genai.text.generation.generation_service module#

pydantic model genai.text.generation.generation_service.BaseConfig[source]#

Bases: BaseServiceConfig

Config:

  • extra: str = forbid

  • validate_assignment: bool = True

  • validate_default: bool = True

field create_execution_options: CreateExecutionOptions = CreateExecutionOptions(throw_on_error=True, ordered=True, concurrency_limit=None, callback=None)#
pydantic model genai.text.generation.generation_service.BaseServices[source]#

Bases: BaseServiceServices

Config:

  • extra: str = forbid

  • validate_assignment: bool = True

  • validate_default: bool = True

field LimitService: type[LimitService] = <class 'genai.text.generation.limits.limit_service.LimitService'>#
pydantic model genai.text.generation.generation_service.CreateExecutionOptions[source]#

Bases: BaseModel

field callback: Callable[[TextGenerationStreamCreateResponse | TextGenerationCreateResponse], None] | None = None#

Callback which is called everytime the response comes.

field concurrency_limit: int | None = None#

Upper bound for concurrent executions (in case the passed value is higher than the API allows, the API’s limit will be used).

Constraints:

  • ge = 1

field ordered: bool = True#

Items will be yielded in the order they were passed in, although they may be processed on the server in different order.

field throw_on_error: bool = True#

Flag indicating whether to throw an error if any error occurs during execution (if disabled, ‘None’ may be returned in case of error).

class genai.text.generation.generation_service.GenerationService[source]#

Bases: BaseService[BaseConfig, BaseServices]

Config#

alias of BaseConfig

Services#

alias of BaseServices

__init__(*, api_client, services=None, config=None)[source]#
Parameters:
compare(*, request, compare_parameters=None, name=None)[source]#
Raises:
Parameters:

  • request (TextGenerationComparisonCreateRequestRequest) –

  • compare_parameters (dict | TextGenerationComparisonParameters | None) –

  • name (str | None) –

Return type:

TextGenerationComparisonCreateResponse

create(*, model_id=None, prompt_id=None, input=None, inputs=None, parameters=None, moderations=None, data=None, execution_options=None)[source]#
Parameters:

  • model_id (str | None) – The ID of the model.

  • prompt_id (str | None) – The ID of the prompt which should be used.

  • input (str | None) – Prompt to process. It is recommended not to leave any trailing spaces.

  • inputs (str | list[str] | None) – Prompt/prompts to process. It is recommended not to leave any trailing spaces.

  • parameters (dict | TextGenerationParameters | None) – Parameters for text generation.

  • moderations (dict | ModerationParameters | None) – Parameters for moderation.

  • data (dict | PromptTemplateData | None) – An optional data object for underlying prompt.

  • execution_options (dict | CreateExecutionOptions | None) – An optional configuration how SDK should work (error handling, limits, callbacks, …)

Yields:

TextGenerationCreateResponse object (server response without modification).

Raises:
Return type:

Generator[TextGenerationCreateResponse, None, None]

Note

To limit number of concurrent requests or change execution procedure, see ‘execute_options’ parameter.

create_stream(*, input=None, model_id=None, prompt_id=None, parameters=None, moderations=None, data=None)[source]#
Yields:

TextGenerationStreamCreateResponse (raw server response object)

Raises:
Parameters:

  • input (str | None) –

  • model_id (str | None) –

  • prompt_id (str | None) –

  • parameters (dict | TextGenerationParameters | None) –

  • moderations (dict | ModerationParameters | None) –

  • data (dict | PromptTemplateData | None) –

Return type:

Generator[TextGenerationStreamCreateResponse, None, None]