realtime-rag-pipeline / generator /extract_attributes.py
Gourisankar Padihary
Logging enabled
afa7a1b
raw
history blame
737 Bytes
import json
from generator.create_prompt import create_prompt
from generator.initialize_llm import initialize_llm
# Initialize the LLM
llm = initialize_llm()
# Function to extract attributes
def extract_attributes(question, relevant_docs, response):
# Format documents into a string by accessing the `page_content` attribute of each Document
formatted_documents = "\n".join([f"Doc {i+1}: {doc.page_content}" for i, doc in enumerate(relevant_docs)])
#print(f'Formated documents: {formatted_documents}')
attribute_prompt = create_prompt(formatted_documents, question, response)
# Instead of using BaseMessage, pass the formatted prompt directly to invoke
result = llm.invoke(attribute_prompt)
return result