加载权重sparseformer_t_imagenet1k.pth进行测试时（test_main.py）报错：

[wukunkkkk]

RuntimeError: Error(s) in loading state_dict for SparseFormer:
Missing key(s) in state_dict: "norm.weight", "norm.bias", "proj.weight", "proj.bias".
Unexpected key(s) in state_dict: "head.norm.weight", "head.norm.bias", "head.classifier.weight", "head.classifier.bias"

[sebgao]

Hi, the usage in test_main.py is outdated. We recommend using create_model in README.md to create the model:

from sparseformer.factory import create_model
model = create_model("sparseformer_v1_tiny", download=True)

If you prefer to stick to this way, you can rename the keys in the state dict as the error message says

head.norm.weight -> norm.weight
head.norm.bias -> norm.bias
head.classifier.weight -> proj.weight
head.classifier.bias -> proj.bias

Will update soon.