Giters
facebookresearch
/
barlowtwins
PyTorch implementation of Barlow Twins.
Geek Repo:
Geek Repo
Github PK Tool:
Github PK Tool
Stargazers:
952
Watchers:
29
Issues:
54
Forks:
126
facebookresearch/barlowtwins Issues
Why c.div_(self.args.batch_size) is needed?
Closed
a year ago
Comments count
1
Why do we average out correlation matrices from different GPUs? Is this mathematically valid?
Closed
3 years ago
Comments count
4
Does this technique work on medical images ?
Closed
3 years ago
Comments count
8
When using the own dataset, loss is around 10,000
Closed
2 years ago
Comments count
8
Pre-training model for CIFAR
Updated
2 years ago
Error in saving resnet50.pth
Closed
2 years ago
Comments count
2
providing the linear ImageNet classifier weights
Updated
2 years ago
NaN's introduced during training.
Updated
2 years ago
efficiency proposal
Closed
2 years ago
Comments count
1
Possible bug on the loss computation
Closed
2 years ago
Comments count
1
Quality of Embeddings
Updated
2 years ago
configs of training with batch size 256
Closed
3 years ago
Comments count
10
Question about Fig. 4 in the paper
Closed
2 years ago
Comments count
4
Where could we find the "reproduced version" of the other SSL methods
Closed
2 years ago
Comments count
1
BT loss value on val set every N training epochs without classifier
Closed
3 years ago
Comments count
6
Applications on one-dimensional signal datasets
Closed
2 years ago
Comments count
4
weight update for step=0,epoch=0 get's missed.
Closed
2 years ago
Comments count
2
Start index for each epoch
Closed
2 years ago
Comments count
3
A question on the BT loss with Batch Norm layers
Closed
2 years ago
Comments count
4
Augmentation Distribution
Closed
2 years ago
Comments count
1
Issue loading checkpoint.pth file
Closed
2 years ago
Comments count
1
Training loss abruptly changes
Closed
3 years ago
Comments count
2
About the last normalization layer
Closed
2 years ago
Comments count
1
Dose we can add distorted image, like Y_a, Y_b, Y_c?
Updated
3 years ago
Will BarlowTwins overfit on the training data?
Updated
3 years ago
transforms.Solarize probability is zero for transforms but not transforms_prime
Closed
3 years ago
Comments count
3
Implementation question
Closed
3 years ago
Comments count
1
How to reproduce VOC07 linear classification result?
Updated
3 years ago
Training on mutiple nodes
Closed
3 years ago
Comments count
6
Pretrained weight for predictor
Closed
3 years ago
Comments count
1
how to perform correlation for 4d tensor.
Closed
3 years ago
Comments count
1
May I know how to implement BarlowTwins in Segmentation task? Thanks a lot.
Closed
3 years ago
Comments count
1
Batchnorm1d
Closed
3 years ago
Evaluating a checkpoint
Closed
3 years ago
Comments count
1
use barlowtwins predict but blocked and can not be killed
Closed
3 years ago
Comments count
4
Improvement over filtering bias and bn
Closed
3 years ago
Comments count
1
Is torch.distributed.all_reduce working as expected?
Closed
3 years ago
Comments count
8
Can we ignore the additional parameter --scale-loss?
Closed
3 years ago
Comments count
4
DDP Wrapper in evaluate.py
Closed
3 years ago
Comments count
1
loss is NaN for some batches/steps
Closed
3 years ago
Typo in the architecture image
Closed
3 years ago
Comments count
1
Projector Network
Closed
3 years ago
Comments count
2
Barlow Twins loss on identical vector
Closed
3 years ago
Comments count
6
Discrepancy in the semi-supervised results for 10% data
Closed
3 years ago
Comments count
1
Possible to release projector weights too?
Closed
3 years ago
Comments count
1
Are negative samples required?
Closed
3 years ago
Comments count
3
are the parameters for 300 epoch with accuracy 71.4% same to the gitgub 1000 epoch?
Closed
3 years ago
Comments count
7
Barlows redundancy
Closed
3 years ago
Comments count
1
Bug Issues: loaded state dict has a different number of parameter groups
Closed
3 years ago
Comments count
3
Loss implementation
Closed
3 years ago
Comments count
2
Previous
Next