3 ms·

How does this compare on embedded systems for performance? For example PyTorch vs tinygrad, or Darknet vs tinygrad?