search

MCG-NJU / VideoMAE

[NeurIPS 2022 Spotlight] VideoMAE: Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Training
https://arxiv.org/abs/2203.12602
Other
1.35k stars 136 forks source link

why do not you use [CLS] token? #78

Open LinB203 opened 1 year ago

LinB203 commented 1 year ago

wonder work! but I'm confuse about that why use pooling token instead of [CLS] token? Is performance become worse? or anything I miss?