The Versus Project

Project Overview

I love @versus, and that's why I fine-tuned an LLM on versus' captions and made an end-to-end AI content tool with it. Low-cost deployment for projects featuring local LLM's is essentially non-existent, so I decided against deployment.

Tech Stack

• Backend: Python, Flask, Flask-JWT-Extended, Flask-Bcrypt
• Frontend: React (with Vite), Tailwind CSS, shadcn/ui
• Database: PostgreSQL (managed with Docker Compose)
• AI & APIs: Google Gemini, Instagrapi, Newspaper3k
• Containerization: Docker

Local Development Setup

Prerequisites

• Python>=3.10
• Node.js and npm
• Docker and Docker Compose
• Kaggle account

Step 1: Set Up the Backend

1. Create and activate a Python virtual environment:

   python3 -m venv .venv
   source .venv/bin/activate

2. Install Python dependencies:

   pip install -r requirements.txt

3. Create an Instagram session:

• Tip: Login to the same instagram account on your browser as well, and browse around normally, every once in a while. This will keep your session from being tagged as "abnormal".

  python3 insta_login.py

Step 2: Set Up the Frontend

1. Navigate to the frontend directory:

   cd frontend

2. Install Node.js dependencies:

   npm install

3. Return to the root directory:

   cd ..

Step 3: Configure Environment Variables

• Create a .env file in the root of the project and add the following variables:

  # --- Gemini API Key ---
  GEMINI=<your-gemini-api-key>
  
  # --- Instagram Credentials ---
  INSTA_USERNAME=<your_instagram_username>
  INSTA_PASSWORD=<your_instagram_password>
  
  # --- Kaggle Inference Server URL ---
  KAGGLE_INFERENCE_URL=https://your-ngrok-url.ngrok-free.app
  
  # --- PostgreSQL Database Credentials ---
  POSTGRES_USER=versusdb
  POSTGRES_PASSWORD=<your_secure_password>
  POSTGRES_DB=versus_db
  POSTGRES_PORT=5433
  
  # --- RSS Feed URL ---
  RSS_FEED=https://www.theguardian.com/football/rss
  
  # --- JWT Secret Key ---
  JWT_SECRET_KEY=<a_very_strong_and_secret_key>

Step 4: Set Up the Kaggle Notebook

• Kaggle Config:

  T4/P100 GPU
  Enable Internet
  
  # -- Secretes --
  NGROK_AUTHTOKEN=<your-ngrok-auth-token>
  NGROK_DOMAIN=https://your-ngrok-url.ngrok-free.app
  HF_TOKEN=<your-huggingface-access-token>

• Sample Inference Server:

  # ---------------------------------- Cell #1 ----------------------------------
  !pip uninstall -y torch torchvision torchaudio numpy
  
  
  # ---------------------------------- Cell #2 ----------------------------------
  !pip install numpy==1.26.4
  
  
  # ---------------------------------- Cell #3 ----------------------------------
  !pip install torch==2.1.0 torchvision==0.16.0 torchaudio==2.1.0 --index-url https://download.pytorch.org/whl/cu121
  
  
  # ---------------------------------- Cell #4 ----------------------------------
  !pip install --upgrade transformers peft accelerate bitsandbytes datasets trl flask pyngrok
  
  
  # ---------------------------------- Cell #5 ----------------------------------
  # Refer inference_example.py
  # Make sure to "Restart and Clear All Cell Outputs" after installing all the dependencies.

Step 5: Start the Services

1. Start the PostgreSQL Database:

• Open a terminal in the project root and run:

  docker compose up -d

2. Start the Backend Server:

• In a new terminal, make sure your Python virtual environment is activated, and run the Flask application:

  python app_ig.py

3. Start the Frontend Development Server:

• In a third terminal, navigate to the frontend directory.
• Run the Vite development server:

  npm run dev