Abracudabra: Convert Tensors, Arrays and DataFrames Between CPU and CUDA

Abracudabra is a Python package designed to simplify the conversion of dataframes, arrays, and tensors between CPU and CUDA devices.

It provides a consistent API to move data between various libraries (NumPy, CuPy, pandas, cuDF, and PyTorch) by handling the necessary device conversions behind the scenes.

Install the package with:

pip install abracudabra

Contents:

• Usage Guide
  • Supported Data Types and Libraries
  • Device Management
  • Conversion Functions
  • Device Management
  • Library Selection Helpers
• Development
  • Setting up the Environment
  • Running Tests
• Changelog
  • v0.1.0 (2025-03-09)
• Credits

API Reference

• abracudabra
  • Submodules
  • Classes
  • Functions
  • Package Contents