LTH14 / mar

PyTorch implementation of MAR+DiffLoss https://arxiv.org/abs/2406.11838

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

About masking ratio

RohollahHS opened this issue · comments

The masking ratio is randomly sampled from [0.7, 1]. Did you conduct an ablation study on the effect of different ranges for the masking ratio?

Thanks.

Similar to MAE we sweep the masking ratio and find 0.7 is the best. For different tasks/data, the optimal masking ratio is typically different.