                             0x00007f1dc3951a70,                                         ),                                     ),                                 ],                             ),                         },                         validator: WithDefault(                             WithDefaultValidator {                                 default: Default(                                     Py(                                         0x00007f1dc39393b0,                                     ),                                 ),                                 on_error: Raise,                                 validator: Str(                                     StrValidator {                                         strict: false,                                         coerce_numbers_to_str: false,                                     },                                 ),                                 validate_default: false,                                 copy_default: false,                                 name: "default[str]",                                 undefined: Py(                                     0x00007f1ea71db950,                                 ),                             },                         ),                         frozen: false,                     },                     Field {                         kw_only: false,                         name: "model_name",                         py_name: Py(                             0x00007f1ea6162b70,                         ),                         init: true,                         init_only: false,                         lookup_key: Simple {                             key: "model_name",                             py_key: Py(                                 0x00007f1dc38b6ab0,                             ),                             path: LookupPath(                                 [                                     S(                                         "model_name",                                         Py(                                             0x00007f1dc39328b0,                                         ),                                     ),                                 ],                             ),                         },                         validator: WithDefault(                             WithDefaultValidator {                                 default: Default(                                     Py(                                         0x00007f1dc3937640,                                     ),                                 ),                                 on_error: Raise,                                 validator: Str(                                     StrValidator {                                         strict: false,                                         coerce_numbers_to_str: false,                                     },                                 ),                                 validate_default: false,                                 copy_default: false,                                 name: "default[str]",                                 undefined: Py(                                     0x00007f1ea71db950,                                 ),                             },                         ),                         frozen: false,                     },                 ],                 positional_count: 2,                 init_only_count: None,                 dataclass_name: "TopicsPredictor",                 validator_name: "dataclass-args[TopicsPredictor]",                 extra_behavior: Ignore,                 extras_validator: None,                 loc_by_alias: true,             },         ),         class: Py(             0x0000561868521020,         ),         fields: [             Py(                 0x00007f1ea52ed250,             ),             Py(                 0x00007f1ea6162b70,             ),         ],         post_init: None,         revalidate: Never,         name: "TopicsPredictor",         frozen: false,         slots: true,     }, ), definitions=[], cache_strings=True)

__repr__()

Return repr(self).

__signature__ = <Signature (algorithm_version: str = 'dbpedia', model_name: str = 'facebook/bart-large-mnli') -> None>

__wrapped__

alias of TopicsPredictor