TFX equivalent in pytorch


I’m trying o learn how to use PyTorch in production. There are a lot of tools out there and I’m confused a little bit. I see most of the books and tutorials on Tensorflow. It has TFX for example. Is torchscript the equivalent to TFX? What pipeline and tools do you suggest to use for PyTorch in production? What is the drawback of using sth like flask+docker+kubernetes to put the trained model in production?

Any link or reference with a complete comparison would be appreciated.

Thank you

I found this and it seems it is similar to TFX. One more question? Is it again (using torchserve) slower than using a saved model using torchscript in C++ or JS?

There also is TorchX

1 Like