Ask HN: What is TensorFlow still good for now?
I heard that in the ML community most people work now in Pytorch.
My understanding of TF was that it was more than an ML framework and more of a general purpose high performance, GPU framework.
What to you think? Doesn't a large majority of Pytorch work still require tensorflow libraries? I might be wrong but when I was finetuning a model using QLora and Pytorch in AWS Sagemaker, a lot of Tensorflow libraries were needed