Not an issue - livepython fyi

[johndpope]

I know notebooks are most popular for ml - but consider there's some that prefer to drop in to command line and run python code. In this case - there's this library that is quite amazing that could potentially be plugged in. https://github.com/agermanidis/livepython

[parrt]

Yes I definitely use the command line with this library. Thanks for the pointer in case people are interested.

[johndpope]

It maybe 10 years from now - but I wish the pytorch team would take this livepython and add it to their roadmap where you could run the training or inference but have(maybe configure/plugin) a digestable simplistic overview / latex - maths and sub disection of the of what's going on under the hood with some kind of ai explanatation of linear algebra. Or said another way - read the pdf papare - show the pytorch code - explain it / bridge the understanding

• ask the user questions to see if they understand thing. It's not necessarily - going to be a compiler - or even give / verify the correct results - but given this branch of ai (relevant papers / formulas that underpin it) - help to explain it for newbies.

this code is running lines 50 - 100 - and it's working on this equation in the paper - based off the work of bengio et el (highlight picture of equation).