I think it was just a matter of time till TF would get superseded by PyTorch. The only reason we kept TF on prod is the java api which allowed us to quickly loa

I think it was just a matter of time till TF would get superseded by PyTorch. The only reason we kept TF on prod is the java api which allowed us to quickly load and serve TF models. I spent so many sleepless nights trying to port Torch model to TF back in the days and make it work the same as Lua based prototype. Whole TF "experience" made us switch to plain Python services model throwing away all the boilerplate Scala/Java code for TF. It doesn't happen often in tech that a better engineered product gets more traction and recognition eventually and I am glad that PyTorch did.