HobbitLong/SupContrast issues and pull requests

#51 - Where is the negative pairs loss?

Issue - State: open - Opened by XiXiRuPan almost 4 years ago - 1 comment

#50 - loss is nan

Issue - State: open - Opened by passionvery almost 4 years ago - 3 comments

#49 - running CE loss does not work

Issue - State: open - Opened by rabeehkarimimahabadi almost 4 years ago - 1 comment

#48 - Missing l2-normalization in the embedding function of ResNet(dim=2048)

Issue - State: closed - Opened by bastian1209 almost 4 years ago - 2 comments

#47 - Create LICENSE

Pull Request - State: closed - Opened by HobbitLong almost 4 years ago

#46 - About contrast mode

Issue - State: open - Opened by Hiiragi0107 almost 4 years ago - 1 comment

#45 - Loss saturates

Issue - State: open - Opened by tarunn2799 almost 4 years ago - 1 comment

#44 - How to understand the figure 4

Issue - State: open - Opened by slchenchn almost 4 years ago - 1 comment

#43 - questions about effect of number of positives

Issue - State: open - Opened by slchenchn almost 4 years ago - 2 comments

#42 - Supervised contrastive loss saturation from the beginning on custom data

Issue - State: open - Opened by zaid478 almost 4 years ago - 4 comments

#41 - No L2 Normalization for the output of the encoder network?

Issue - State: open - Opened by killawhale2 almost 4 years ago - 9 comments

#41 - No L2 Normalization for the output of the encoder network?

Issue - State: open - Opened by killawhale2 almost 4 years ago - 9 comments

#40 - Pretrained models

Issue - State: open - Opened by MyLtYkRiTiK almost 4 years ago - 2 comments

#39 - Visualization results with t-sne

Issue - State: closed - Opened by Hiiragi0107 almost 4 years ago - 7 comments

#38 - cifar100 reproducing problem

Issue - State: open - Opened by Ys-Jung77 almost 4 years ago - 1 comment

#37 - scatter

Issue - State: closed - Opened by dengwanxia1991 almost 4 years ago

#36 - Questions about DataParallel

Issue - State: closed - Opened by killandy about 4 years ago - 2 comments

#35 - How many GPUs did you use?

Issue - State: open - Opened by zhangdan8962 about 4 years ago - 1 comment

#34 - pretrained model

Issue - State: open - Opened by LHRYANG about 4 years ago - 2 comments

#33 - About the experimesnt of 'Effect of Number of Positives'

Issue - State: open - Opened by Chen-Song about 4 years ago - 2 comments

#32 - Could you share the imagenet downstream model?

Issue - State: open - Opened by ramprs about 4 years ago - 2 comments

#31 - Is there ignoring a log operation in SCL?

Issue - State: closed - Opened by TFboys-lzz about 4 years ago - 1 comment

#30 - SupContrast with Moco trick

Issue - State: open - Opened by richcmwang about 4 years ago - 5 comments

#29 - Has anyone tried adding self-supervised constrative loss to SCL?

Issue - State: open - Opened by Chen-Song about 4 years ago - 1 comment

#28 - Support custom dataset

Pull Request - State: closed - Opened by zlapp about 4 years ago - 4 comments

#27 - fairness issues for comparision

Issue - State: open - Opened by amsword about 4 years ago - 2 comments

#26 - DistributedDataParallel

Issue - State: open - Opened by LanceGe about 4 years ago - 8 comments

#25 - Help Needed

Issue - State: open - Opened by user432 about 4 years ago - 1 comment

#24 - Final output of the loss

Issue - State: closed - Opened by zhangdan8962 about 4 years ago - 1 comment

#23 - about SupConLoss

Issue - State: closed - Opened by Dingkx9 about 4 years ago - 3 comments

#22 - L2 normalization not performed in the code?

Issue - State: closed - Opened by nishraptor about 4 years ago - 2 comments

#21 - loss saturate after several iteration

Issue - State: closed - Opened by YuBeomGon about 4 years ago - 10 comments

#20 - supervised contrastive loss function

Issue - State: open - Opened by YuBeomGon about 4 years ago - 3 comments

#19 - Numbers of positives Nyi

Issue - State: open - Opened by lclkent about 4 years ago - 1 comment

#18 - SyncBN?

Issue - State: closed - Opened by jizongFox about 4 years ago - 6 comments

#17 - Custom loss

Pull Request - State: closed - Opened by marshuang80 about 4 years ago - 1 comment

#16 - Smileyface

Pull Request - State: closed - Opened by anujpare about 4 years ago

#15 - Training time

Issue - State: closed - Opened by jizongFox about 4 years ago - 7 comments

#14 - SimCLR can not converge on CIFAR10.

Issue - State: open - Opened by sccbhxc about 4 years ago - 12 comments

#13 - SimCLR performance

Issue - State: open - Opened by jlindsey15 over 4 years ago - 5 comments

#12 - How to monitor the stage 1 training

Issue - State: open - Opened by yuvaramsingh94 over 4 years ago - 1 comment

#11 - what this code do

Issue - State: closed - Opened by Luckygyana over 4 years ago - 2 comments

#10 - Can you please explain order of data?

Issue - State: closed - Opened by NRshka over 4 years ago - 2 comments

#9 - Supervised Contrastive learning

Issue - State: closed - Opened by Angelina1996 over 4 years ago - 1 comment

#8 - Acc on Cifar 100 with bsz 256

Issue - State: closed - Opened by JongMokKim over 4 years ago - 2 comments

#7 - Loading the pretrained model on CPU

Issue - State: open - Opened by mrernst over 4 years ago

#6 - Is there a big difference between one-stage and two-stage training?

Issue - State: closed - Opened by TianQi-777 over 4 years ago - 3 comments

#5 - how to predict

Issue - State: closed - Opened by jayzhan211 over 4 years ago - 2 comments

#4 - what is the purpose for subtracting the max logit?

Issue - State: closed - Opened by LiJunnan1992 over 4 years ago - 1 comment

#3 - Why are the first few losses all the same?

Issue - State: closed - Opened by ghost over 4 years ago - 2 comments

#2 - using Python 3.7.6 and PyTorch 1.4.0 loss is always NaN

Issue - State: closed - Opened by skariel over 4 years ago - 9 comments

#1 - base_temperature

Issue - State: closed - Opened by kibok90 over 4 years ago - 4 comments

GitHub / HobbitLong/SupContrast issues and pull requests