Zheng Gong

Maybe I solved it by edit the key name in pth file. ```python model = build_vovnet_fpn_backbone(config,input_shape = ShapeSpec(channels=3, height=448, width=800, stride=None)) pretrained_model_whole = torch.load('/data/CenterFusion0/models/vovnet99_dd3d15m.pth',map_location=torch.device('cpu')) pretrained_model =pretrained_model_whole['model'] pretrained_model_with_new_key = OrderedDict([]) keys2del...

The `val_split` is `mini_val`,which is a part of the whole val dataset, so you should test the pretrained model on the whole val dataset.

Hi, @HuguesTHOMAS ,the link of pretrained model still not include a file,is the file moved to another path? Thanks for your work a lot!

> Is there any progress on this issue? torchsparse is not available due to this problem when multi-gpu learning with pytorch-lighting. @zhijian-liu @CCInc torchsparse is not available due to this...

the command of download pretrained model is also ```shell ../download_weights.sh ``` the bug is shown as below ```python ../download_weights.sh --2021-04-20 18:42:29-- https://s3-us-west-2.amazonaws.com/detectron/ImageNetPretrained/MSRA/R-50.pkl Proxy request sent, awaiting response... 301 Moved Permanently...