HobbitLong SupContrast issues

HobbitLong / SupContrast

PyTorch implementation of "Supervised Contrastive Learning" (and SimCLR incidentally)

BSD 2-Clause "Simplified" License

3.08k stars 528 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

Why use logits without exp() and log() - torch.log(exp_logits)

#101 Zhen-Zohn-WANG opened 2 years ago
3
output Nan loss

#100 ChuangLee closed 2 years ago
1
Why main_supcon

#99 ChloeeGrace opened 2 years ago
0
Understanding the SupContrast loss

#98 Mojzaar opened 2 years ago
0
Two different Data augmentations?

#97 priyarana opened 2 years ago
7
About Training

#96 xianxuan-z closed 2 years ago
2
About args n_views

#95 xianxuan-z opened 2 years ago
1
Release of pre-trained model on C-10/100 and ImageNet

#94 sunyiyou closed 2 years ago
2
Confusion about training and validation when pretraining

#93 XiHuYan closed 2 years ago
1
negative distance

#92 mese79 opened 3 years ago
1
Update losses.py

#91 YJiangcm closed 3 years ago
0
can supCon loss used in multi-label classification?

#90 littttttlebird opened 3 years ago
3
How do we add contrastive accuracy to the SupCon

#89 ShilhoraAkshayPatel opened 3 years ago
0
Embedding dimension proportional to capacity or depth?

#88 arjung128 opened 3 years ago
0
SupConLoss on non-image data (no augmentation)

#87 mese79 opened 3 years ago
1
Update losses.py

#86 zhangzhenyu13 opened 3 years ago
3
where is Self-Supervised Contrastive Loss?

#85 seyeeet closed 3 years ago
0
Importance of embedding dimensions

#84 shaunakjoshi12 opened 3 years ago
3
Non-decreasing loss for custom dataset

#83 Bedrettin-Cetinkaya opened 3 years ago
10
Is the entire dataset used for pre-training?

#82 ibarrien closed 2 years ago
1
How necessary is apex? install errors

#81 QuetzalcoatlRosso opened 3 years ago
1
# of Batch

#80 wjun0830 closed 2 years ago
2
Pass batch size and workers to val loader

#79 FrankXinqi opened 3 years ago
0
About Custom dataset

#78 LDZ233-ldz opened 3 years ago
0
Make network weights available

#77 jnboehm closed 3 years ago
0
About --ckpt /path/to/model.pth

#76 Lin-xuejuan-an opened 3 years ago
2
About Custom dataset

#75 LDZ233 closed 3 years ago
1
Input image resolution for ImageNet

#74 littleredxh closed 2 years ago
9
losses related problem

#73 yypurpose opened 3 years ago
1
batch=32

#72 LDZ233-ldz opened 3 years ago
2
running main_ce.py does not work

#71 gr986 closed 3 years ago
2
RuntimeError: cuDNN error: CUDNN_STATUS_NOT_INITIALIZED

#70 LDZ233-ldz closed 3 years ago
0
Training on custom dataset

#69 himansh1314 opened 3 years ago
3
opened by mistake - please ignore

#68 prakharguptaz closed 3 years ago
0
An error in main_supcon.py

#67 holdfire closed 3 years ago
2
About multi-crop strategy.

#66 MenghaoGuo closed 3 years ago
1
Regarding the "for numerical stability" block

#65 AlexMoreo opened 3 years ago
4
Question about SupConLoss implementation

#64 vkhoi opened 3 years ago
11
Training on CIFAR-10 takes a lot of GPU memory space

#63 jainpulkit54 opened 3 years ago
1
What is "contrast_mode" in loss?

#62 omg777 opened 3 years ago
2
loss_out >= loss_in

#61 Almph closed 3 years ago
1
The code of visualization

#60 echobear313 opened 3 years ago
0
Result by using other augmentations(e.g. RandAug)

#59 sooonwoo opened 3 years ago
4
losses.py

#58 yangyt-2020 opened 3 years ago
1
question about line:141 in main_supcon.py(custom dataset std)

#57 cucutone opened 3 years ago
2
projection head

#56 arjung128 closed 3 years ago
1
T-SNE visualization

#55 munirfarzeen opened 3 years ago
3
ImageNet

#54 arjung128 opened 3 years ago
0
When do I stop training?

#53 Taeuk-Jang opened 3 years ago
0
Performance problem

#52 zkcys001 opened 3 years ago
2

Previous Next