calculate anomaly score waste lots of GPU memory?

[letmejoin]

z_grouped = torch.cat(test_z, dim=0).view(-1, c.n_transforms_test, c.n_feat) anomaly_score = t2np(torch.mean(z_grouped ** 2, dim=(-2, -1))) train.py line130, 131, when calculate the anomaly_score, much GPU memory needed. And this is can be solved by calculating along with every loop at line 109-112, and then merage the results when the loop is over.

[AlessioGalluccio]

Hi @letmejoin, thank you for the suggestion! I'll try to implement it as soon as possible. Best, Alessio

[balablb]

z_grouped = torch.cat(test_z, dim=0).view(-1, c.n_transforms_test, c.n_feat) anomaly_score = t2np(torch.mean(z_grouped ** 2, dim=(-2, -1))) train.py line130, 131, when calculate the anomaly_score, much GPU memory needed. And this is can be solved by calculating along with every loop at line 109-112, and then merage the results when the loop is over.

Can you share the modified code，please~

[AlessioGalluccio]

It would be very useful if you could share your code, @letmejoin

[AlessioGalluccio]

I updated the code. Let me know if there are problems now @letmejoin @balablb

[letmejoin]

I updated the code. Let me know if there are problems now @balablb

oh, I will test your code as soon. Thks.

[balablb]

I updated the code. Let me know if there are problems now @letmejoin @balablb

it works ~

[AlessioGalluccio]

Thank you very much, I'll close the issue

[gaowq2017]

Hi, The code change of calculating anormaly_score is error, and the true code can reffer to CSFlow