salesforce/PCL
PyTorch code for "Prototypical Contrastive Learning of Unsupervised Representations"
PythonMIT
Issues
- 2
About Loss of InfoNCE and Cluter_results
#22 opened by lsyysl9711 - 1
- 0
Apply to CIFAR-10
#25 opened by kroegern1 - 12
Question about eq(9) in your paper.
#17 opened by randydkx - 8
about run-kmeans
#6 opened by Hiiragi0107 - 0
Question about the equation 12 in the paper
#23 opened by Hugo-cell111 - 0
Does eval_loader need to load all the data?
#21 opened by MarahHalawa - 2
Single GPU
#19 opened by Hepta-Col - 1
Last prototype never sampled as negative
#20 opened by liux16wfu - 1
Questioning about the cluster method?
#18 opened by Odysseussambition - 2
- 1
Imbalanced datasets
#15 opened by millanp95 - 2
question about concentration around a prototype
#16 opened by vmmm123 - 2
The settings for KNN classifier?
#13 opened by Hzzone - 1
- 3
- 3
about cifar10
#4 opened by Hiiragi0107 - 8
- 2
- 3
About the cluster size analysis and the uniform distribution assumption for each cluster
#9 opened by ChangLee0903 - 2
inconsistency between the InfoNCE's negative samples and its corresponding clusters
#8 opened by ChangLee0903 - 5
- 1
About NCE
#3 opened by MajorDavidZhang - 2
The training time of PCL?
#2 opened by zzw-zwzhang - 2
Questions about concentration estimation
#1 opened by Knarf04