lucidrains/pixel-level-contrastive-learning
Implementation of Pixel-level Contrastive Learning, proposed in the paper "Propagate Yourself", in Pytorch
PythonMIT
Issues
- 0
Apply PixPro to resnet50 with FPN
#19 opened by simonebonato - 3
- 5
positive pairs
#14 opened by jancylee - 1
Loss values are not reducing
#18 opened by NamburiSrinath - 0
Could you report performance?
#16 opened by jianlong-yuan - 6
Alpha - Combination of PixPro loss and PixCL
#4 opened by WilhelmT - 0
How to train PixelCL on multi-gpu?
#15 opened by ALEX13679173326 - 11
- 2
Same size models -> wildly different RAM usage
#13 opened by makarr - 3
Data Augumentation not same with BYOL?
#10 opened by guanfuchen - 13
O(n^4) memory usage in PPM
#8 opened by neonbjb - 1
about the pos-neg pair selection
#7 opened by 1144181135 - 1
about distributed
#6 opened by datuizhuang - 6
Help needed
#1 opened by lucidrains - 9
- 4
Does using RandomResizedCrop/HorizontalFlip in the augmentaion pipeline break the pixel alignment?
#2 opened by cjrd