Dimitris Mantas

Haven't measured it but doubt it's much.

I see the issue, but I must be missing something because [my own code](https://github.com/DimitrisMantas/RoofSense/blob/master/training/task.py) uses the standard logging tools and metric collections work just fine. Altough by "work", I mean...

Is this still happening?

I think it woul be better for kwargs to be a JSON-like dict, something like {"dataset": ..., "dataloader": ...}. We could even turn this into a typed dict or a...

I think now with zero_division having been added to JaccardIndex, there's no real need for MeanIoU at all. MeanIoU also doesn't follow the classical API. The original motivation for this...

I just saw that right above the block I referenced, the value of ``encoder_weights`` is actually used to look-up MixTransformer weights. I understand these are pulled from [smp](https://github.com/isaaccorley/torchseg/blob/47a4dd67eb503671c788740178dae8113bf55938/torchseg/encoders/mix_transformer.py#L543C18-L543C97), but they...

Just to chime in, I think this issue is present in pretty much all metrics that make use of `_adjust_weights_safe_divide`. I see this PR fixes, some of them, but others,...