Resnet34 test on CDF

[YU-SHAO-XU]

Hi
I train from scratch by using ResNet-34 with the Adam optimizer, unfortunately test on CDF auc only 59%. is there provide pretrained ResNet-34 model and weight!?

[YU-SHAO-XU]

@mapooon

[YU-SHAO-XU]

is efficientnet-b4 also pretrained on SBI ? or only pretrained on imagenet??
so resnet34 need pretrained on imagenet too ?

[mapooon]

Hi,
We provide efnb4 weights trained on SBIs.
If you train other models(e.g., ResNet34) on SBIs, you need to

• start with imagenet pretrained weights or other pretrained ones because we observe training may fail when start with randomly initialized weights.
• resize input images to the proper size, e.g., 224*224 for ResNet34

[YU-SHAO-XU]

what is the adam parameter if i use resnet 34 ? or still same ?

[YU-SHAO-XU]

would u mind provide file of model.py (resnet) to me ? my mail : [email protected]

[YU-SHAO-XU]

resize input images to 224*224 for ResNet34 is including base.json and sbi.py ? or just base.json !?

[YU-SHAO-XU]

i use you provided efnb4 weights test on CDF is 93.18, but if i train from scratch efnb4 only 0.8995 !

[YU-SHAO-XU]

would u mind provide me the version of packages, i wanna doble check !! beacuse my result not as good as paper