Question about sub-validation set

anonymous47823493 / EagleEye

(ECCV'2020 Oral)EagleEye: Fast Sub-net Evaluation for Efficient Neural Network Pruning

304 stars 68 forks source link

Question about sub-validation set #35

Closed junhyukso closed 3 years ago

junhyukso commented 3 years ago

Thanks for your great research and code. I'm just curious about why you use sub-validation set as small amount of training data, instead of same amout of validation data. is it unlikely that the model will be over-fitting to the training data and will be measured with high accuracy?

Bowenwu1 commented 3 years ago

Hi We can not access the validation data during searching and training, since this will cause the data leak issue. Actually, for such a big dataset, the over-fitting problem is very weak. We have verified this by experiments.