langchain zero to hero agents

?? LangChain Nol-ke-Pahlawan / ? Agen

Mulailah dengan Agen LangChain, bagian dari seri zero-to-hero.

Sebelum Anda mulai

• Tutorial ini menggunakan terminal untuk menginstal dependensi dan menjalankan skrip Python.
• Ketika Anda melihat ? emoji sebelum serangkaian perintah terminal, buka proses terminal baru.
• Saat Anda melihat emoji ♻️ sebelum serangkaian perintah terminal, Anda dapat menggunakan kembali terminal yang sama yang Anda gunakan terakhir kali.

Prasyarat

1. Unduh dan instal Puisi.

2. Siapkan lingkungan Puisi:

?

poetry init --no-interaction --python= " ^3.11 " --dependency=langchain --dependency=langchain-openai --dependency=langchainhub --dependency= " langserve[all] " --dependency=duckduckgo-search
poetry install

3. Dapatkan kunci API OpenAI dan simpan sebagai variabel lingkungan (misalnya: dengan DirEnv):

 export OPENAI_API_KEY=...

Memulai

1. Mari buat skrip agen sederhana.

• Buat file langchain_zero_to_hero_agents/src/main.py dan buat agen sederhana (kode dimodifikasi dari Buku Masak Agen LangChain)

 from langchain import hub
from langchain . agents import AgentExecutor , tool
from langchain . agents . output_parsers import XMLAgentOutputParser
from langchain_openai import ChatOpenAI
from langchain_community . tools import DuckDuckGoSearchResults

#######################
# LangChain Agent Code
#######################

model = ChatOpenAI ( model = "gpt-3.5-turbo" , temperature = 0 , streaming = True )


@ tool
def search ( query : str ) -> str :
    """Search things about current events."""
    search = DuckDuckGoSearchResults ()
    return search . run ( query )


tool_list = [ search ]

prompt = hub . pull ( "hwchase17/xml-agent-convo" )


def convert_intermediate_steps ( intermediate_steps ):
    log = ""
    for action , observation in intermediate_steps :
        log += (
            f"<tool> { action . tool } </tool><tool_input> { action . tool_input } "
            f"</tool_input><observation> { observation } </observation>"
        )
    return log


def convert_tools ( tools ):
    return " n " . join ([ f" { tool . name } : { tool . description } " for tool in tools ])


agent = (
    {
        "input" : lambda x : x [ "input" ],
        "agent_scratchpad" : lambda x : convert_intermediate_steps (
            x [ "intermediate_steps" ]
        ),
    }
    | prompt . partial ( tools = convert_tools ( tool_list ))
    | model . bind ( stop = [ "</tool_input>" , "</final_answer>" ])
    | XMLAgentOutputParser ()
)

agent_executor = AgentExecutor ( agent = agent , tools = tool_list , verbose = True )

if __name__ == "__main__" :
    print ( agent_executor . invoke ({ "input" : "whats the weather in New york?" }))

2. Coba jalankan skrip di terminal untuk mengujinya berfungsi:

♻️

poetry run python langchain_zero_to_hero_agents/src/main.py

3. Agar agen ini berguna, kita dapat menyiapkan API sederhana dengan LangServe:

 from fastapi import FastAPI
from langchain . pydantic_v1 import BaseModel
from langserve import add_routes
from typing import Any

#######################
# LangChain Agent Code
#######################

# ...

######################
# LangServe API Code
######################

class Input ( BaseModel ):
    input : str

class Output ( BaseModel ):
    output : Any

app = FastAPI (
    title = "DuckDuckGo Agent" ,
    version = "1.0" ,
    description = "API for accessing a simple LangChain agent that can query the web with DuckDuckGo." ,
)

add_routes (
    app ,
    agent_executor . with_types ( input_type = Input , output_type = Output ). with_config (
        { "run_name" : "agent" }
    ),
    path = "/agent"
)

if __name__ == "__main__" :
    import uvicorn

    uvicorn . run ( app , host = "localhost" , port = 8000 )

4. Mulai server API LangServe Anda:

♻️

poetry run python langchain_zero_to_hero_agents/src/main.py

5. Kunjungi http://localhost:8000/agent/playground/ untuk mengakses UI sederhana untuk berinteraksi dengan agen Anda.

6. Buat skrip pengujian ( langchain_zero_to_hero_agents/tests/test.py ) untuk bereksperimen mengakses API Anda melalui Python:

 import requests

response = requests . post (
    "http://localhost:8000/agent/invoke" ,
    json = { 'input' : { "input" : "what is the weather in new york" }}
)

print ( response . json ())

7. Jalankan skrip pengujian Anda dan amati keluaran JSON terstruktur:

?

poetry run python langchain_zero_to_hero_agents/tests/test.py

Lisensi

MIT