When training scripts fail, it's hard to tell what happens because the output is written to a log file without any indication of where it goes. This has caused confusion for people: https://github.com/pcyin/tranX/issues/4
I think it'd be better to either:
Both pipe the output to log files, and display it on screen.
Provide an indication that output will be written to a log file.
(or both).
When training scripts fail, it's hard to tell what happens because the output is written to a log file without any indication of where it goes. This has caused confusion for people: https://github.com/pcyin/tranX/issues/4
I think it'd be better to either:
I can send a PR for either of these pending https://github.com/pcyin/tranX/pull/12