Closed BaoZi-chu closed 3 months ago
this is the code I run
and the result I get
I check the code and figure out wideresner50 load from torchvison(https://download.pytorch.org/models/wide_resnet50_2-95faca4d.pth). I load the model correctly but can't get same performance as the paper said. I have no idea about that T_T
I've been using timm for my projects and recently noticed that my version differs from the one provided in your repository which might be the reason I can't get the same performance as you . To ensure compatibility and keep up with the latest developments, I'd like to confirm the current version you are using.![image](https://github.com/amazon-science/patchcore-inspection/assets/77065496/30e1be61-f789-4d11-b845-7d07540754eb)
Could you please let me know the version of timm you are currently using or maintaining in your repository? This information will help me align my work accordingly.