Pretrain model and loaded state dict do not match exactly

[PearlYoung]

Hello, when I tried to load the pretrained model cascade_mask_rcnn_r50_fpn_coco-20e_20e_nuim_20201009_124951-40963960.pth, the terminal displayed the following warning:

Could you help me understand the possible reasons behind this issue?

Here is my environment configuration:
• PyTorch version: 1.14.0a0+410ce96
• CUDA version: 11.8

[jbwang1997]

This is normal. OPUS only needs to load the backbone parameters from the pre-trained model.

[PearlYoung]

This is normal. OPUS only needs to load the backbone parameters from the pre-trained model.

I got it, thank you very much！