Archives/langchain
2
0
Fork 1
mirror of https://github.com/hwchase17/langchain synced 2024-11-18 09:25:54 +00:00
Code Issues Packages Projects Releases Wiki Activity

community[patch]: Fix agenttoolkits imports (#14559)

Browse Source
This commit is contained in:
Bagatur 2023-12-11 14:19:25 -08:00 committed by GitHub
parent 3b5b0f16c6
commit a844b495c4
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
5 changed files with 33 additions and 195 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

88
.scripts/community_split/libs/community/langchain_community/agent_toolkits/conversational_retrieval/openai_functions.py
View File

@ -1,88 +0,0 @@
from __future__ import annotations
from typing import Any, List, Optional, TYPE_CHECKING
from langchain_core.language_models import BaseLanguageModel
from langchain_core.memory import BaseMemory
from langchain_core.messages import SystemMessage
from langchain_core.prompts.chat import MessagesPlaceholder
from langchain_core.tools import BaseTool
if TYPE_CHECKING:
from langchain.agents.agent import AgentExecutor
def _get_default_system_message() -> SystemMessage:
return SystemMessage(
content=(
"Do your best to answer the questions. "
"Feel free to use any tools available to look up "
"relevant information, only if necessary"
)
)
def create_conversational_retrieval_agent(
llm: BaseLanguageModel,
tools: List[BaseTool],
remember_intermediate_steps: bool = True,
memory_key: str = "chat_history",
system_message: Optional[SystemMessage] = None,
verbose: bool = False,
max_token_limit: int = 2000,
**kwargs: Any,
) -> AgentExecutor:
"""A convenience method for creating a conversational retrieval agent.
Args:
llm: The language model to use, should be ChatOpenAI
tools: A list of tools the agent has access to
remember_intermediate_steps: Whether the agent should remember intermediate
steps or not. Intermediate steps refer to prior action/observation
pairs from previous questions. The benefit of remembering these is if
there is relevant information in there, the agent can use it to answer
follow up questions. The downside is it will take up more tokens.
memory_key: The name of the memory key in the prompt.
system_message: The system message to use. By default, a basic one will
be used.
verbose: Whether or not the final AgentExecutor should be verbose or not,
defaults to False.
max_token_limit: The max number of tokens to keep around in memory.
Defaults to 2000.
Returns:
An agent executor initialized appropriately
"""
from langchain.agents.agent import AgentExecutor
from langchain.agents.openai_functions_agent.agent_token_buffer_memory import (
AgentTokenBufferMemory,
)
from langchain.agents.openai_functions_agent.base import OpenAIFunctionsAgent
from langchain.memory.token_buffer import ConversationTokenBufferMemory
if remember_intermediate_steps:
memory: BaseMemory = AgentTokenBufferMemory(
memory_key=memory_key, llm=llm, max_token_limit=max_token_limit
)
else:
memory = ConversationTokenBufferMemory(
memory_key=memory_key,
return_messages=True,
output_key="output",
llm=llm,
max_token_limit=max_token_limit,
)
_system_message = system_message or _get_default_system_message()
prompt = OpenAIFunctionsAgent.create_prompt(
system_message=_system_message,
extra_prompt_messages=[MessagesPlaceholder(variable_name=memory_key)],
)
agent = OpenAIFunctionsAgent(llm=llm, tools=tools, prompt=prompt)
return AgentExecutor(
agent=agent,
tools=tools,
memory=memory,
verbose=verbose,
return_intermediate_steps=remember_intermediate_steps,
**kwargs,
)

103
.scripts/community_split/libs/community/langchain_community/agent_toolkits/vectorstore/base.py
View File

@ -1,103 +0,0 @@
"""VectorStore agent."""
from __future__ import annotations
from typing import Any, Dict, Optional, TYPE_CHECKING
from langchain_core.callbacks import BaseCallbackManager
from langchain_core.language_models import BaseLanguageModel
from langchain_community.agent_toolkits.vectorstore.prompt import PREFIX, ROUTER_PREFIX
from langchain_community.agent_toolkits.vectorstore.toolkit import (
VectorStoreRouterToolkit,
VectorStoreToolkit,
)
if TYPE_CHECKING:
from langchain.agents.agent import AgentExecutor
def create_vectorstore_agent(
llm: BaseLanguageModel,
toolkit: VectorStoreToolkit,
callback_manager: Optional[BaseCallbackManager] = None,
prefix: str = PREFIX,
verbose: bool = False,
agent_executor_kwargs: Optional[Dict[str, Any]] = None,
**kwargs: Any,
) -> AgentExecutor:
"""Construct a VectorStore agent from an LLM and tools.
Args:
llm (BaseLanguageModel): LLM that will be used by the agent
toolkit (VectorStoreToolkit): Set of tools for the agent
callback_manager (Optional[BaseCallbackManager], optional): Object to handle the callback [ Defaults to None. ]
prefix (str, optional): The prefix prompt for the agent. If not provided uses default PREFIX.
verbose (bool, optional): If you want to see the content of the scratchpad. [ Defaults to False ]
agent_executor_kwargs (Optional[Dict[str, Any]], optional): If there is any other parameter you want to send to the agent. [ Defaults to None ]
**kwargs: Additional named parameters to pass to the ZeroShotAgent.
Returns:
AgentExecutor: Returns a callable AgentExecutor object. Either you can call it or use run method with the query to get the response
""" # noqa: E501
from langchain.agents.agent import AgentExecutor
from langchain.agents.mrkl.base import ZeroShotAgent
from langchain.chains.llm import LLMChain
tools = toolkit.get_tools()
prompt = ZeroShotAgent.create_prompt(tools, prefix=prefix)
llm_chain = LLMChain(
llm=llm,
prompt=prompt,
callback_manager=callback_manager,
)
tool_names = [tool.name for tool in tools]
agent = ZeroShotAgent(llm_chain=llm_chain, allowed_tools=tool_names, **kwargs)
return AgentExecutor.from_agent_and_tools(
agent=agent,
tools=tools,
callback_manager=callback_manager,
verbose=verbose,
**(agent_executor_kwargs or {}),
)
def create_vectorstore_router_agent(
llm: BaseLanguageModel,
toolkit: VectorStoreRouterToolkit,
callback_manager: Optional[BaseCallbackManager] = None,
prefix: str = ROUTER_PREFIX,
verbose: bool = False,
agent_executor_kwargs: Optional[Dict[str, Any]] = None,
**kwargs: Any,
) -> AgentExecutor:
"""Construct a VectorStore router agent from an LLM and tools.
Args:
llm (BaseLanguageModel): LLM that will be used by the agent
toolkit (VectorStoreRouterToolkit): Set of tools for the agent which have routing capability with multiple vector stores
callback_manager (Optional[BaseCallbackManager], optional): Object to handle the callback [ Defaults to None. ]
prefix (str, optional): The prefix prompt for the router agent. If not provided uses default ROUTER_PREFIX.
verbose (bool, optional): If you want to see the content of the scratchpad. [ Defaults to False ]
agent_executor_kwargs (Optional[Dict[str, Any]], optional): If there is any other parameter you want to send to the agent. [ Defaults to None ]
**kwargs: Additional named parameters to pass to the ZeroShotAgent.
Returns:
AgentExecutor: Returns a callable AgentExecutor object. Either you can call it or use run method with the query to get the response.
""" # noqa: E501
from langchain.agents.agent import AgentExecutor
from langchain.agents.mrkl.base import ZeroShotAgent
from langchain.chains.llm import LLMChain
tools = toolkit.get_tools()
prompt = ZeroShotAgent.create_prompt(tools, prefix=prefix)
llm_chain = LLMChain(
llm=llm,
prompt=prompt,
callback_manager=callback_manager,
)
tool_names = [tool.name for tool in tools]
agent = ZeroShotAgent(llm_chain=llm_chain, allowed_tools=tool_names, **kwargs)
return AgentExecutor.from_agent_and_tools(
agent=agent,
tools=tools,
callback_manager=callback_manager,
verbose=verbose,
**(agent_executor_kwargs or {}),
)

4
libs/community/langchain_community/agent_toolkits/__init__.py
View File

@ -18,9 +18,6 @@ from langchain_community.agent_toolkits.amadeus.toolkit import AmadeusToolkit
from langchain_community.agent_toolkits.azure_cognitive_services import ( from langchain_community.agent_toolkits.azure_cognitive_services import (
AzureCognitiveServicesToolkit, AzureCognitiveServicesToolkit,
) )
from langchain_community.agent_toolkits.conversational_retrieval.openai_functions import ( # noqa: E501
create_conversational_retrieval_agent,
)
from langchain_community.agent_toolkits.file_management.toolkit import ( from langchain_community.agent_toolkits.file_management.toolkit import (
FileManagementToolkit, FileManagementToolkit,
) )
@ -74,5 +71,4 @@ __all__ = [
"create_pbi_chat_agent", "create_pbi_chat_agent",
"create_spark_sql_agent", "create_spark_sql_agent",
"create_sql_agent", "create_sql_agent",
"create_conversational_retrieval_agent",
] ]

0
libs/community/tests/unit_tests/agent_toolkits/__init__.py Normal file
View File

33
libs/community/tests/unit_tests/agent_toolkits/test_imports.py Normal file
View File

@ -0,0 +1,33 @@
from langchain_community.agent_toolkits import __all__
EXPECTED_ALL = [
"AINetworkToolkit",
"AmadeusToolkit",
"AzureCognitiveServicesToolkit",
"FileManagementToolkit",
"GmailToolkit",
"JiraToolkit",
"JsonToolkit",
"MultionToolkit",
"NasaToolkit",
"NLAToolkit",
"O365Toolkit",
"OpenAPIToolkit",
"PlayWrightBrowserToolkit",
"PowerBIToolkit",
"SlackToolkit",
"SteamToolkit",
"SQLDatabaseToolkit",
"SparkSQLToolkit",
"ZapierToolkit",
"create_json_agent",
"create_openapi_agent",
"create_pbi_agent",
"create_pbi_chat_agent",
"create_spark_sql_agent",
"create_sql_agent",
]
def test_all_imports() -> None:
assert set(__all__) == set(EXPECTED_ALL)