agent

Module Contents

Classes

Agent	Agent is the basic unit of the system. It is responsible for
AsyncAgentMixin
AsyncAgent	Asynchronous variant of the Agent class
StreamingAgentMixin	Component that makes agent calling output a streaming response.
AsyncStreamingAgentMixin	Component that makes asynchronous agent calling output a streaming response.
StreamingAgent	Streaming variant of the Agent class
AsyncStreamingAgent	Streaming variant of the AsyncAgent class
Sequential	Sequential is an agent container that forwards messages to each agent
AsyncSequential	Sequential is an agent container that forwards messages to each agent
StreamingSequential	Streaming variant of the Sequential class
AsyncStreamingSequential	Streaming variant of the AsyncSequential class
AgentContainerMixin
AgentList	Agent is the basic unit of the system. It is responsible for
AgentDict	Agent is the basic unit of the system. It is responsible for

class lagent.agents.agent.Agent(llm=None, template=None, memory=dict(type=Memory), output_format=None, aggregator=dict(type=DefaultAggregator), name=None, description=None, hooks=None)

Agent is the basic unit of the system. It is responsible for communicating with the LLM, managing the memory, and handling the message aggregation and parsing. It can also be extended with hooks

参数:

• llm (Union[BaseLLM, Dict]) – The language model used by the agent.

• template (Union[PromptTemplate, str]) – The template used to format the messages.

• memory (Dict) – The memory used by the agent.

• output_format (Dict) – The output format used by the agent.

• aggregator (Dict) – The aggregator used by the agent.

• name (Optional[str]) – The name of the agent.

• description (Optional[str]) – The description of the agent.

• hooks (Optional[Union[List[Dict], Dict]]) – The hooks used by the agent.

返回:

The response message.

返回类型:

AgentMessage

name = 'Agent'

llm: lagent.llms.BaseLLM = None

memory: lagent.memory.MemoryManager

output_format: lagent.prompts.parsers.StrParser = None

template = None

description = None

aggregator: lagent.agents.aggregator.DefaultAggregator

update_memory(message, session_id=0)

forward(*message, session_id=0, **kwargs)

参数:

message (lagent.schema.AgentMessage) –

返回类型:

Union[lagent.schema.AgentMessage, str]

state_dict(session_id=None, prefix='', destination=None)

返回类型:

Dict

load_state_dict(state_dict, session_id=0)

参数:

state_dict (Dict) –

register_hook(hook)

参数:

hook (Callable) –

reset(session_id=0, keypath=None, recursive=False)

参数:

• keypath (Optional[str]) –

• recursive (bool) –

get_messages(session_id=0, keypath=None)

Get OpenAI format messages from memory.

参数:

• session_id (int) – The session id of the memory.

• keypath (Optional[str]) – The keypath of the sub-agent to get messages from. Default is None.

返回:

The messages from the memory including the sub-agent’s system prompt.

返回类型:

List[dict]

class lagent.agents.agent.AsyncAgentMixin

async forward(*message, session_id=0, **kwargs)

参数:

message (lagent.schema.AgentMessage) –

返回类型:

Union[lagent.schema.AgentMessage, str]

class lagent.agents.agent.AsyncAgent(llm=None, template=None, memory=dict(type=Memory), output_format=None, aggregator=dict(type=DefaultAggregator), name=None, description=None, hooks=None)

Bases: AsyncAgentMixin, Agent

Asynchronous variant of the Agent class

参数:

• llm (Union[lagent.llms.BaseLLM, Dict]) –

• template (Union[lagent.prompts.prompt_template.PromptTemplate, str, dict, List[dict]]) –

• memory (Dict) –

• output_format (Optional[Dict]) –

• aggregator (Dict) –

• name (Optional[str]) –

• description (Optional[str]) –

• hooks (Optional[Union[List[Dict], Dict]]) –

class lagent.agents.agent.StreamingAgentMixin

Component that makes agent calling output a streaming response.

forward(*message, session_id=0, **kwargs)

参数:

message (lagent.schema.AgentMessage) –

返回类型:

Generator[Union[lagent.schema.AgentMessage, Tuple[lagent.schema.ModelStatusCode, str]], None, None]

class lagent.agents.agent.AsyncStreamingAgentMixin

Component that makes asynchronous agent calling output a streaming response.

async forward(*message, session_id=0, **kwargs)

参数:

message (lagent.schema.AgentMessage) –

返回类型:

AsyncGenerator[Union[lagent.schema.AgentMessage, Tuple[lagent.schema.ModelStatusCode, str]], None]

class lagent.agents.agent.StreamingAgent(llm=None, template=None, memory=dict(type=Memory), output_format=None, aggregator=dict(type=DefaultAggregator), name=None, description=None, hooks=None)

Bases: StreamingAgentMixin, Agent

Streaming variant of the Agent class

参数:

• llm (Union[lagent.llms.BaseLLM, Dict]) –

• template (Union[lagent.prompts.prompt_template.PromptTemplate, str, dict, List[dict]]) –

• memory (Dict) –

• output_format (Optional[Dict]) –

• aggregator (Dict) –

• name (Optional[str]) –

• description (Optional[str]) –

• hooks (Optional[Union[List[Dict], Dict]]) –

class lagent.agents.agent.AsyncStreamingAgent(llm=None, template=None, memory=dict(type=Memory), output_format=None, aggregator=dict(type=DefaultAggregator), name=None, description=None, hooks=None)

Bases: AsyncStreamingAgentMixin, Agent

Streaming variant of the AsyncAgent class

参数:

• llm (Union[lagent.llms.BaseLLM, Dict]) –

• template (Union[lagent.prompts.prompt_template.PromptTemplate, str, dict, List[dict]]) –

• memory (Dict) –

• output_format (Optional[Dict]) –

• aggregator (Dict) –

• name (Optional[str]) –

• description (Optional[str]) –

• hooks (Optional[Union[List[Dict], Dict]]) –

class lagent.agents.agent.Sequential(*agents, **kwargs)

Bases: Agent

Sequential is an agent container that forwards messages to each agent in the order they are added.

参数:

agents (Union[Agent, Iterable]) –

add_agent(name, agent)

参数:

• name (str) –

• agent (Agent) –

forward(*message, session_id=0, exit_at=None, **kwargs)

参数:

• message (lagent.schema.AgentMessage) –

• exit_at (Optional[int]) –

返回类型:

lagent.schema.AgentMessage

class lagent.agents.agent.AsyncSequential(*agents, **kwargs)

Bases: AsyncAgentMixin, Sequential

Sequential is an agent container that forwards messages to each agent in the order they are added.

参数:

agents (Union[Agent, Iterable]) –

async forward(*message, session_id=0, exit_at=None, **kwargs)

参数:

• message (lagent.schema.AgentMessage) –

• exit_at (Optional[int]) –

返回类型:

lagent.schema.AgentMessage

class lagent.agents.agent.StreamingSequential(*agents, **kwargs)

Bases: StreamingAgentMixin, Sequential

Streaming variant of the Sequential class

参数:

agents (Union[Agent, Iterable]) –

forward(*message, session_id=0, exit_at=None, **kwargs)

参数:

• message (lagent.schema.AgentMessage) –

• exit_at (Optional[int]) –

class lagent.agents.agent.AsyncStreamingSequential(*agents, **kwargs)

Bases: AsyncStreamingAgentMixin, Sequential

Streaming variant of the AsyncSequential class

参数:

agents (Union[Agent, Iterable]) –

async forward(*message, session_id=0, exit_at=None, **kwargs)

参数:

• message (lagent.schema.AgentMessage) –

• exit_at (Optional[int]) –

class lagent.agents.agent.AgentContainerMixin

class lagent.agents.agent.AgentList(agents=None)

Bases: Agent, collections.UserList, AgentContainerMixin

Agent is the basic unit of the system. It is responsible for communicating with the LLM, managing the memory, and handling the message aggregation and parsing. It can also be extended with hooks

参数:

• llm (Union[BaseLLM, Dict]) – The language model used by the agent.

• template (Union[PromptTemplate, str]) – The template used to format the messages.

• memory (Dict) – The memory used by the agent.

• output_format (Dict) – The output format used by the agent.

• aggregator (Dict) – The aggregator used by the agent.

• name (Optional[str]) – The name of the agent.

• description (Optional[str]) – The description of the agent.

• hooks (Optional[Union[List[Dict], Dict]]) – The hooks used by the agent.

• agents (Optional[Iterable[Agent]]) –

返回:

The response message.

返回类型:

AgentMessage

name = None

class lagent.agents.agent.AgentDict(agents=None)

Bases: Agent, collections.UserDict, AgentContainerMixin

Agent is the basic unit of the system. It is responsible for communicating with the LLM, managing the memory, and handling the message aggregation and parsing. It can also be extended with hooks

参数:

• llm (Union[BaseLLM, Dict]) – The language model used by the agent.

• template (Union[PromptTemplate, str]) – The template used to format the messages.

• memory (Dict) – The memory used by the agent.

• output_format (Dict) – The output format used by the agent.

• aggregator (Dict) – The aggregator used by the agent.

• name (Optional[str]) – The name of the agent.

• description (Optional[str]) – The description of the agent.

• hooks (Optional[Union[List[Dict], Dict]]) – The hooks used by the agent.

• agents (Optional[Mapping[str, Agent]]) –

返回:

The response message.

返回类型:

AgentMessage

name = None