Hello,
when trying to interrupt the training of my models in jupyternb, nothing happens, due to the multiprocessing in the background. My question is: is there any handy method that allows to propagate the keyboardinterrupt to all employed background processes, such that restarting the notebook is not required, anymore? Just restarting wouldnt even be so bad, but, in my case (maybe also for others), the kernel dies permanently, when I interrupt training by restarting the notebook. So I have to restart the entire notebook application every time. Do this ten times in half an hour, and you will be writing this post here 
Thanks!
Best, JZ