Baseline ++ validation wrongly reported in paper?

[DianeBouchacourt]

In the paper, it seems you perform validation on novel classes for Baseline and Baseline++ too, as per the sentence
" To reduce the risk of over-fitting, we use the validation set to select the epoch or episode
with the best accuracy for all methods, including baseline and baseline++"

But looking at the code, it seems you actually do not validation and set acc=-1 for Baseline (see

CloserLookFewShot/train.py

Line 39 in e03aca8

if acc > max_acc : #for baseline and baseline++, we don't use validation in default and we let acc = -1, but we allow options to validate with DB index

and

CloserLookFewShot/methods/baselinetrain.py

Line 80 in e03aca8

#Emperically, this only works for CUB dataset but not for miniImagenet dataset

)

Could you explain?