app.py

# !pip install langchain
# !pip install langchain_community
# !pip install langchain_text_splitters
# !pip install langchain-google-genai
# !pip install gradio
# !pip install openai
# !pip install pypdf
# !pip install chromadb
# !pip install tiktoken
# !pip install python-dotenv

import initialize
from langchain_openai import ChatOpenAI
from langchain.memory import ConversationBufferMemory
from langchain.chains import ConversationalRetrievalChain
from langchain.chains import VectorDBQA
from langchain.llms import OpenAI


import gradio as gr
import os
import requests

import sys
sys.path.append('../..')

# For Google Colab
'''
from google.colab import userdata
OPENAI_API_KEY = userdata.get('OPENAI_API_KEY')
hf_token = userdata.get('hf_token')
GEMINI_API_KEY = userdata.get('GEMINI_API_KEY')
# For Desktop
from dotenv import load_dotenv, find_dotenv
_ = load_dotenv(find_dotenv()) # Read local .env file
OPENAI_API_KEY = os.environ['OPENAI_API_KEY']
hf_token = os.environ['hf_token']
GEMINI_API_KEY = os.environ['GEMINI_API_KEY']
'''

# For Hugging Face
OPENAI_API_KEY = os.environ.get('OPENAI_API_KEY')
hf_token = os.environ.get('hf_token')
GEMINI_API_KEY = os.environ.get('GEMINI_API_KEY')
fs_token = os.environ.get('fs_token')

llm_name = "gpt-3.5-turbo"

vectordb = initialize.initialize()


#-------------------------------------------



from langchain import HuggingFacePipeline, PromptTemplate, LLMChain, RetrievalQA
from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline
import torch

quantization_config = {
    "load_in_4bit": True,
    "bnb_4bit_compute_dtype": torch.float16,
    "bnb_4bit_quant_type": "nf4",
    "bnb_4bit_use_double_quant": True,
}

llm = HuggingFacePipeline(pipeline=pipeline)
model_id = "mistralai/Mistral-7B-Instruct-v0.1"
model_4bit = AutoModelForCausalLM.from_pretrained(
    model_id, device="cuda", quantization_config=quantization_config
)

tokenizer = AutoTokenizer.from_pretrained(model_id)

pipeline = pipeline(
    "text-generation",
    model=model_4bit,
    tokenizer=tokenizer,
    use_cache=True,
    device=0,  # '0' is for GPU, 'cpu' for CPU
    max_length=500,
    do_sample=True,
    top_k=5,
    num_return_sequences=1,
    eos_token_id=tokenizer.eos_token_id,
    pad_token_id=tokenizer.eos_token_id,
)

template = """[INST] You are a helpful, respectful and honest assistant. Answer exactly in few words from the context
Answer the question below from the context below:
{context}
{question} [/INST]
"""




def chat_query(retrieverQA, text_query):
    
    retrieverQA = RetrievalQA.from_chain_type(llm=llm, chain_type="retrieval", retriever=vectordb.as_retriever(), verbose=True)
    result = retrieverQA.run(text_query)
    return result



    
#-------------------------------------------

    
# def chat_query(question, history):
    
#     llm = ChatOpenAI(model=llm_name, temperature=0.1, api_key = OPENAI_API_KEY)

#     # Conversation Retrival Chain with Memory
#     memory = ConversationBufferMemory(memory_key="chat_history", return_messages=True)
#     retriever=vectordb.as_retriever()
#     qa = ConversationalRetrievalChain.from_llm(llm, retriever=retriever, memory=memory)

#     # Replace input() with question variable for Gradio
#     result = qa({"question": question})
#     return result['answer']

#     # Chatbot only answers based on Documents
#     # qa = VectorDBQA.from_chain_type(llm=OpenAI(openai_api_key = OPENAI_API_KEY, ), chain_type="stuff", vectorstore=vectordb)
#     # result = qa.run(question)
#     # return result

    


# logo_path = os.path.join(os.getcwd(), "Logo.png")

iface = gr.ChatInterface(
    fn=chat_query,
    title=" -----:  Conversational BOT for Model-TS  :----- ",
    description="""-- Welcome to the Language Model trained on `Model Technical Specifications` of Engineering Dept.\n\n
                   -- The Model tries to answer the Query based on TS, GTR & Specific Requirements. \n\n
                   -- For precise reply, please input `Specific Keywords` in your Query. \n\n """,
    concurrency_limit = None,
    examples = ["What should be the GIB height outside the GIS hall ?", "Tell me about coupling device for Outdoor Mounting." ],
    #theme=gr.themes.Glass(),
)

iface.launch(share=True, debug=True)

# What should be the GIB height outside the GIS hall ?