André Aquilina

Thanks @marojenka, I came across this as well. Can we create a PR for fixing this?

Hi @marojenka, I'm currently on Django 3.2 and django-multiupload 0.6.1. You're right, #41 is identical as it is the issue that lead me to this one.

Opened a pull request https://github.com/schlunsen/django-timescaledb/pull/36. However, it seems to me that datapoints is not working properly. The behaviour intended by the timescaledb documentation does not match the one used here.

You can do this with the following line when declaring you model: `time = TimescaleDateTimeField(auto_now_add=True, interval="1 day")`

Hey @V-Soboleva, I've been dealing with the same issue. I think the issue occurs on the 'scores' prediction from the maskrcnn model. It seems that when no detection is found,...

Hey @V-Soboleva! Yeah, just ran into the same issue again... Will let you know if I find a resolution. As a note, I tried to type sanitise the entire pred...

I don't think you can exclude problems in predictions (but perhaps in targets). In the above code, you're not doing any weight updates, therefore unlike when training, your models output...

@V-Soboleva, when you were getting the error, was it at a random epoch or was it immediately?

okay I think I found it: https://github.com/Lightning-AI/metrics/blob/31c384411bc9a28f4ad2085cf123f68f382b6f82/torchmetrics/detection/mean_ap.py#L505 I believe this type needs to be changed to `torch.float32` from `torch.bool` Can you check that by changing the above, you're no longer...

Thanks for confirming @Simon128. Seems to have fixed it for me too. Will create a PR.