You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I don't love this solution. We put a LOT of effort into cross testing the code for many python versions and pytorch lightning versions, so that we wouldn't be like those tf 1.x people.
If you download the package and have lightning 2.0, changing one line is an inconvenience. If you download the package and have lightnign 1.x, downporting it is definitely scary business because you don't know if there will be a regression.
I propose:
a) Put back all the testing for all the older versions
b) Do a try except block to first try lightning and the pl
c) Since we don't use lightning, and also hacking torchsynth to work with lightning is janky (I can show you my code and what I'm doing), we consider actually making a torchsynth-lightning package and getting rid of the lightning dep, with nicer lightning wrapping.
The text was updated successfully, but these errors were encountered:
88bda58
I don't love this solution. We put a LOT of effort into cross testing the code for many python versions and pytorch lightning versions, so that we wouldn't be like those tf 1.x people.
If you download the package and have lightning 2.0, changing one line is an inconvenience. If you download the package and have lightnign 1.x, downporting it is definitely scary business because you don't know if there will be a regression.
I propose:
a) Put back all the testing for all the older versions
b) Do a try except block to first try lightning and the pl
c) Since we don't use lightning, and also hacking torchsynth to work with lightning is janky (I can show you my code and what I'm doing), we consider actually making a torchsynth-lightning package and getting rid of the lightning dep, with nicer lightning wrapping.
The text was updated successfully, but these errors were encountered: