Warning: This document is for the development version of Rasa. The latest version is 1.10.8.

Agent

class rasa.core.agent.Agent(domain=None, policies=None, interpreter=None, generator=None, tracker_store=None, lock_store=None, action_endpoint=None, fingerprint=None, model_directory=None, model_server=None, remote_storage=None, path_to_model_archive=None)

The Agent class provides a convenient interface for the most important Rasa functionality.

This includes training, handling messages, loading a dialogue model, getting the next action, and handling a channel.

create_processor(preprocessor=None)

Instantiates a processor based on the set state of the agent.

Return type

MessageProcessor

async execute_action(sender_id, action, output_channel, policy, confidence)

Handle a single message.

Return type

DialogueStateTracker

async handle_message(message, message_preprocessor=None, **kwargs)

Handle a single message.

Return type

Optional[List[Dict[str, Any]]]

async handle_text(text_message, message_preprocessor=None, output_channel=None, sender_id='default')

Handle a single message.

If a message preprocessor is passed, the message will be passed to that function first and the return value is then used as the input for the dialogue engine.

The return value of this function depends on the output_channel. If the output channel is not set, set to None, or set to CollectingOutputChannel this function will return the messages the bot wants to respond.

Example
>>> from rasa.core.agent import Agent
>>> from rasa.core.interpreter import RasaNLUInterpreter
>>> agent = Agent.load("examples/moodbot/models")
>>> await agent.handle_text("hello")
[u'how can I help you?']
Return type

Optional[List[Dict[str, Any]]]

is_core_ready()

Check if all necessary components and policies are ready to use the agent.

Return type

bool

is_ready()

Check if all necessary components are instantiated to use agent.

Policies might not be available, if this is an NLU only agent.

Return type

bool

classmethod load(model_path, interpreter=None, generator=None, tracker_store=None, lock_store=None, action_endpoint=None, model_server=None, remote_storage=None, path_to_model_archive=None)

Load a persisted model from the passed path.

Return type

Agent

async load_data(training_resource, remove_duplicates=True, unique_last_num_states=None, augmentation_factor=50, tracker_limit=None, use_story_concatenation=True, debug_plots=False, exclusion_percentage=None)

Load training data from a resource.

Return type

List[DialogueStateTracker]

async log_message(message, message_preprocessor=None, **kwargs)

Append a message to a dialogue - does not predict actions.

Return type

DialogueStateTracker

async parse_message_using_nlu_interpreter(message_data, tracker=None)

Handles message text and intent payload input messages.

The return value of this function is parsed_data.

Parameters
  • message_data (Text) – Contain the received message in text or intent payload format.

  • tracker (DialogueStateTracker) – Contains the tracker to be used by the interpreter.

Returns

The parsed message.

Example

{ “text”: ‘/greet{“name”:”Rasa”}’, “intent”: {“name”: “greet”, “confidence”: 1.0}, “intent_ranking”: [{“name”: “greet”, “confidence”: 1.0}], “entities”: [{“entity”: “name”, “start”: 6, “end”: 21, “value”: “Rasa”}], }

Return type

Dict[str, Any]

persist(model_path)

Persists this agent into a directory for later loading and usage.

Return type

None

async predict_next(sender_id, **kwargs)

Handle a single message.

Return type

Optional[Dict[str, Any]]

toggle_memoization(activate)

Toggles the memoization on and off.

If a memoization policy is present in the ensemble, this will toggle the prediction of that policy. When set to False the Memoization policies present in the policy ensemble will not make any predictions. Hence, the prediction result from the ensemble always needs to come from a different policy (e.g. TEDPolicy). Useful to test prediction capabilities of an ensemble when ignoring memorized turns from the training data.

Return type

None

train(training_trackers, **kwargs)

Train the policies / policy ensemble using dialogue data from file.

Parameters
  • training_trackers – trackers to train on

  • **kwargs – additional arguments passed to the underlying ML trainer (e.g. keras parameters)

Return type

None

async trigger_intent(intent_name, entities, output_channel, tracker)

Trigger a user intent, e.g. triggered by an external event.

Return type

None