quick one, got access to a range of TPUs in the next weeks, wanted to run there fast.ai code that is prepared for cuda GPU with .cuda(); to run that code on TPUs, is that possible? if so what changes do we need to make in the code so that it runs on a TPU? or maybe is not possible at the moment?
thank you for any tips 
1 Like
You would need to get pytorch working with the tpus, they are apparently working on that:
2 Likes
thank you j.laute, got you, so no use for now I guess
I think it might be possible, i cant try unfortunately
1 Like