The loss is still around 0.7, but I can’t train it for longer lest the weights will turn into NaNs.
Putes to bed
Mtars in Shoy, https://youtube.
home was are not a wasteof user (and it stop chalbutta wasteof on ToPT)
boringn to think of this bill
ratio maos is a RTO PUnt
love my banner (cfussil Pls "Orng slo
Al the ringy be
3ying my minutes
ELUole pronones
os
I think it’s better if the 512 units version was trained from scratch.