Custom Language Model using Python3, Openai

I was also following this article, and found that much of the code was out of date and the features had been deprecated. After a lot of trial and error and reviewing the documentation I now have a working custom LLM (on MacOS - the software and packages may differ on other OS).

The following software and packages are required:

1. Install Python from https://www.python.org/downloads/
2. python3 -m pip install -U pip
3. pip3 install openai
4. pip3 install llama-index
5. pip3 install PyPDF2
6. pip3 install gradio

• Importing is now from llama_index.core
• GPTSimpleVectorIndex is deprecated in favour of GPTVectorStoreIndex
• LLMPredictor and ServiceContext are deprecated in favour of Settings, StorageContext and load_index_from_storage
• index.save_to_disk is deprecated in favour of
• index.storage_context.persist()
• index.query is deprecated in favour of query_engine

My updated and functioning code now reads as follows:

from llama_index.core import SimpleDirectoryReader, GPTVectorStoreIndex, Settings, StorageContext, load_index_from_storage
from llama_index.llms.openai import OpenAI
import gradio as gr
import os

os.environ["OPENAI_API_KEY"] = '--enter API Key---'

def construct_index(directory_path):
    num_outputs = 512

    Settings.llm = OpenAI(temperature=0.7, model_name="gpt-3.5-turbo", max_tokens=num_outputs)

    docs = SimpleDirectoryReader(directory_path).load_data()

    index = GPTVectorStoreIndex.from_documents(docs)

    index.storage_context.persist("storage")

    return index

def chatbot(input_text):
    storage_context = StorageContext.from_defaults(persist_dir="storage")
    index = load_index_from_storage(storage_context)
    query_engine = index.as_query_engine()
    response = query_engine.query(input_text)
    return response.response

iface = gr.Interface(fn=chatbot, inputs=gr.Textbox(lines=7, label="Enter your text"), outputs="text", title="Custom LLM Bot")

iface.launch(share=True)