Training saturates early?

[ajoshi80]

I'm using the same hypers but seeing this for my training curve. Why would this happen? Looks like the LR is too high but your curve with the same lr seems fine.

[ajoshi80]

This seems to be an issue with the python 3 data loading where everything is mapped to OOV

[jivatneet]

@ajoshi80 were you able to resolve this?

[malteos]

@jivatneet @ajoshi80

The issue is that since Python 3 all tokens are byte variables and therefore they are recognized as OOV. Changing everything back to string representations seems to work.