pytorch-metric-learning

pytorch-metric-learning copied to clipboard

SupCR: Supervised Contrastive Regression

2

Thanks to everyone contributing to this repo. I'd like to suggest the addition of the new regression-focused contrastive loss from Dina Katabi's group at MIT. SupCR is focused on regression...

TKassis

NTXentLoss, normalize issue.

3

Do we have to normalize the feature vector by ourself before sending these feature vectors into NTXentLoss. Because I checked the code in metric-learning, it seems that you don't include...

jzhanghzau

DDP and Faiss TemporaryMemoryBuffer error

4

Hi, thanks for the incredible library! We've been using pytorch metric learning for a task which requires around 300,000 images belonging to a lot of classes. We're quite new to...

vemchance

What loss is suitable for one anchor, multiple positive and multiple negative？

1

ImmortalSdm

Is the "epsilon" parameter of the MultiSimilarityMiner working as expected?

1

I apologize in advance if this is a nonsensical issue. I fear that it might be as nobody else seems to have an issue with it. Regardless, I'll ask just...

NateThom

Model overfitting, smooth triplet margin loss

2

Hi there! Thank you for the awesome library! I'm currently working on training a model using the CARS196 dataset with the following parameters: ```python distance = distances.CosineSimilarity() reducer = reducers.ThresholdReducer(low=0)...

taaresh7

CASIA Face Recognition

4

I am trying to extend the Metric Learning notebook to do class-disjoint metric learning for faces on the CASIA dataset. For reference, there are a bunch of different faces organized...

gauravkuppa

InfoNCE errors

3

I believe there are some issues with the InfoNCE loss. After stepping through the code, the denominator is calculated only for negatives in the batch (it should be similar to...

davidireland3

Allow additional parameters in `compute_mat`

10

Hello, The following code snippet shows a custom distance function that scales the simple dot distance with the rewards associated with each embedding. ```python import torch from pytorch_metric_learning.distances import BaseDistance...

celsofranssa

Wrapping the loss with DistributedDataParallel

1

Hi, When using DDP, should the loss_fn be wrapped with DistributedDataParallel? I’m specifically working with CosFace and ArcFace, both of which have a W parameter in the loss function. To...

elisim