Auto-PyTorch
Auto-PyTorch copied to clipboard
Many errors are squeezed by try except
Errors during the training are completely squeezed by try except and they do not come up on the terminal.
This aspect is quite user-unfriendly, so it should be fixed soon.
It happens when we use ExecuteTaFuncWithQueue
(or pynisher
).
So, these errors are handled well as they are added to the additional_info for the configuration. What we can do is also log it to terminal by pushing error logs to std:err
Do you think we can close this issue?