Importing Libraries:
from langchain_openai import ChatOpenAI
from langchain.chains import ConversationalRetrievalChain
Define a function make_chain that creates a chain of langchain components.
def make_chain():
Now write further code inside this function.
Initialize LLM that will be used to answer queries. We will use gpt-3.5-turbo model, which is default model used in ChatGPT:
model = ChatOpenAI(
model_name="gpt-3.5-turbo",
temperature=0.0,
verbose=True
)
Initialize prompt using the prompt template we defined:
prompt = get_prompt()
Initialize Retriever. We will initialize our Chroma vector store as retriever.
vector_store = get_chroma_client()
retriever = vector_store.as_retriever(search_type="mmr", verbose=True)
Create the chain of these components. We will use ConversationalRetrievalChain from langchain which is generally used for building chatbots:
chain = ConversationalRetrievalChain.from_llm(
model,
retriever=retriever,
return_source_documents=True,
combine_docs_chain_kwargs=dict(prompt=prompt),
verbose=True,
rephrase_question=False,
)
return chain
Above return_source_documents when set to True, returns the documents retrieved from the vector store along with the answer. rephrase_question when set to True, rephrase the user question according to the whole conversation or we can say chat_history. For this demonstration, we will not be using chat_history, that's why we have set it to False.
Note - Having trouble with the assessment engine? Follow the steps listed here
Loading comments...