llm-chatbot

Contributing

• Feel free to create your version of this repo by creating a new branch with your name and what you plan to do, e.g. haoze/multi-files

• Don't directly make any changes on the main branch, this is the "stable" branch for deployment.

• One more thing, you may need an openai gpt api key to get the response from them. If you only work on UI and front end, you might not need this.

• Please add your contributions with meaningful change message to the change log below

• Need help setting up: contact me at [email protected]

Steps to Run Application

1. Start the Backend Server: Navigate to the backend directory and run the command to launch the backend server.

2. Launch the Frontend: Open a new terminal window, go to the saas-frontend directory, and start the frontend application.

3. Access the Application: Once both servers are running, open your web browser and visit http://localhost:3000 to access the application.

4. Access different chatbot streams: Navigate to http://localhost:3000/virtualTA for the TA Chatbot and http://localhost:3000/courseEvaluation for the Course Evaluation Chatbot

Running the FrontEnd

1. Navigate to the folder saas-frontend

cd saas-frontend

2. Add the following .env File under the saas-frontend folder (optional: customize for your school and IP adress)

# .env file for NC State University
# @author Sanjit Verma

#VCL: default set to localhost
REACT_APP_API_URL=http://127.0.0.1:8000 

# NAVBAR
REACT_APP_NAVBAR_HEADER="Virtual TA"
REACT_APP_FEEDBACK_FORM_URL="https://forms.gle/5swJdyyfSdQxGww69"

# LANDING PAGE
REACT_APP_MODALBODYTEXT="Virtual TA answers your basic questions, giving you more support for your learning journey."
REACT_APP_BACKGROUND_IMAGE_URL="/_MAH0122.jpg"
REACT_APP_LFOOTER=""
REACT_APP_RFOOTER=""
REACT_APP_FRONT_LOGO="/ncstate-brick-4x1-blk-max.png"

# CHAT PAGE
REACT_APP_CHAT_WELCOME="Hello, "
REACT_APP_CHAT_WELCOME_TEXT="How can I help you today?"


# NAVBAR CE
REACT_APP_CENAVBAR_HEADER="Course Evaluation LLM"

# CHAT PAGE CE
REACT_APP_CECHAT_WELCOME="Hello, "
REACT_APP_CECHAT_WELCOME_TEXT="Please upload your course evaluation."

# COURSE ENVIRONMENT
REACT_APP_COURSE_NUMBER="COURSE_NUMBER_HERE"
REACT_APP_COURSE_TITLE="COURSE_TITLE_HERE"

3. Install the packages for the frontend.

npm install

4. Start the server.

npm start

Running the Backend

Contact Sanjit Verma ([email protected]) to be added to the MongoDB database for your username and password

1. Navigate to the folder saas-backend
2. Install the packages from requirements.txt.

pip install -r requirements.txt

(Optional) The requirments.txt might be out of date. Install the package prompted from the terminal.

3. Add the following .env File under the saas-backend folder

MONGODB_URI="mongodb+srv://<username>:<password>@chatbot.3zncvhn.mongodb.net/?retryWrites=true&w=majority&appName=chatbot"
MONGODB_DATABASE=chatbot
MONGODB_VECTORS= MONGODB_VECTORS
MONGODB_VECTOR_INDEX= MONGODB_VECTOR_INDEX
MONGODB_USERS= MONGODB_USERS
MONGODB_SUGGESTIONS= MONGODB_SUGGESTIONS

MONGODB_VECTORS_COURSEEVAL = MONGODB_VECTORS_COURSEEVALUATION
MONGODB_VECTOR_INDEX_COURSEEVAL = MONGODB_VECTOR_INDEX_COURSEEVALUATION

MONGODB_VECTORS_COURSEEVALUATION_DOCS = MONGODB_VECTORS_COURSEEVALUATION_DOCS
MONGODB_VECTOR_INDEX_TEMPUSER_DOC = vector_index

MONGODB_TEMPUSER = MONGODB_TEMPUSER
MONGODB_VECTOR_INDEX_TEMPUSER = MONGODB_VECTOR_INDEX_TEMPUSER

MONGODB_WHITELIST_USERS = MONGODB_WHITELIST_USERS
MONGODB_ACCESSCODES = MONGODB_ACCESSCODES

MONGODB_VECTORS_COURSEWEBSITE= MONGODB_VECTORS_COURSEWEBSITE
MONGODB_VECTOR_INDEX_WEBSITE= Course_Website

OPENAI_API_KEY= <YOU API KEY HERE>
LANGFUSE_PUBLIC_KEY = <YOU API KEY HERE>
LANGFUSE_SECRET_KEY = <YOU API KEY HERE>
LANGFUSE_HOST = "https://us.cloud.langfuse.com"
MAIL_USERNAME = "[email protected]"
MAIL_PASSWORD = "uoop ghkv iaom cfna" 

4. Start the server.

python app.py

Setting up LangFuse

We track our LLM performance using Langfuse.

What it is: Langfuse is a platform designed for logging, monitoring, and observability of large language models (LLMs). It is not a testing framework. Instead, Langfuse provides tools for tracking the performance, usage, and behavior of LLMs in real-time. Use it to optimize models perfomance, cost and debug issues.

Please contact Nirmal Joji ([email protected]) to be added to the project on LangFuse and get the API Keys or if you need assistance with LangFuse

Authenticating Users

Contact Sanjit Verma ([email protected]) for assistance. Users can be authenticated through the Virtual TA Admin Panel

To access the panel:

1. Navigate to the folder auth
2. Add the following .env File under the auth folder

MONGODB_URI="mongodb+srv://<username>:<password>@chatbot.3zncvhn.mongodb.net/?retryWrites=true&w=majority&appName=chatbot"
MONGODB_DATABASE=chatbot
MONGODB_WHITELIST_USERS = MONGODB_WHITELIST_USERS
MONGODB_ACCESSCODES = MONGODB_ACCESSCODES

3. Run the tachatbotWhiteList.py app, install any required python libraries
4. Either add the user directly via email through the Authenticated Users tab or generate an access code to send to the user (no email required)