boxx

Why we included this project

Anyone working in scientific Python eventually notices that numpy, torch, and matplotlib handle the heavy lifting but leave the small ergonomic stuff to you. boxx is a compact utility library that covers the gap with short helper functions for interactive debugging and inspection, aimed at people juggling numpy arrays, torch tensors, mxnet arrays, and PIL images. Rather than handwriting a timer or digging through a nested dict entry by entry, you call timeit, tree, show, or loga and get what you need in a line or two. In notebooks or Spyder, g() is the standout: it pushes a function's local variables into the interactive console so you can inspect them on the spot. For teams iterating on vision models or numerical experiments, that shorter loop between writing code and understanding what it produced is the real payoff, with no new workflow to learn.