mirror of https://github.com/hwchase17/langchain
Harrison Chase
1 year ago
committed by
GitHub
parent
a795c3d860
commit
6556a8fcfd
@ -0,0 +1,274 @@
|
||||
{
|
||||
"cells": [
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 1,
|
||||
"id": "9926203f",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"import os\n",
|
||||
"\n",
|
||||
"os.environ[\"LANGCHAIN_TRACING_V2\"] = \"true\"\n",
|
||||
"os.environ[\"LANGCHAIN_ENDPOINT\"] = \"https://api.smith.langchain.com\"\n",
|
||||
"os.environ[\"LANGCHAIN_API_KEY\"] = \"\""
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 2,
|
||||
"id": "45bc4149",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"agent_instructions = \"\"\"You are a helpful assistant. Help the user answer any questions.\n",
|
||||
"\n",
|
||||
"You have access to the following tools:\n",
|
||||
"\n",
|
||||
"{tools}\n",
|
||||
"\n",
|
||||
"In order to use a tool, you can use <tool></tool> and <tool_input></tool_input> tags. \\\n",
|
||||
"You will then get back a response in the form <observation></observation>\n",
|
||||
"For example, if you have a tool called 'search' that could run a google search, in order to search for the weather in SF you would respond:\n",
|
||||
"\n",
|
||||
"<tool>search</tool><tool_input>weather in SF</tool_input>\n",
|
||||
"<observation>64 degrees</observation>\n",
|
||||
"\n",
|
||||
"When you are done, respond with a final answer between <final_answer></final_answer>. For example:\n",
|
||||
"\n",
|
||||
"<final_answer>The weather in SF is 64 degrees</final_answer>\n",
|
||||
"\n",
|
||||
"Begin!\n",
|
||||
"\n",
|
||||
"Question: {question}\"\"\""
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 3,
|
||||
"id": "4da4c0d2",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stderr",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"/Users/harrisonchase/.pyenv/versions/3.9.1/envs/langchain/lib/python3.9/site-packages/deeplake/util/check_latest_version.py:32: UserWarning: A newer version of deeplake (3.6.14) is available. It's recommended that you update to the latest version using `pip install -U deeplake`.\n",
|
||||
" warnings.warn(\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"from langchain.chat_models import ChatAnthropic\n",
|
||||
"from langchain.prompts import ChatPromptTemplate, AIMessagePromptTemplate\n",
|
||||
"from langchain.agents import tool"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 4,
|
||||
"id": "b81e9120",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"model = ChatAnthropic(model=\"claude-2\")"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 5,
|
||||
"id": "5271f612",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"prompt_template = ChatPromptTemplate.from_template(agent_instructions) + AIMessagePromptTemplate.from_template(\"{intermediate_steps}\")"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 6,
|
||||
"id": "83780d81",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"chain = prompt_template | model.bind(stop=[\"</tool_input>\", \"</final_answer>\"])"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 7,
|
||||
"id": "c091d0e1",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"@tool\n",
|
||||
"def search(query: str) -> str:\n",
|
||||
" \"\"\"Search things about current events.\"\"\"\n",
|
||||
" return \"32 degrees\""
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 8,
|
||||
"id": "1e81b05d",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"tool_list = [search]"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 9,
|
||||
"id": "5f0d986f",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"from langchain.agents import Tool, AgentExecutor, BaseSingleActionAgent\n",
|
||||
"from typing import List, Tuple, Any, Union\n",
|
||||
"from langchain.schema import AgentAction, AgentFinish\n",
|
||||
"\n",
|
||||
"\n",
|
||||
"class AnthropicAgent(BaseSingleActionAgent):\n",
|
||||
" \n",
|
||||
" tools: List[Tool]\n",
|
||||
" chain: Any\n",
|
||||
"\n",
|
||||
" @property\n",
|
||||
" def input_keys(self):\n",
|
||||
" return [\"input\"]\n",
|
||||
"\n",
|
||||
" def plan(\n",
|
||||
" self, intermediate_steps: List[Tuple[AgentAction, str]], **kwargs: Any\n",
|
||||
" ) -> Union[AgentAction, AgentFinish]:\n",
|
||||
" \"\"\"Given input, decided what to do.\n",
|
||||
"\n",
|
||||
" Args:\n",
|
||||
" intermediate_steps: Steps the LLM has taken to date,\n",
|
||||
" along with observations\n",
|
||||
" **kwargs: User inputs.\n",
|
||||
"\n",
|
||||
" Returns:\n",
|
||||
" Action specifying what tool to use.\n",
|
||||
" \"\"\"\n",
|
||||
" log = \"\"\n",
|
||||
" for action, observation in intermediate_steps:\n",
|
||||
" log += f\"<tool>{action.tool}</tool><tool_input>{action.tool_input}</tool_input><observation>{observation}</observation>\"\n",
|
||||
" tools = \"\"\n",
|
||||
" for tool in self.tools:\n",
|
||||
" tools += f\"{tool.name}: {tool.description}\\n\"\n",
|
||||
" response = self.chain.invoke({\"intermediate_steps\": log, \"tools\": tools, \"question\": kwargs[\"input\"]})\n",
|
||||
" if \"</tool>\" in response.content:\n",
|
||||
" t, ti = response.content.split(\"</tool>\")\n",
|
||||
" _t = t.split(\"<tool>\")[1]\n",
|
||||
" _ti = ti.split(\"<tool_input>\")[1]\n",
|
||||
" return AgentAction(tool=_t, tool_input=_ti, log=response.content)\n",
|
||||
" elif \"<final_answer>\" in response.content:\n",
|
||||
" t, ti = response.content.split(\"<final_answer>\")\n",
|
||||
" return AgentFinish(return_values={\"output\": ti}, log=response.content)\n",
|
||||
" else:\n",
|
||||
" raise ValueError\n",
|
||||
"\n",
|
||||
" async def aplan(\n",
|
||||
" self, intermediate_steps: List[Tuple[AgentAction, str]], **kwargs: Any\n",
|
||||
" ) -> Union[AgentAction, AgentFinish]:\n",
|
||||
" \"\"\"Given input, decided what to do.\n",
|
||||
"\n",
|
||||
" Args:\n",
|
||||
" intermediate_steps: Steps the LLM has taken to date,\n",
|
||||
" along with observations\n",
|
||||
" **kwargs: User inputs.\n",
|
||||
"\n",
|
||||
" Returns:\n",
|
||||
" Action specifying what tool to use.\n",
|
||||
" \"\"\"\n",
|
||||
" raise ValueError"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 10,
|
||||
"id": "315361c5",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"agent = AnthropicAgent(tools=tool_list, chain=chain)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 11,
|
||||
"id": "bca6096f",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"agent_executor = AgentExecutor(agent=agent, tools=tool_list, verbose=True)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 12,
|
||||
"id": "71b872b1",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"\n",
|
||||
"\n",
|
||||
"\u001b[1m> Entering new AgentExecutor chain...\u001b[0m\n",
|
||||
"\u001b[32;1m\u001b[1;3m <tool>search</tool>\n",
|
||||
"<tool_input>weather in new york\u001b[0m\u001b[36;1m\u001b[1;3m32 degrees\u001b[0m\u001b[32;1m\u001b[1;3m\n",
|
||||
"\n",
|
||||
"<final_answer>The weather in New York is 32 degrees\u001b[0m\n",
|
||||
"\n",
|
||||
"\u001b[1m> Finished chain.\u001b[0m\n"
|
||||
]
|
||||
},
|
||||
{
|
||||
"data": {
|
||||
"text/plain": [
|
||||
"'The weather in New York is 32 degrees'"
|
||||
]
|
||||
},
|
||||
"execution_count": 12,
|
||||
"metadata": {},
|
||||
"output_type": "execute_result"
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"agent_executor.run(\"whats the weather in New york?\")"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "cca87246",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": []
|
||||
}
|
||||
],
|
||||
"metadata": {
|
||||
"kernelspec": {
|
||||
"display_name": "Python 3 (ipykernel)",
|
||||
"language": "python",
|
||||
"name": "python3"
|
||||
},
|
||||
"language_info": {
|
||||
"codemirror_mode": {
|
||||
"name": "ipython",
|
||||
"version": 3
|
||||
},
|
||||
"file_extension": ".py",
|
||||
"mimetype": "text/x-python",
|
||||
"name": "python",
|
||||
"nbconvert_exporter": "python",
|
||||
"pygments_lexer": "ipython3",
|
||||
"version": "3.9.1"
|
||||
}
|
||||
},
|
||||
"nbformat": 4,
|
||||
"nbformat_minor": 5
|
||||
}
|
||||
@ -0,0 +1,118 @@
|
||||
from typing import Any, List, Tuple, Union
|
||||
|
||||
from langchain.agents.agent import AgentOutputParser, BaseSingleActionAgent
|
||||
from langchain.agents.xml.prompt import agent_instructions
|
||||
from langchain.callbacks.base import Callbacks
|
||||
from langchain.chains.llm import LLMChain
|
||||
from langchain.prompts.chat import AIMessagePromptTemplate, ChatPromptTemplate
|
||||
from langchain.schema import AgentAction, AgentFinish
|
||||
from langchain.tools.base import BaseTool
|
||||
|
||||
|
||||
class XMLAgentOutputParser(AgentOutputParser):
|
||||
def parse(self, text: str) -> Union[AgentAction, AgentFinish]:
|
||||
if "</tool>" in text:
|
||||
tool, tool_input = text.split("</tool>")
|
||||
_tool = tool.split("<tool>")[1]
|
||||
_tool_input = tool_input.split("<tool_input>")[1]
|
||||
return AgentAction(tool=_tool, tool_input=_tool_input, log=text)
|
||||
elif "<final_answer>" in text:
|
||||
_, answer = text.split("<final_answer>")
|
||||
return AgentFinish(return_values={"output": answer}, log=text)
|
||||
else:
|
||||
raise ValueError
|
||||
|
||||
def get_format_instructions(self) -> str:
|
||||
raise NotImplementedError
|
||||
|
||||
@property
|
||||
def _type(self) -> str:
|
||||
return "xml-agent"
|
||||
|
||||
|
||||
class XMLAgent(BaseSingleActionAgent):
|
||||
"""Agent that uses XML tags.
|
||||
|
||||
Args:
|
||||
tools: list of tools the agent can choose from
|
||||
llm_chain: The LLMChain to call to predict the next action
|
||||
|
||||
Examples:
|
||||
|
||||
.. code-block:: python
|
||||
|
||||
from langchain.agents import XMLAgent
|
||||
from langchain
|
||||
|
||||
tools = ...
|
||||
model =
|
||||
|
||||
|
||||
"""
|
||||
|
||||
tools: List[BaseTool]
|
||||
"""List of tools this agent has access to."""
|
||||
llm_chain: LLMChain
|
||||
"""Chain to use to predict action."""
|
||||
|
||||
@property
|
||||
def input_keys(self) -> List[str]:
|
||||
return ["input"]
|
||||
|
||||
@staticmethod
|
||||
def get_default_prompt() -> ChatPromptTemplate:
|
||||
return ChatPromptTemplate.from_template(
|
||||
agent_instructions
|
||||
) + AIMessagePromptTemplate.from_template("{intermediate_steps}")
|
||||
|
||||
@staticmethod
|
||||
def get_default_output_parser() -> XMLAgentOutputParser:
|
||||
return XMLAgentOutputParser()
|
||||
|
||||
def plan(
|
||||
self,
|
||||
intermediate_steps: List[Tuple[AgentAction, str]],
|
||||
callbacks: Callbacks = None,
|
||||
**kwargs: Any,
|
||||
) -> Union[AgentAction, AgentFinish]:
|
||||
log = ""
|
||||
for action, observation in intermediate_steps:
|
||||
log += (
|
||||
f"<tool>{action.tool}</tool><tool_input>{action.tool_input}"
|
||||
f"</tool_input><observation>{observation}</observation>"
|
||||
)
|
||||
tools = ""
|
||||
for tool in self.tools:
|
||||
tools += f"{tool.name}: {tool.description}\n"
|
||||
inputs = {
|
||||
"intermediate_steps": log,
|
||||
"tools": tools,
|
||||
"question": kwargs["input"],
|
||||
"stop": ["</tool_input>", "</final_answer>"],
|
||||
}
|
||||
response = self.llm_chain(inputs, callbacks=callbacks)
|
||||
return response[self.llm_chain.output_key]
|
||||
|
||||
async def aplan(
|
||||
self,
|
||||
intermediate_steps: List[Tuple[AgentAction, str]],
|
||||
callbacks: Callbacks = None,
|
||||
**kwargs: Any,
|
||||
) -> Union[AgentAction, AgentFinish]:
|
||||
log = ""
|
||||
for action, observation in intermediate_steps:
|
||||
log += (
|
||||
f"<tool>{action.tool}</tool><tool_input>{action.tool_input}"
|
||||
f"</tool_input><observation>{observation}</observation>"
|
||||
)
|
||||
tools = ""
|
||||
for tool in self.tools:
|
||||
tools += f"{tool.name}: {tool.description}\n"
|
||||
inputs = {
|
||||
"intermediate_steps": log,
|
||||
"tools": tools,
|
||||
"question": kwargs["input"],
|
||||
"stop": ["</tool_input>", "</final_answer>"],
|
||||
}
|
||||
response = await self.llm_chain.acall(inputs, callbacks=callbacks)
|
||||
return response[self.llm_chain.output_key]
|
||||
@ -0,0 +1,21 @@
|
||||
# flake8: noqa
|
||||
agent_instructions = """You are a helpful assistant. Help the user answer any questions.
|
||||
|
||||
You have access to the following tools:
|
||||
|
||||
{tools}
|
||||
|
||||
In order to use a tool, you can use <tool></tool> and <tool_input></tool_input> tags. \
|
||||
You will then get back a response in the form <observation></observation>
|
||||
For example, if you have a tool called 'search' that could run a google search, in order to search for the weather in SF you would respond:
|
||||
|
||||
<tool>search</tool><tool_input>weather in SF</tool_input>
|
||||
<observation>64 degrees</observation>
|
||||
|
||||
When you are done, respond with a final answer between <final_answer></final_answer>. For example:
|
||||
|
||||
<final_answer>The weather in SF is 64 degrees</final_answer>
|
||||
|
||||
Begin!
|
||||
|
||||
Question: {question}"""
|
||||
File diff suppressed because one or more lines are too long
Loading…
Reference in New Issue