Very good that Pytorch emerged as a serious contender to TF. While TF still provided more production grade tools (TFX, TensorRT, TF serving), Pytorch continue to evolve and hope soon we have a more complete ecosystem

I really like JAX as well: https://github.com/google/jax. It's younger than PyTorch and TF, but feels cleaner and more expressive. It has a very nice autodiff implementation (based on https://github.com/HIPS/autograd) and performance is comparable to TF in my experience.