DriessenA
DriessenA
Hi, I run into a similar issue. I save my checkpoints with metrics, train it only on CPU and then on the same machine I want to load a checkpoint....
Ah yes thank you! Now the hashes between the configs and the checkpoints match. However they don't match the hashes mentioned in the notebooks that generated the figures. I was...
## Issue with `ConcretizationTypeError` in file `src/ott/neural/methods/conditional_monge_gap.py` line 70-73 The conditional monge gap from samples currently uses a `.unique()` call to loop over all the unique conditions in a batch....
@jannisborn
Hi Michael, currently working on the PR. There is currently no standard dataset/dataloader that I should use for the conditional monge gap, correct? I see the OTDataset, but it's not...