Humanize-AI-Server

Humanize AI is a text rewriting and paraphrasing service that leverages advanced T5-based models from Hugging Face. The goal is to generate text that preserves meaning while making it sound more natural and less detectable by AI detectors.

Note: This project is still under active development and not all features are finalized.

Note: Visit the frontend site with the link https://humanize-ai-frontend.vercel.app

---

Project Structure

.
├── api
│   ├── index.py              # Main Flask server
│   ├── download_model.py     # Script to download and save T5 model locally
│   ├── models.txt            # List of model names or references
│   ├── requirements.txt      # Python dependencies
│   ├── vercel.json           # Vercel deployment configuration
│   ├── makefile              # (Optional) Make commands for building/deployment
│   └── ...
├── .env                      # Environment variables (HF_TOKEN, etc.)
└── .gitignore

• index.py: Contains your Flask application with endpoints (/, /hello, /paraphrase).
• download_model.py: Downloads the t5-small model and tokenizer to a local folder.
• models.txt: Lists models or references to be used or downloaded.
• requirements.txt: Lists Python dependencies (Flask, requests, python-dotenv, etc.).
• vercel.json: Configuration file if you choose to deploy on Vercel.

---

Features

• Paraphrasing
  Leverages a T5-based paraphrasing model (e.g., humarin/chatgpt_paraphraser_on_T5_base or Vamsi/T5_Paraphrase_Paws) via the Hugging Face Inference API.

• Custom Models
  You can switch out the model URL in index.py (the HF_API_URL) to any other Hugging Face model endpoint.

• Environment Variables
  Uses a .env file to store sensitive tokens (e.g., HF_TOKEN).

• Logging
  Uses Python’s logging module to debug and monitor API behavior.

---

Prerequisites

1. Python 3.11+
2. Hugging Face Account & Token
   • Generate a token from your Hugging Face profile settings.
   • Create a .env file and add HF_TOKEN=your_huggingface_token_here.

---

Installation & Setup

1. Clone the repository

   git clone https://github.com/ADEMOLA200/Humanize-AI-Server.git
   cd Humanize-AI-Server/api

2. Create a virtual environment (recommended)

   python -m venv .venv
   source .venv/bin/activate  # On Windows: .venv\Scripts\activate

3. Install Python dependencies

   pip install -r requirements.txt

4. Set up environment variables

   • Create a .env file in the project root or api folder with:

     HF_TOKEN=your_huggingface_token_here

   • Make sure to keep this file secret or add it to your .gitignore.

5. (Optional) Download a local T5 model

   • If you want to download and store a local T5 model (e.g., t5-small), run:

     python download_model.py

   • This will create a folder ./models/t5-small with the tokenizer and model weights.
   • Note: The current Flask code in index.py uses the Hugging Face Inference API, so downloading the model locally is optional.

---

Running the Flask App

From inside the api directory, run:

python index.py

The Flask server will start on http://127.0.0.1:5001 by default (or another port if you set the PORT environment variable).

Endpoints

1. Health Check

   • GET /
   • Example: curl http://127.0.0.1:5001/
   • Response:

     Server operational!
     

2. Hello Test

   • GET /hello
   • Example: curl http://127.0.0.1:5001/hello
   • Response:

     Hello, Vercel deployment works!
     

3. Paraphrase

   • POST /paraphrase
   • Request Body (JSON):

     {
       "text": "Your text to paraphrase goes here."
     }

   • Response (JSON):

     {
       "paraphrased": "The generated paraphrased text.",
       "success": true
     }

---

Deploying on Vercel

This project includes a vercel.json configuration for deploying a Python server.

1. Install the Vercel CLI.
2. Run vercel in the project root and follow the prompts.

Note: Python/Flask on Vercel typically uses the Serverless Functions approach. Make sure your vercel.json and project structure match Vercel’s Python requirements.

---

Contributing

1. Fork the repository.
2. Create a new branch for your feature or bug fix.
3. Commit your changes with clear messages.
4. Submit a Pull Request to the main branch.

---

License

This project is licensed under the MIT License.

---

Contact

For any questions or suggestions, please open an issue in the repository or contact [email protected].

---

Important Note

This project is still under active development. The paraphrasing and “humanization” features may not be perfect, and the underlying model can be changed or improved at any time. Feel free to experiment with different Hugging Face models for better performance.