search

gidariss / FewShotWithoutForgetting

MIT License
519 stars 110 forks source link

Obtained test accuracy higher than reported for miniImageNet #9

Closed vidit09 closed 6 years ago

vidit09 commented 6 years ago

Hello , thank you for the implementation of your work. For the first stage, I used the ResNetLikeCosineClassifier configuration and changed the batch size to 6 (due to memory limitation of my gpu). Later with Attention based 5way 1-shot training(untouched configuration), I get 56.89+/-0.83 on novel class and 78.70 on base class for the test set. What I'm wondering is, if there is any additional step involved ?

vidit09 commented 6 years ago

Ok, just observed the updated test accuracies.