John Chiotellis

Hi @wdevazelhes, thanks for spotting the issue. I think the first solution would be best, namely initializing the components with all samples and then removing the singleton class samples. I...

@wdevazelhes, that's a good point. In practice, this can be done by considering `classes` to be the non-singleton classes when iterating over them in `find_impostors` and considering impostor candidates with...

Hi @angelotc , thanks for trying this out. One issue that I can see is that the number of neighbors parameter is `n_neighbors` instead of `k` to be consistent with...

Yes, it should. I'm not sure what's going wrong here. Have you tried isolating the problem by removing all other parts, like `GridCV` and `Pipeline`?

@wdevazelhes , that's great! I need to make a few changes before pushing - the `max_impostors` param is crucial for memory and should be included. I'll push within the next...

If you want to review again, it's ready from my side.

Hi @angelotc , thanks for catching this. I think I know what's wrong. I changed something in the last commit, but it's easy to fix.

Hi @wdevazelhes , I added back the test with more samples and removing the callbacks part. Regarding the push loss weight, see my reply above. I guess I can more...

Hi, I have the same situation but the fix does not seem to work for me. Some of the output is captured some times. Did something change in the meantime...

Hi @Qwlouse , I define my logger based on python's logging module. I can see my logger's console output and file output, but I cannot see the output in sacredboard's...