ReflectionAnyLLM

Click here to watch the video sorry for the quality

This project demonstrates a basic chain-of-thought interaction with any LLM (Large Language Model), whether local or hosted remotely, provided it supports the OpenAI-compatible API. You can easily swap out the endpoint and API key to work with various providers such as:

LMSTUDIO
Groq
OpenRouter
OpenAI API

Getting Started

To run this project, you'll need a local web server such as:

XAMPP (Windows)
MAMP (Mac)
LAMP stack (Linux)

While MySQL is not required, you will need a web server capable of running PHP with the curl extension. Simply running php -S won't work out of the box because the project depends on the curl extension.

Steps to Set Up:

Download and install the required web server (XAMPP, MAMP, or LAMP).
Copy the index.html and chat.php files to your public web directory:
- XAMPP: htdocs
- MAMP: htdocs
- LAMP: /var/www
Ensure the PHP curl extension is enabled on your server.

Accessing the Application:

Local Setup: If you are using a local web server, navigate to http://localhost/ in your browser.
Online Hosting: If deploying on an online server, navigate to your domain and the folder where the files are uploaded.

File Structure:

├── index.html   # The main HTML file for the front-end
└── chat.php     # PHP script to handle the backend logic

Disclaimer

This project is purely for fun, and I usually don’t share this type of work, but I was asked to do so on Reddit. If you want to fork it and improve upon it, I’ll make sure to accept your commits.

A few things to note:

The chain-of-thought implementation is only 2 steps because my M1 Mac is slow, and I can’t afford the wait. You can expand it as much as you want by calling the function repeatedly.
Important: The front-end and back-end code is not secure and should not be used for production environments.
I uploaded this in a hurry. If you'd like me to continue working on it, let me know—but I don’t believe it's worth much at this point.