You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hi, thanks for the question. Note that the TOP1 that's being reported is the Nearest-Neighbor (NN) evaluation, which is typically lower than the linear-probing performance (83.4). You can find the NN performance in the bottom part of the table, where we obtained 77.6.
Hi, I run the cifar.sh
in the end, the results output like this:
this result is much less than 83.4 reported in the paper. Can someone tell me why? Thank you very much!
The text was updated successfully, but these errors were encountered: