app.py

from llama_index import GPTVectorStoreIndex, SimpleDirectoryReader, ServiceContext, set_global_service_context, load_index_from_storage, StorageContext, PromptHelper
from llama_index.llms import OpenAI
import gradio as gr
import sys
import os

try:
    from Config import openai_key
    os.environ["OPENAI_API_KEY"] = openai_key
except:
    pass

"""
Code adopted from Beebom article: "How to Train an AI Chatbot With Custom Knowledge Base Using ChatGPT API" by Arjun Sha
https://beebom.com/how-train-ai-chatbot-custom-knowledge-base-chatgpt-api/
"""

max_input_size = 4096
num_outputs = 512
chunk_size_limit = 600

prompt_helper = PromptHelper(context_window=max_input_size, num_output=num_outputs, chunk_overlap_ratio=0.1, chunk_size_limit=chunk_size_limit)

llm = OpenAI(model="gpt-3.5-turbo", temperature=0.5, max_tokens=num_outputs)

service_context = ServiceContext.from_defaults(llm=llm, prompt_helper=prompt_helper)
set_global_service_context(service_context)

def retrieve_index(index_path):
    storage_context = StorageContext.from_defaults(persist_dir=index_path)
    index = load_index_from_storage(storage_context)
    return index

def chatbot(input_text):
    response = QE.query(input_text)
    response_stream = ""
    for r in response.response_gen:
        response_stream += r
        yield response_stream

if __name__ == "__main__":
    iface = gr.Interface(fn=chatbot,
                         inputs=gr.components.Textbox(lines=7, label="Enter your text"),
                         outputs="text",
                         title="AI Chatbot for the Doing What Works Library")

    index = retrieve_index("dww_vectors")
    QE = index.as_query_engine(streaming=True)
    iface.launch(share=False)