I started with tensorflow back in 2017 or so, then in my current job I am touching a lot of pytorch, and as such I say a pox on both their houses. But keras is

I started with tensorflow back in 2017 or so, then in my current job I am touching a lot of pytorch, and as such I say a pox on both their houses. But keras is OK and I greatly appreciate that you can (sometimes) serialize everything to hdf5.