nemo_voice_agent.evaluation.scenarios

Subpackages

nemo_voice_agent.evaluation.scenarios.data

Submodules

nemo_voice_agent.evaluation.scenarios.classes

Package Contents

Functions

Name	Description
`get_eval_scenario`	Get an evaluation scenario by name.
`list_eval_scenarios`	List all evaluation scenarios.
`register_eval_scenario`	Class decorator that registers a scenario class into ALL_EVAL_SCENARIOS.

Data

ALL_EVAL_SCENARIOS

END_CONVERSATION_GUIDELINE

EXECUTION_HONESTY_GUIDELINE

API

nemo_voice_agent.evaluation.scenarios.get_eval_scenario(
    name: str,
    kwargs = {}
) -> typing.Optional[nemo_voice_agent.evaluation.scenarios.classes.Scenario]

Get an evaluation scenario by name.

nemo_voice_agent.evaluation.scenarios.list_eval_scenarios() -> typing.List[str]

List all evaluation scenarios.

nemo_voice_agent.evaluation.scenarios.register_eval_scenario(
    cls
)

Class decorator that registers a scenario class into ALL_EVAL_SCENARIOS.

nemo_voice_agent.evaluation.scenarios.ALL_EVAL_SCENARIOS: Dict[str, Scenario] = {}

nemo_voice_agent.evaluation.scenarios.END_CONVERSATION_GUIDELINE = "When the user has indicated they have no further requests and you have exchange...

nemo_voice_agent.evaluation.scenarios.EXECUTION_HONESTY_GUIDELINE = "Distinguish 'discussed' from 'done': a step counts as completed only when you s...