nn-explorer

NN Explorer is an interactive tool for creating, visualizing, and simulating neural network architectures. It allows you to design models layer-by-layer, connect them visually, automatically calculate shape transformations, and export the architecture as ready-to-use PyTorch code. Ideal for quickly prototyping and understanding model structures without writing code manually.

Features

• Drag-and-Drop: Add layers from a palette to a central whiteboard.
• Visual Connections: Link layers with automatic validation.
• Properties Panel: Edit layer parameters and set the global input shape.
• Shape Propagation: Automatically calculate input/output sizes for each layer.
• Templates: Load predefined architectures (Simple MLP, LeNet-like CNN).
• PyTorch Export: Generate corresponding torch.nn Python code for the designed network.

License

This project is released under the MIT License.