langgraph/examples/rag/langgraph_rag_agent_llama3_local.ipynb

Local RAG agent with LLaMA3

We'll combine ideas from paper RAG papers into a RAG agent:

• Routing: Adaptive RAG (paper). Route questions to different retrieval approaches
• Fallback: Corrective RAG (paper). Fallback to web search if docs are not relevant to query
• Self-correction: Self-RAG (paper). Fix answers w/ hallucinations or don’t address question

Local models

Embedding

GPT4All Embeddings:

pip install langchain-nomic

LLM

Use Ollama and llama3:

ollama pull llama3

Prompt -

https://llama.meta.com/docs/model-cards-and-prompt-formats/meta-llama-3/

In [ ]:

%%capture --no-stderr
%pip install -U langchain-nomic langchain_community tiktoken langchainhub chromadb langchain langgraph tavily-python nomic[local] langchain-text-splitters

Tracing (optional)

In [ ]:

import os

os.environ["LANGCHAIN_TRACING_V2"] = "true"
os.environ["LANGCHAIN_ENDPOINT"] = "https://api.smith.langchain.com"
os.environ["LANGCHAIN_API_KEY"] = "<your-api-key>"

In [4]:

### LLM

local_llm = "llama3"

In [5]:

### Index

from langchain_community.document_loaders import WebBaseLoader
from langchain_community.vectorstores import Chroma
from langchain_nomic.embeddings import NomicEmbeddings
from langchain_text_splitters import RecursiveCharacterTextSplitter

urls = [
    "https://lilianweng.github.io/posts/2023-06-23-agent/",
    "https://lilianweng.github.io/posts/2023-03-15-prompt-engineering/",
    "https://lilianweng.github.io/posts/2023-10-25-adv-attack-llm/",
]

docs = [WebBaseLoader(url).load() for url in urls]
docs_list = [item for sublist in docs for item in sublist]

text_splitter = RecursiveCharacterTextSplitter.from_tiktoken_encoder(
    chunk_size=250, chunk_overlap=0
)
doc_splits = text_splitter.split_documents(docs_list)

# Add to vectorDB
vectorstore = Chroma.from_documents(
    documents=doc_splits,
    collection_name="rag-chroma",
    embedding=NomicEmbeddings(model="nomic-embed-text-v1.5", inference_mode="local"),
)
retriever = vectorstore.as_retriever()

In [4]:

### Retrieval Grader

from langchain_community.chat_models import ChatOllama
from langchain_core.output_parsers import JsonOutputParser
from langchain_core.prompts import PromptTemplate

# LLM
llm = ChatOllama(model=local_llm, format="json", temperature=0)

prompt = PromptTemplate(
    template="""<|begin_of_text|><|start_header_id|>system<|end_header_id|> You are a grader assessing relevance 
    of a retrieved document to a user question. If the document contains keywords related to the user question, 
    grade it as relevant. It does not need to be a stringent test. The goal is to filter out erroneous retrievals. \n
    Give a binary score 'yes' or 'no' score to indicate whether the document is relevant to the question. \n
    Provide the binary score as a JSON with a single key 'score' and no premable or explanation.
     <|eot_id|><|start_header_id|>user<|end_header_id|>
    Here is the retrieved document: \n\n {document} \n\n
    Here is the user question: {question} \n <|eot_id|><|start_header_id|>assistant<|end_header_id|>
    """,
    input_variables=["question", "document"],
)

retrieval_grader = prompt | llm | JsonOutputParser()
question = "agent memory"
docs = retriever.invoke(question)
doc_txt = docs[1].page_content
print(retrieval_grader.invoke({"question": question, "document": doc_txt}))

{'score': 'yes'}

In [5]:

### Generate

from langchain_core.output_parsers import StrOutputParser
from langchain_core.prompts import PromptTemplate

# Prompt
prompt = PromptTemplate(
    template="""<|begin_of_text|><|start_header_id|>system<|end_header_id|> You are an assistant for question-answering tasks. 
    Use the following pieces of retrieved context to answer the question. If you don't know the answer, just say that you don't know. 
    Use three sentences maximum and keep the answer concise <|eot_id|><|start_header_id|>user<|end_header_id|>
    Question: {question} 
    Context: {context} 
    Answer: <|eot_id|><|start_header_id|>assistant<|end_header_id|>""",
    input_variables=["question", "document"],
)

llm = ChatOllama(model=local_llm, temperature=0)


# Post-processing
def format_docs(docs):
    return "\n\n".join(doc.page_content for doc in docs)


# Chain
rag_chain = prompt | llm | StrOutputParser()

# Run
question = "agent memory"
docs = retriever.invoke(question)
generation = rag_chain.invoke({"context": docs, "question": question})
print(generation)

The context mentions that the memory component of an LLM-powered autonomous agent system includes a long-term memory module (external database) that records a comprehensive list of agents' experience in natural language, referred to as "memory stream". This suggests that the agent has some form of memory or recall mechanism.

In [10]:

### Hallucination Grader

# LLM
llm = ChatOllama(model=local_llm, format="json", temperature=0)

# Prompt
prompt = PromptTemplate(
    template=""" <|begin_of_text|><|start_header_id|>system<|end_header_id|> You are a grader assessing whether 
    an answer is grounded in / supported by a set of facts. Give a binary 'yes' or 'no' score to indicate 
    whether the answer is grounded in / supported by a set of facts. Provide the binary score as a JSON with a 
    single key 'score' and no preamble or explanation. <|eot_id|><|start_header_id|>user<|end_header_id|>
    Here are the facts:
    \n ------- \n
    {documents} 
    \n ------- \n
    Here is the answer: {generation}  <|eot_id|><|start_header_id|>assistant<|end_header_id|>""",
    input_variables=["generation", "documents"],
)

hallucination_grader = prompt | llm | JsonOutputParser()
hallucination_grader.invoke({"documents": docs, "generation": generation})

Out [10]:

{'score': 'yes'}

In [11]:

### Answer Grader

# LLM
llm = ChatOllama(model=local_llm, format="json", temperature=0)

# Prompt
prompt = PromptTemplate(
    template="""<|begin_of_text|><|start_header_id|>system<|end_header_id|> You are a grader assessing whether an 
    answer is useful to resolve a question. Give a binary score 'yes' or 'no' to indicate whether the answer is 
    useful to resolve a question. Provide the binary score as a JSON with a single key 'score' and no preamble or explanation.
     <|eot_id|><|start_header_id|>user<|end_header_id|> Here is the answer:
    \n ------- \n
    {generation} 
    \n ------- \n
    Here is the question: {question} <|eot_id|><|start_header_id|>assistant<|end_header_id|>""",
    input_variables=["generation", "question"],
)

answer_grader = prompt | llm | JsonOutputParser()
answer_grader.invoke({"question": question, "generation": generation})

Out [11]:

{'score': 'yes'}

In [15]:

### Router

from langchain_community.chat_models import ChatOllama
from langchain_core.output_parsers import JsonOutputParser
from langchain_core.prompts import PromptTemplate

# LLM
llm = ChatOllama(model=local_llm, format="json", temperature=0)

prompt = PromptTemplate(
    template="""<|begin_of_text|><|start_header_id|>system<|end_header_id|> You are an expert at routing a 
    user question to a vectorstore or web search. Use the vectorstore for questions on LLM  agents, 
    prompt engineering, and adversarial attacks. You do not need to be stringent with the keywords 
    in the question related to these topics. Otherwise, use web-search. Give a binary choice 'web_search' 
    or 'vectorstore' based on the question. Return the a JSON with a single key 'datasource' and 
    no premable or explanation. Question to route: {question} <|eot_id|><|start_header_id|>assistant<|end_header_id|>""",
    input_variables=["question"],
)

question_router = prompt | llm | JsonOutputParser()
question = "llm agent memory"
docs = retriever.get_relevant_documents(question)
doc_txt = docs[1].page_content
print(question_router.invoke({"question": question}))

{'datasource': 'vectorstore'}

In [6]:

### Search
from langchain_community.tools.tavily_search import TavilySearchResults

web_search_tool = TavilySearchResults(k=3)

We'll implement these as a control flow in LangGraph.

In [16]:

from pprint import pprint
from typing import List

from langchain_core.documents import Document
from typing_extensions import TypedDict

from langgraph.graph import END, StateGraph, START

### State


class GraphState(TypedDict):
    """
    Represents the state of our graph.

    Attributes:
        question: question
        generation: LLM generation
        web_search: whether to add search
        documents: list of documents
    """

    question: str
    generation: str
    web_search: str
    documents: List[str]


### Nodes


def retrieve(state):
    """
    Retrieve documents from vectorstore

    Args:
        state (dict): The current graph state

    Returns:
        state (dict): New key added to state, documents, that contains retrieved documents
    """
    print("---RETRIEVE---")
    question = state["question"]

    # Retrieval
    documents = retriever.invoke(question)
    return {"documents": documents, "question": question}


def generate(state):
    """
    Generate answer using RAG on retrieved documents

    Args:
        state (dict): The current graph state

    Returns:
        state (dict): New key added to state, generation, that contains LLM generation
    """
    print("---GENERATE---")
    question = state["question"]
    documents = state["documents"]

    # RAG generation
    generation = rag_chain.invoke({"context": documents, "question": question})
    return {"documents": documents, "question": question, "generation": generation}


def grade_documents(state):
    """
    Determines whether the retrieved documents are relevant to the question
    If any document is not relevant, we will set a flag to run web search

    Args:
        state (dict): The current graph state

    Returns:
        state (dict): Filtered out irrelevant documents and updated web_search state
    """

    print("---CHECK DOCUMENT RELEVANCE TO QUESTION---")
    question = state["question"]
    documents = state["documents"]

    # Score each doc
    filtered_docs = []
    web_search = "No"
    for d in documents:
        score = retrieval_grader.invoke(
            {"question": question, "document": d.page_content}
        )
        grade = score["score"]
        # Document relevant
        if grade.lower() == "yes":
            print("---GRADE: DOCUMENT RELEVANT---")
            filtered_docs.append(d)
        # Document not relevant
        else:
            print("---GRADE: DOCUMENT NOT RELEVANT---")
            # We do not include the document in filtered_docs
            # We set a flag to indicate that we want to run web search
            web_search = "Yes"
            continue
    return {"documents": filtered_docs, "question": question, "web_search": web_search}


def web_search(state):
    """
    Web search based based on the question

    Args:
        state (dict): The current graph state

    Returns:
        state (dict): Appended web results to documents
    """

    print("---WEB SEARCH---")
    question = state["question"]
    documents = state["documents"]

    # Web search
    docs = web_search_tool.invoke({"query": question})
    web_results = "\n".join([d["content"] for d in docs])
    web_results = Document(page_content=web_results)
    if documents is not None:
        documents.append(web_results)
    else:
        documents = [web_results]
    return {"documents": documents, "question": question}


### Conditional edge


def route_question(state):
    """
    Route question to web search or RAG.

    Args:
        state (dict): The current graph state

    Returns:
        str: Next node to call
    """

    print("---ROUTE QUESTION---")
    question = state["question"]
    print(question)
    source = question_router.invoke({"question": question})
    print(source)
    print(source["datasource"])
    if source["datasource"] == "web_search":
        print("---ROUTE QUESTION TO WEB SEARCH---")
        return "websearch"
    elif source["datasource"] == "vectorstore":
        print("---ROUTE QUESTION TO RAG---")
        return "vectorstore"


def decide_to_generate(state):
    """
    Determines whether to generate an answer, or add web search

    Args:
        state (dict): The current graph state

    Returns:
        str: Binary decision for next node to call
    """

    print("---ASSESS GRADED DOCUMENTS---")
    state["question"]
    web_search = state["web_search"]
    state["documents"]

    if web_search == "Yes":
        # All documents have been filtered check_relevance
        # We will re-generate a new query
        print(
            "---DECISION: ALL DOCUMENTS ARE NOT RELEVANT TO QUESTION, INCLUDE WEB SEARCH---"
        )
        return "websearch"
    else:
        # We have relevant documents, so generate answer
        print("---DECISION: GENERATE---")
        return "generate"


### Conditional edge


def grade_generation_v_documents_and_question(state):
    """
    Determines whether the generation is grounded in the document and answers question.

    Args:
        state (dict): The current graph state

    Returns:
        str: Decision for next node to call
    """

    print("---CHECK HALLUCINATIONS---")
    question = state["question"]
    documents = state["documents"]
    generation = state["generation"]

    score = hallucination_grader.invoke(
        {"documents": documents, "generation": generation}
    )
    grade = score["score"]

    # Check hallucination
    if grade == "yes":
        print("---DECISION: GENERATION IS GROUNDED IN DOCUMENTS---")
        # Check question-answering
        print("---GRADE GENERATION vs QUESTION---")
        score = answer_grader.invoke({"question": question, "generation": generation})
        grade = score["score"]
        if grade == "yes":
            print("---DECISION: GENERATION ADDRESSES QUESTION---")
            return "useful"
        else:
            print("---DECISION: GENERATION DOES NOT ADDRESS QUESTION---")
            return "not useful"
    else:
        pprint("---DECISION: GENERATION IS NOT GROUNDED IN DOCUMENTS, RE-TRY---")
        return "not supported"


workflow = StateGraph(GraphState)

# Define the nodes
workflow.add_node("websearch", web_search)  # web search
workflow.add_node("retrieve", retrieve)  # retrieve
workflow.add_node("grade_documents", grade_documents)  # grade documents
workflow.add_node("generate", generate)  # generatae

Graph Build

In [17]:

# Build graph
workflow.add_conditional_edges(
    START,
    route_question,
    {
        "websearch": "websearch",
        "vectorstore": "retrieve",
    },
)

workflow.add_edge("retrieve", "grade_documents")
workflow.add_conditional_edges(
    "grade_documents",
    decide_to_generate,
    {
        "websearch": "websearch",
        "generate": "generate",
    },
)
workflow.add_edge("websearch", "generate")
workflow.add_conditional_edges(
    "generate",
    grade_generation_v_documents_and_question,
    {
        "not supported": "generate",
        "useful": END,
        "not useful": "websearch",
    },
)

In [18]:

# Compile
app = workflow.compile()

# Test

inputs = {"question": "What are the types of agent memory?"}
for output in app.stream(inputs):
    for key, value in output.items():
        pprint(f"Finished running: {key}:")
pprint(value["generation"])

---ROUTE QUESTION---
What are the types of agent memory?
{'datasource': 'vectorstore'}
vectorstore
---ROUTE QUESTION TO RAG---
---RETRIEVE---
'Finished running: retrieve:'
---CHECK DOCUMENT RELEVANCE TO QUESTION---
---GRADE: DOCUMENT RELEVANT---
---GRADE: DOCUMENT RELEVANT---
---GRADE: DOCUMENT RELEVANT---
---GRADE: DOCUMENT RELEVANT---
---ASSESS GRADED DOCUMENTS---
---DECISION: GENERATE---
'Finished running: grade_documents:'
---GENERATE---
---CHECK HALLUCINATIONS---
---DECISION: GENERATION IS GROUNDED IN DOCUMENTS---
---GRADE GENERATION vs QUESTION---
---DECISION: GENERATION ADDRESSES QUESTION---
'Finished running: generate:'
('According to the provided context, there are several types of memory '
 'mentioned:\n'
 '\n'
 '1. Sensory Memory: This is the earliest stage of memory, providing the '
 'ability to retain impressions of sensory information (visual, auditory, etc) '
 'after the original stimuli have ended.\n'
 '2. Maximum Inner Product Search (MIPS): This is a long-term memory module '
 "that records a comprehensive list of agents' experience in natural "
 'language.\n'
 '\n'
 'These are the types of agent memory mentioned in the context.')

Trace:

https://smith.langchain.com/public/8d449b67-6bc4-4ecf-9153-759cd21df24f/r

In [21]:

from pprint import pprint

# Compile
app = workflow.compile()
inputs = {"question": "Who are the Bears expected to draft first in the NFL draft?"}
for output in app.stream(inputs):
    for key, value in output.items():
        pprint(f"Finished running: {key}:")
pprint(value["generation"])

---ROUTE QUESTION---
Who are the Bears expected to draft first in the NFL draft?
{'datasource': 'web_search'}
web_search
---ROUTE QUESTION TO WEB SEARCH---
---WEB SEARCH---
'Finished running: websearch:'
---GENERATE---
---CHECK HALLUCINATIONS---
---DECISION: GENERATION IS GROUNDED IN DOCUMENTS---
---GRADE GENERATION vs QUESTION---
---DECISION: GENERATION ADDRESSES QUESTION---
'Finished running: generate:'
('The Bears are expected to draft USC star and 2022 Heisman Trophy winner '
 'Caleb Williams with the No. 1 overall pick.')

Trace:

https://smith.langchain.com/public/c785f9c0-f519-4a38-ad5a-febb59a2139c/r

In [ ]: