morpheus_llm.llm.services.nemo_llm_service.NeMoLLMService

class NeMoLLMService(

*,

api_key=None,

org_id=None,

base_url=None,

retry_count=5,

)

Bases: LLMService

A service for interacting with NeMo LLM models, this class should be used to create a client for a specific model.

Parameters:

api_keystr, optional

The API key for the LLM service, by default None. If None the API key will be read from the NGC_API_KEY environment variable. If neither are present an error will be raised., by default None

org_idstr, optional

The organization ID for the LLM service, by default None. If None the organization ID will be read from the NGC_ORG_ID environment variable. This value is only required if the account associated with the api_key is a member of multiple NGC organizations, by default None

base_urlstr, optional

The api host url, by default None. If None the url will be read from the NGC_BASE_URL environment variable. If neither are present the NeMo default will be used, by default None

retry_countint, optional

The number of times to retry a request before raising an exception, by default 5

Methods

create(service_type, *service_args, ...)	Returns a service for interacting with LLM models.
get_client(*, model_name, **model_kwargs)	Returns a client for interacting with a specific model.

APIKey
BaseURL
OrgId

class APIKey(value=None, use_env=True)

Bases: EnvConfigValue

Attributes:

source

use_env

value

class BaseURL(value=None, use_env=True)

Bases: EnvConfigValue

Attributes:

source

use_env

value

class OrgId(value=None, use_env=True)

Bases: EnvConfigValue

Attributes:

source

use_env

value

static create(service_type, *service_args, **service_kwargs)

Returns a service for interacting with LLM models.

Parameters:

service_typestr

The type of the service to create

service_argslist

Additional arguments to pass to the service.

service_kwargsdict[str, typing.Any]

Additional keyword arguments to pass to the service.

get_client(*, model_name, **model_kwargs)

Returns a client for interacting with a specific model. This method is the preferred way to create a client.

Parameters:

model_namestr

The name of the model to create a client for.

model_kwargsdict[str, typing.Any]

Additional keyword arguments to pass to the model when generating text.