Hi,
I find it pretty scary that Pytorch does not have loaders for SQL or aws S3 or Parquet formats, does this mean no one is actually using it for large scale data sets?
Is any one using Pytorch for Kafka streams or other streaming systems?
Theoretically we should be able to build these components on our own, so can we write our own operating system, and develop our own programming language. But should we not expect to see a certain level of maturity from a tool that is competing with Tensorflow?
Regards,
Gourav Sengupta