Welcome to the first steps of creating a PLEXOS AI

First Step: Choosing A Large Language Model(LLM)

In this script, the llama 3.1 LLM was used with the help of Ollama to run the LLM locally Go to https://ollama.com/ and install ollama Choose an LLM, run the following command in the terminan/cmd in order to download the LLM

ollama pull llama3.1

To test, run the following and type a prompt:

ollama run llama3.1 

Second Step: Gather Data

Create a folder/directory named Extracted_Data, place all the files you want to train your LLM on in this case, we will be using well documented PLEXOS API code.

Third Step: RAG Script

Install the following dependencies

pip install llama-index
pip install llama-index-llms-ollama
pip install llama_index.embeddings.huggingface

note if you are using windows you have to do this extra step:

pip uninstall torch
pip install torch==2.2

The RAG.py has three functions

construct_index()
    this function takes in a dirctory path, loads the files, then generates an index
    and saves it in the model Folder/Directory

load_index()
    this function takes the model Folder/Directory from the storage returns it as an index

save_to_excel():
    Saves query results to an Excel file. If the file exists, it prompts whether to append or overwrite.

Fourth Step: Run and create training data

The first time using the program you must call the construct_index() to create an index. After it has been created you can call the load_index() instead.

You can type a prompt press enter and a answer will be generated. It will ask if you want to save it by typing yes/y. After you create enough data, we can upload this data for it used to fine tune a model.