Issues with hub

[edogrigqv2]

• from connectors are slow
• 2nd time dataset is accessed error is happening if it is not fully uploaded - try to reproducable, need to come up with more redundant logic @edogrigqv2
• upload dataset in default mode gives an error - @davidbuniat
• sequences are not working properly, .compute() is in the middle. - in progress
• error for specific chunk sizes
• slow data loader from_pytorch
• text is working properly
• test coverage needs to be increased
• we need to add more datasets that are trainable
• many users ask conda support (especially from omdena)
• benchmarking with CI/CD
• documention issues, simplify, explain better - @kristinagrig06 @mikayelh We have to optimize the documentation layout, so more users go and read it.
• create colab tutorials for uploading, accessing, training model. Using transform as well.