Bedrock을 이용해 chat을 설정합니다.
from botocore.config import Config
from langchain_aws import ChatBedrock
bedrock_region = 'us-east-1'
modelId = "anthropic.claude-3-sonnet-20240229-v1:0"
boto3_bedrock = boto3.client(
service_name='bedrock-runtime',
region_name=bedrock_region,
config=Config(
retries = {
'max_attempts': 30
}
)
)
HUMAN_PROMPT = "\n\nHuman:"
AI_PROMPT = "\n\nAssistant:"
maxOutputTokens = 4096
parameters = {
"max_tokens":maxOutputTokens,
"temperature":0.1,
"top_k":250,
"top_p":0.9,
"stop_sequences": [HUMAN_PROMPT]
}
chat = ChatBedrock(
model_id=modelId,
client=boto3_bedrock,
model_kwargs=parameters,
)class User(BaseModel):
name: str
age: int
structured_llm = chat.with_structured_output(User, include_raw=True)
info = structured_llm.invoke("Jason is 25 years old.")
user_info = info['parsed']
print('name: ', user_info.name)
print('age: ', user_info.age)이때의 결과는 아래와 같습니다.
name: Jason
age: 25
LLM의 parsing 결과는 아래와 같습니다.
{
"raw":"AIMessage(content=""",
"additional_kwargs="{
"usage":{
"prompt_tokens":322,
"completion_tokens":50,
"total_tokens":372
},
"stop_reason":"tool_use",
"model_id":"anthropic.claude-3-sonnet-20240229-v1:0"
},
"response_metadata="{
"usage":{
"prompt_tokens":322,
"completion_tokens":50,
"total_tokens":372
},
"stop_reason":"tool_use",
"model_id":"anthropic.claude-3-sonnet-20240229-v1:0"
},
"id=""run-94e916c0-db53-4bd7-aeca-eb30605981a6-0",
"tool_calls="[
{
"name":"User",
"args":{
"name":"Jason",
"age":25
},
"id":"toolu_bdrk_01RQkKNJefn6bogdDzQWHhrq"
}
],
"usage_metadata="{
"input_tokens":322,
"output_tokens":50,
"total_tokens":372
}")",
"parsed":"User(name=""Jason",
age=25),
"parsing_error":"None"
}class AnswerWithJustification(BaseModel):
'''An answer to the user question along with justification for the answer.'''
answer: str
justification: str
structured_llm = chat.with_structured_output(AnswerWithJustification, include_raw=True)
info = structured_llm.invoke("What weighs more a pound of bricks or a pound of feathers")
parsed_info = info['parsed']
print('answer: ', parsed_info.answer)
print('justification: ', parsed_info.justification)이때의 결과는 아래와 같습니다.
answer: A pound of bricks and a pound of feathers weigh the same.
justification: A pound is a unit of weight or mass, not volume. Since a pound of bricks and a pound of feathers both have the same mass (one pound), they must weigh the same amount. The fact that bricks are denser and take up less volume than feathers for the same weight is irrelevant - their weights are equal when the mass is the same. This is often used as a riddle to trick people into thinking the bricks would be heavier due to their greater density, but by definition of the pound unit, equal masses must have equal weights.
from typing import Annotated, List, Tuple, TypedDict
class Plan(BaseModel):
"""List of steps. The updated plan should be in the following format:
<plan>
[\"<step>\", \"<step>\", ...]
</plan>"""
steps: List[str] = Field(
description="Action to perform. If you want to respond to user, use Response. "
"If you need to further use tools to get the answer, use Plan."
)
structured_llm = chat.with_structured_output(Plan, include_raw=True)
info = structured_llm.invoke("To find the hometown of the 2024 Australian Open winner, we would need to follow these steps:
1. Wait for the 2024 Australian Open tennis tournament to take place (typically in January 2024 in Melbourne, Australia).
2. Identify the winner of the men's singles or women's singles tournament.
3. Research biographical information about the 2024 Australian Open winner to determine their hometown or place of birth.
4. The hometown or birthplace of the 2024 Australian Open winner is the final answer.
Since the 2024 Australian Open has not happened yet, we cannot provide the actual hometown until the tournament takes place and the winner is determined.
The key steps are to wait for the event, identify the winner, and then research their biographical details to find their hometown or birthplace.")
parsed_info = info['parsed']
parsed_info.steps이때의 결과는 아래와 같습니다.
{'raw': AIMessage(content='', additional_kwargs={'usage': {'prompt_tokens': 561, 'completion_tokens': 131, 'total_tokens': 692}, 'stop_reason': 'tool_use', 'model_id': 'anthropic.claude-3-sonnet-20240229-v1:0'}, response_metadata={'usage': {'prompt_tokens': 561, 'completion_tokens': 131, 'total_tokens': 692}, 'stop_reason': 'tool_use', 'model_id': 'anthropic.claude-3-sonnet-20240229-v1:0'}, id='run-b6259c7e-ffbd-4ef2-956f-e816ea4a5d66-0', tool_calls=[{'name': 'Plan', 'args': {'steps': ['Wait for the 2024 Australian Open tennis tournament to take place in January 2024.', "Identify the winner of the men's singles or women's singles tournament.", 'Research biographical information about the 2024 Australian Open winner.', 'Determine the hometown or place of birth of the winner from the biographical information.', 'Response: The hometown of the 2024 Australian Open winner is [their hometown/birthplace].']}, 'id': 'toolu_bdrk_01E89VxYe4bPmT7Nmc12rjB9'}], usage_metadata={'input_tokens': 561, 'output_tokens': 131, 'total_tokens': 692}),
'parsed': Plan(steps=['Wait for the 2024 Australian Open tennis tournament to take place in January 2024.', "Identify the winner of the men's singles or women's singles tournament.", 'Research biographical information about the 2024 Australian Open winner.', 'Determine the hometown or place of birth of the winner from the biographical information.', 'Response: The hometown of the 2024 Australian Open winner is [their hometown/birthplace].']),
'parsing_error': None}
['Wait for the 2024 Australian Open tennis tournament to take place in January 2024.',
"Identify the winner of the men's singles or women's singles tournament.",
'Research biographical information about the 2024 Australian Open winner.',
'Determine the hometown or place of birth of the winner from the biographical information.',
'Response: The hometown of the 2024 Australian Open winner is [their hometown/birthplace].']
from pydantic import BaseModel
from typing import Type
from langchain_core.tools import StructuredTool
class MultiplyArgsSchema(BaseModel):
a: int
b: int
class MultiplyTool(StructuredTool):
name: str = "multiply"
description: str = "Multiply two numbers."
args_schema: Type[BaseModel] = MultiplyArgsSchema
def _run(self, a: int, b: int) -> int:
return a * b
# Example usage
tool = MultiplyTool()
result = tool._run(a=3, b=4)
print(result)How to return structured output with a ReAct style agent은 좋은 레퍼런스입니다.
from pydantic import BaseModel, Field
from typing import Literal
from langchain_core.tools import tool
from langchain_anthropic import ChatAnthropic
from langgraph.graph import MessagesState
class WeatherResponse(BaseModel):
"""Respond to the user with this"""
temperature: float = Field(description="The temperature in fahrenheit")
wind_directon: str = Field(
description="The direction of the wind in abbreviated form"
)
wind_speed: float = Field(description="The speed of the wind in km/h")
# Inherit 'messages' key from MessagesState, which is a list of chat messages
class AgentState(MessagesState):
# Final structured response from the agent
final_response: WeatherResponse
@tool
def get_weather(city: Literal["nyc", "sf"]):
"""Use this to get weather information."""
if city == "nyc":
return "It is cloudy in NYC, with 5 mph winds in the North-East direction and a temperature of 70 degrees"
elif city == "sf":
return "It is 75 degrees and sunny in SF, with 3 mph winds in the South-East direction"
else:
raise AssertionError("Unknown city")
tools = [get_weather]
model = ChatAnthropic(model="claude-3-opus-20240229")
model_with_tools = model.bind_tools(tools)
model_with_structured_output = model.with_structured_output(WeatherResponse)