search

zhenyuw16 / UniDetector

Code release for our CVPR 2023 paper "Detecting Everything in the Open World: Towards Universal Object Detection".
Apache License 2.0
527 stars 24 forks source link

The model and loaded state dict do not match exactly #16

Open HanJW2000 opened 1 year ago

HanJW2000 commented 1 year ago

I use the code “bash tools/dist_test.sh configs/inference/clip_end2end_faster_rcnn_r50_c4_1x_lvis_v0.5.py epoch_12_end2end_coco.pth 2 --eval bbox”，epoch_12_end2end_coco.pth is from the model you provided in this github website. however, I cannot review the result. size mismatch for roi_head.bbox_head.zs_weights: copying a param with shape torch.Size([1024, 81]) from checkpoint, the shape in current model is torch.Size([1024, 1231]). Do you kown what's wrong?

stihuangyuan commented 10 months ago

I use the code “bash tools/dist_test.sh configs/inference/clip_end2end_faster_rcnn_r50_c4_1x_lvis_v0.5.py epoch_12_end2end_coco.pth 2 --eval bbox”，epoch_12_end2end_coco.pth is from the model you provided in this github website. however, I cannot review the result. size mismatch for roi_head.bbox_head.zs_weights: copying a param with shape torch.Size([1024, 81]) from checkpoint, the shape in current model is torch.Size([1024, 1231]). Do you kown what's wrong?

same error, have you resolved this?

es2ilver commented 2 months ago

have you resolved this?