Inspect your runnables

Once you create a runnable with LCEL, you may often want to inspect it to get a better sense for what is going on. This notebook covers some methods for doing so.

First, let’s create an example LCEL. We will create one that does retrieval

!pip install langchain openai faiss-cpu tiktoken

from operator import itemgetter

from langchain.prompts import ChatPromptTemplate
from langchain.vectorstores import FAISS
from langchain_core.output_parsers import StrOutputParser
from langchain_core.runnables import RunnableLambda, RunnablePassthrough
from langchain_openai import ChatOpenAI, OpenAIEmbeddings

vectorstore = FAISS.from_texts(
    ["harrison worked at kensho"], embedding=OpenAIEmbeddings()
)
retriever = vectorstore.as_retriever()

template = """Answer the question based only on the following context:
{context}

Question: {question}
"""
prompt = ChatPromptTemplate.from_template(template)

model = ChatOpenAI()

chain = (
    {"context": retriever, "question": RunnablePassthrough()}
    | prompt
    | model
    | StrOutputParser()
)

Get a graph

You can get a graph of the runnable

chain.get_graph()

Graph(nodes={'7308e6063c6d40818c5a0cc1cc7444f2': Node(id='7308e6063c6d40818c5a0cc1cc7444f2', data=<class 'pydantic.main.RunnableParallel<context,question>Input'>), '292bbd8021d44ec3a31fbe724d9002c1': Node(id='292bbd8021d44ec3a31fbe724d9002c1', data=<class 'pydantic.main.RunnableParallel<context,question>Output'>), '9212f219cf05488f95229c56ea02b192': Node(id='9212f219cf05488f95229c56ea02b192', data=VectorStoreRetriever(tags=['FAISS', 'OpenAIEmbeddings'], vectorstore=<langchain_community.vectorstores.faiss.FAISS object at 0x117334f70>)), 'c7a8e65fa5cf44b99dbe7d1d6e36886f': Node(id='c7a8e65fa5cf44b99dbe7d1d6e36886f', data=RunnablePassthrough()), '818b9bfd40a341008373d5b9f9d0784b': Node(id='818b9bfd40a341008373d5b9f9d0784b', data=ChatPromptTemplate(input_variables=['context', 'question'], messages=[HumanMessagePromptTemplate(prompt=PromptTemplate(input_variables=['context', 'question'], template='Answer the question based only on the following context:\n{context}\n\nQuestion: {question}\n'))])), 'b9f1d3ddfa6b4334a16ea439df22b11e': Node(id='b9f1d3ddfa6b4334a16ea439df22b11e', data=ChatOpenAI(client=<class 'openai.api_resources.chat_completion.ChatCompletion'>, openai_api_key='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', openai_proxy='')), '2bf84f6355c44731848345ca7d0f8ab9': Node(id='2bf84f6355c44731848345ca7d0f8ab9', data=StrOutputParser()), '1aeb2da5da5a43bb8771d3f338a473a2': Node(id='1aeb2da5da5a43bb8771d3f338a473a2', data=<class 'pydantic.main.StrOutputParserOutput'>)}, edges=[Edge(source='7308e6063c6d40818c5a0cc1cc7444f2', target='9212f219cf05488f95229c56ea02b192'), Edge(source='9212f219cf05488f95229c56ea02b192', target='292bbd8021d44ec3a31fbe724d9002c1'), Edge(source='7308e6063c6d40818c5a0cc1cc7444f2', target='c7a8e65fa5cf44b99dbe7d1d6e36886f'), Edge(source='c7a8e65fa5cf44b99dbe7d1d6e36886f', target='292bbd8021d44ec3a31fbe724d9002c1'), Edge(source='292bbd8021d44ec3a31fbe724d9002c1', target='818b9bfd40a341008373d5b9f9d0784b'), Edge(source='818b9bfd40a341008373d5b9f9d0784b', target='b9f1d3ddfa6b4334a16ea439df22b11e'), Edge(source='2bf84f6355c44731848345ca7d0f8ab9', target='1aeb2da5da5a43bb8771d3f338a473a2'), Edge(source='b9f1d3ddfa6b4334a16ea439df22b11e', target='2bf84f6355c44731848345ca7d0f8ab9')])

Print a graph

While that is not super legible, you can print it to get a display that’s easier to understand

chain.get_graph().print_ascii()

           +---------------------------------+         
           | Parallel<context,question>Input |         
           +---------------------------------+         
                    **               **                
                 ***                   ***             
               **                         **           
+----------------------+              +-------------+  
| VectorStoreRetriever |              | Passthrough |  
+----------------------+              +-------------+  
                    **               **                
                      ***         ***                  
                         **     **                     
           +----------------------------------+        
           | Parallel<context,question>Output |        
           +----------------------------------+        
                             *                         
                             *                         
                             *                         
                  +--------------------+               
                  | ChatPromptTemplate |               
                  +--------------------+               
                             *                         
                             *                         
                             *                         
                      +------------+                   
                      | ChatOpenAI |                   
                      +------------+                   
                             *                         
                             *                         
                             *                         
                   +-----------------+                 
                   | StrOutputParser |                 
                   +-----------------+                 
                             *                         
                             *                         
                             *                         
                +-----------------------+              
                | StrOutputParserOutput |              
                +-----------------------+              

Get the prompts

An important part of every chain is the prompts that are used. You can get the graphs present in the chain:

chain.get_prompts()

[ChatPromptTemplate(input_variables=['context', 'question'], messages=[HumanMessagePromptTemplate(prompt=PromptTemplate(input_variables=['context', 'question'], template='Answer the question based only on the following context:\n{context}\n\nQuestion: {question}\n'))])]