How well can Large Language Models Learn?

About

This project aims to assess the capabilities of Large Language Models (LLMs) in a learning environment such as classrooms. The LLMs will take in lecture slides from a particular module and attempt to answer exam like questions based on the topics in the lecture slides provided.

Setting up Requirements

1. Install the required dependencies using this command.

pip install -r requirements.txt

Source File Breakdown

1. ChatBot interface

   • Run the agents.py file to launch the interface
   • To use the interface, upload a pdf file containing highlitable text to the documents section and type in a question regarding the document uploaded before hitting submit
   • Wait for the program to process your file before providing an answer

2. Experiments

   • The experimental data was created using modelData.py, promptData.py, and trainingData.py.
   • After creating the data, run experiments.py to run the evaluations on the projects you've created
   • The evaluation functions and LLM is stored in evalFunctions.py

3. Context

   • evalSlides, evalTranscripts and lectures contain the chroma which the model uses to perform retrieval augmented generation.

4. Downloading the LLMs

   • To download the LLMs, run modelDownload.py to download the model .gguf files in this directory.

Cloning the Repository

Use the code below to clone the GitHub repository

git clone https://github.com/japhialoo/LLMs.git