ChatGPTの限界を超えろ LangChain【概要編】

現状のChatGPT の限界

LangChainとは

Welcome to LangChain — 🦜🔗 LangChain 0.0.152

https://python.langchain.com/en/latest/index.html

どのようなモジュールがあるのか？

• Models
• LLMs、Chat Models、Text Embeddings Models

• Prompts
• Prompt Templates、Chat Prompt Template、Example Selectors、Output Parsers

• Memory

• Indexes
• Document Loaders、Text Splitters、Vectorstores、Retrievers

• Chains

• Agents
• Tools、Agents、Toolkits、Agent Executors

Models

Models — 🦜🔗 LangChain 0.0.156

https://python.langchain.com/en/latest/modules/models.html

from langchain.llms import OpenAI
llm = OpenAI(model_name="text-ada-001", n=2, best_of=2)

llm("Tell me a joke")

'\n\nWhy did the chicken cross the road?\n\nTo get to the other side.'

Prompts

Prompts — 🦜🔗 LangChain 0.0.152

https://python.langchain.com/en/latest/modules/prompts.html

from langchain import PromptTemplate


template = """
I want you to act as a naming consultant for new companies.
What is a good name for a company that makes {product}?
"""

prompt = PromptTemplate(
    input_variables=["product"],
    template=template,
)

prompt.format(product="colorful socks")
# -> I want you to act as a naming consultant for new companies.
# -> What is a good name for a company that makes colorful socks?

Memory

Memory — 🦜🔗 LangChain 0.0.152

https://python.langchain.com/en/latest/modules/memory.html

from langchain.memory import ChatMessageHistory

history = ChatMessageHistory()

history.add_user_message("hi!")
history.add_ai_message("whats up?")

history.messages

[HumanMessage(content='hi!', additional_kwargs={}),
 AIMessage(content='whats up?', additional_kwargs={})]

Indexes

Indexes — 🦜🔗 LangChain 0.0.152

https://python.langchain.com/en/latest/modules/indexes.html

1. テキストデータの読み込み

2. テキストデータの分割

3. 分割されたチャンクのベクトル化

4. ベクトル化したデータの保存

5. 保存したデータの取得

テキストデータの読み込み: Document loader
テキストデータの分割: Text Splitters
ベクトル化データの保存: Vectorstores
保存したデータの取得: Retrievers

from langchain.text_splitter import CharacterTextSplitter

documents = loader.load()
text_splitter = CharacterTextSplitter(chunk_size=1000, chunk_overlap=0)
texts = text_splitter.split_documents(documents)

Chains

Chains — 🦜🔗 LangChain 0.0.157

https://python.langchain.com/en/latest/modules/chains.html

from langchain.prompts import PromptTemplate
from langchain.llms import OpenAI

llm = OpenAI(temperature=0.9)
prompt = PromptTemplate(
    input_variables=["product"],
    template="What is a good name for a company that makes {product}?",
)

from langchain.chains import LLMChain
chain = LLMChain(llm=llm, prompt=prompt)

# Run the chain only specifying the input variable.
print(chain.run("colorful socks"))

from langchain.chains import SimpleSequentialChain
overall_chain = SimpleSequentialChain(chains=[chain, chain_two], verbose=True)

# Run the chain specifying only the input variable for the first chain.
catchphrase = overall_chain.run("colorful socks")
print(catchphrase)

Agents

Agents — 🦜🔗 LangChain 0.0.157

https://python.langchain.com/en/latest/modules/agents.html#

from langchain.tools import Tool
from langchain.utilities import GoogleSearchAPIWrapper

search = GoogleSearchAPIWrapper()

tool = Tool(
    name = "Google Search",
    description="Search Google for recent results.",
    func=search.run
)

tool.run("Obama's first name?")

# Construct the agent. We will use the default agent type here
agent = initialize_agent(tools, llm, 
agent=AgentType.ZERO_SHOT_REACT_DESCRIPTION, verbose=True)

まとめ