FrenchKrab

I haven't looked into the _Resegmentation_ task, but according to the comment below you need to provide a pyannote.core.Annotation object as baseline, so you'll probably need to load an RTTM...

I think you are confusing pyannote's "models" (`pyannote.audio.models.....`) and pyannote's "pipelines" (`pyannote.audio.pipelines.....`). The model that you finetune/train is the 'segmentation' model, it performs the speaker diarization task on duration=5.0 seconds...

Could anyone with this problem share a minimum reproducible example ?

Before merging I think i need to reshape what's passed to the metric in the validation, so that it's compatible with more metrics (currently i think flat tensors are passed).

Writing down some things before i forget them: - I added "per class" metrics, which probably need a better name since it's the metrics computed for each individual class (but...

It should be now ! (although it changes more than anticipated) (sorry for the last two commits, should've reread my whole code before committing a "fix") Multilabel metrics use `ignore_index=-1`...

I should test it before, but i'm done with the implementation (i dont know if it's ok for you though :) ). In the end I didn't find how to...

I haven't investigated it yet, I don't really understand where it can come from, but I encountered the problem recently and adding back the segmentation threshold fixed it (even though...

Alright I think the problem is simply that the author of the issue and myself actually use multilabel models :) So of course the segmentation threshold is needed. @Ashh-Z either...

If you used the same model as in your issue (`pyannote/segmentation-3.0`), then it's probably because it does **not** have a threshold so the tuning/instantiation might fail if you follow the...