Skip to content

README.md

Latest commit

 

History

History
12 lines (9 loc) · 708 Bytes

README.md

File metadata and controls

12 lines (9 loc) · 708 Bytes
Raw

DesignTorch

DesignTorch is a visual model builder for PyTorch. The goal is to let users create neural network architectures by adding nodes for layers, activation functions, and other components, and connecting them visually to define the model's data flow.

Roadmap

  • Connect the React frontend to a Python backend using PyTorch.
  • Implement functional nodes for layers, activations, and other common components.
  • Allow users to connect nodes and define data flow dynamically.
  • Enable exporting of visual models to executable PyTorch nn.Module code.
  • Add support for advanced layers (Conv, RNN, etc.) and custom parameters.
  • Improve UI/UX with better node interactions and visual feedback.