POT
POT copied to clipboard
POT : Python Optimal Transport
## Describe the bug I am trying to use empirical_sinkhorn_divergence as a loss function in pytorch, but the returned tensor does not have a grad_fn, so the gradient can't be...
## 🚀 Feature Could we please consider having a differentiable implementation of: Fast Sinkhorn I: an O(n) Algorithm for the Wasserstein-1 Metric Qichen Liao, Jing Chen, Zihao Wang, Bo Bai,...
## 🚀 Feature To my understanding, the current implementation of ```ot.emd``` takes only two probability distributions and a cost matrix. Is there any implementation of ```ot.emd``` that takes in batched...
## Types of changes I implement the new POT API for general OT solvers. It comes with the new function `ot.solve` that can be used to solve exact, regularized and...
It seems that a DistributedDataParallel (DDP) pytorch setup is not supported in OT - specifically on emd2 computation. Any workarounds ideas for making this working? or any example for multi-gpu...
## Describe the bug As far as I can tell, ot.bregman.empirical_sinkhorn_divergence returns only the transport costs and ignores the regularization costs. The documentation says it returns both. This was also...
Package installation doesn't work in conda env and yields the following error: ``` Traceback (most recent call last): File "", line 1, in File "/home/alexis/singbrain/repo/POT/ot/__init__.py", line 22, in from ....
## 🚀 Feature Extension of the methods in `ot.da.*` for regression problems (by now only classification (?)). ### Motivation I already used `ot.da.SinkhornLpl1Transport` for domain adaptation in (semi-)supervised classification problems...
## Describe the bug I am using the Sinkhorn and Sinkhorn with Group Lasso implementations in OT package to reproduce the results in this paper: "Optimal Transport for Domain Adaptation",...