Building Local LLM for document Q&A using LangChain, Sentence Transformers, and FAISS but I keep getting Type Error while processing the query

Question

I would like to implement something like this - https://github.com/wombyz/gpt4all_langchain_chatbots/blob/main/custom_knowledge_chatbot.py

But since GGML models are slower and inefficient, I have decided to go with a gguf model (https://huggingface.co/TheBloke/airoboros-l2-7B-gpt4-2.0-GGUF) and instead of LlamaCPPEmbeddings (https://huggingface.co/Pi3141/alpaca-native-7B-ggml/commit/397e872bf4c83f4c642317a5bf65ce84a105786e), I am using SentenceTransformer("all-mpnet-base-v2")

This is my code -

from pygpt4all.models.gpt4all import GPT4All
from pprint import pprint
#import streamlit as st
from langchain import PromptTemplate, LLMChain
from langchain.document_loaders import TextLoader
from langchain.embeddings import LlamaCppEmbeddings
from langchain.llms import GPT4All
from langchain.text_splitter import RecursiveCharacterTextSplitter
#from langchain.callbacks.base import CallbackManager
from langchain.callbacks.streaming_stdout import StreamingStdOutCallbackHandler
from langchain.vectorstores.faiss import FAISS
from llama_index import download_loader
from langchain.document_loaders import UnstructuredURLLoader
from langchain.docstore.document import Document
from unstructured.cleaners.core import remove_punctuation,clean,clean_extra_whitespace
import PyPDF2
import re

from langchain.chains import ConversationalRetrievalChain
from pathlib import Path
from typing import List, Tuple
import requests
from bs4 import BeautifulSoup



gpt4all_path = './models/airoboros-l2-7B-gpt4-2.0.Q4_K_M.gguf'
llm = GPT4All(model=gpt4all_path,max_tokens=2048, verbose=True,temp=0.1)

def load_documents() -> List[str]:
 loader = TextLoader('./docs/cleaned_q_and_a.txt')
 documents = loader.load()
 texts = [doc.page_content for doc in documents]
 return texts



def load_meta_data_documents() -> List:
 loader = TextLoader('./docs/cleaned_q_and_a.txt')
 return loader.load()

def create_index(texts: List[str], embeddings_model: SentenceTransformer,chunks:List) -> FAISS:
# Generate embeddings for the texts
 embeddings = embeddings_model.encode(texts, show_progress_bar=True)
 metadatas = [doc.metadata for doc in chunks]
# Prepare text_embeddings as a list of tuples for FAISS.from_embeddings
 text_embeddings = [(text, embedding) for text, embedding in zip(texts, 
 embeddings)]

# Call FAISS.from_embeddings
 search_index = FAISS.from_embeddings(text_embeddings=text_embeddings, 
 embedding=embeddings_model, metadatas=metadatas)

return search_index

docs = load_documents()
metadata = load_meta_data_documents()
vector_store = create_index(docs,embeddings_model,metadata)



# Save Index (use this to save the index for later use)

# Comment the line below after running once successfully (IMPORTANT)

vector_store.save_local("q_and_a_index")

index = FAISS.load_local("./q_and_a_index/", embeddings_model)

qa = ConversationalRetrievalChain.from_llm(llm,index.as_retriever(),max_tokens_limit=500)


chat_history=[]

print("Custom Knowledge ChatBot")

while True:

 query = input("Please enter your question: ")



 if query.lower() == 'exit':

  break



#processed_query = embeddings_model.encode([query])

 result = qa({"question":query,"chat_history":chat_history})



 pattern = r'Helpful Answer:.*'

 match = re.search(pattern, result['answer'], re.DOTALL)



 if match:

# Only display the matched part which is the relevant answer

  print(match.group())

 else:

  print("Answer:", result['answer'])

This is the Error I am getting.

I tried encoding the query

processed_query = embeddings_model.encode([query])
result = qa({"question":processed_query,"chat_history":chat_history})

When I tried encoding the query this is the error I received -

trans_features = {'input_ids': features['input_ids'], 'attention_mask': features['attention_mask']}
if 'token_type_ids' in features:
     rans_features['token_type_ids'] = features['token_type_ids']

IndexError: only integers, slices (`:`), ellipsis (`...`), numpy.newaxis (`None`) and integer or boolean arrays are valid indices

packages in my environment - https://pastebin.com/L4wqnwyi