ModelForge is a visual, drag-and-drop interface to build and train deep learning models — no code, no fuss. Design neural networks like stacking Lego blocks and see them come to life with real-time training and visualization. Whether you're a beginner or a prototyping pro, ModelForge turns your ideas into working models faster than you can say "backpropagation."
![ModelForge Banner]("./model-forge/public/modelForge (1).png")
- ⚙️ Sequential Model Builder – Drag and drop layers to visually create a deep learning model.
- 🖼️ Intuitive Interface – React-powered canvas with draggable components and real-time model structure preview.
- 📊 Live Training Feedback – Watch your model train with live loss/accuracy curves.
- 🗃️ Layer Customization – Tweak hyperparameters and settings in a dedicated pane.
- 🧠 Built-in Layer Library – Dense, Conv2D, Dropout, Flatten, and more (with icons!).
- 🧵 Session Memory – Redis-based backend to store and manage model configs pre-training.
- 📦 One-click Training – Train your model via backend endpoints (Express.js + PyTorch/TensorFlow support planned).
| Frontend | Backend | Storage | Extras |
|---|---|---|---|
| Next.js (TypeScript) | Express.js | Redis | Swiper.js, TailwindCSS |
| Framer Motion | REST API | File System (optional) | TensorBoard (planned) |
- Node.js
- Redis
- Python (optional for training module)
git clone https://github.com/yourusername/modelforge.git
cd modelforge
# Frontend
cd frontend
npm install
npm run dev