search

Levigty / AimCLR

This is an official PyTorch implementation of "Contrastive Learning from Extremely Augmented Skeleton Sequences for Self-supervised Action Recognition" in AAAI2022.
MIT License
63 stars 14 forks source link

About KNN evaluation and finetune settings #5

Closed SLime1998 closed 2 years ago

SLime1998 commented 2 years ago

Hi, Thank you for sharing your work! You used KNN evaluation method in your paper, but I did not find the related implementation details in the article, nor did I find the corresponding code. Would please provide some more details of the KNN evaluation? And as to the finetune setting, did you adjust the weight decay when doing finetune? Looking forward to your reply

Levigty commented 2 years ago

Hi. For the KNN evaluation, we use the knn_monitor like other contrastive learning methods. For finetune setting, we set weight decay to 1e-4, learning rate 0.01 or 0.001.

SLime1998 commented 2 years ago

Thanks a lot for your help!