Skip to content

INTERACT-LLM/Interact-LLM

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Interact-LLM: Inference Experiments

This repository contains code for experiments exploring how large language models (LLMs) perform as cognitive tutors for language learning. All experiments are connected to the INTERACT-LLM project. See also the Research section.

Note: The code is currently only intended for internal use and is not production-ready.

🔗 Read about the INTERACT-LLM project (in Danish),

🚀 Overview

The src folder contains:

Folder Description
interact_llm Inference engine and terminal-based chatbot
scripts Experimental setups using the backend linked to specific publications, including alignment-drift created for Almasi & Kristensen-McLachlan (2025).

📝 Research

The Interact-LLM repository uses version tags linked to specific papers. Each paper has its own folder in /src/scripts, and the version tag includes the corresponding paper name:

Paper Interact-LLM Version Suffix scripts folder (Inference code) Paper Repository (Analysis code)
Almasi & Kristensen-McLachlan (2025) vX.X.X-alignment-drift /src/scripts/alignment_drift INTERACT-LLM/alignment-drift-llms

🛠️ Technical Requirements

The code was run on Python 3.12.3 on both a macOS (15.3.1) and Ubuntu system (24.04). The project also requires:

Tool Installation
make Installed via Homebrew
uv Installed through this project's makefile (see Usage)

⚙️ Usage

Setup project

To install uv on macOS/Linux and set up a virtual environment with the required Python dependencies, run in the terminal:

make setup

Add HF Token

If you are interested in running gated models such as Llama-3.1-8B-Instruct, you will need to pass a Hugging Face token with read access.

Create a file called hf_token.txt with the token and place it in the tokens folder. The file will not be pushed to GitHub.

Run Chatbot in Terminal App

To experiment with interacting with a chatbot (prompted to act as a Spanish language tutor currently), run in the terminal

uv run python -m interact_llm 

⚠️ IMPORTANT: This is very early development. Functionality is limited.

Reproduce Experiments

Refer to the individual READMEs in scripts e.g., alignment-drift.

About

Code for experiments requiring LLM inference

Topics

Resources

License

Stars

Watchers

Forks

Packages

No packages published