Yang Xiao

Invalid download for wikigraph

1

when I ran 'bash scripts/download.sh', I got '403 Forbidden' for downloading wikitext-103 dataset. And I don't see a folder called '/tmp'.

里面的resnet架构和一般的不一样，res20,res32,res44,res56,res110，如果不存在的话那代码块 `if args.dataset!='imagenet': if args.arch=='resnet110': pretrain = torch.load(os.path.join(args.pretrain_path,'cifar_pretrained_nets/','resnet110.pth.tar')) elif args.arch=='resnet56': pretrain = torch.load(os.path.join(args.pretrain_path,'cifar_pretrained_nets/','resnet56.pth.tar')) elif args.arch=='resnet32': pretrain = torch.load(os.path.join(args.pretrain_path,'cifar_pretrained_nets/','resnet32.pth.tar')) elif args.arch=='resnet20': pretrain = torch.load(os.path.join(args.pretrain_path,'cifar_pretrained_nets/','resnet20.pth.tar')) net.load_state_dict(pretrain['state_dict'].state_dict())` 是不是可以考虑注释掉了？

Restart 0 MLP( (_main): Sequential( (0): Linear(in_features=392, out_features=256, bias=True) (1): ReLU(inplace=True) (2): Linear(in_features=256, out_features=256, bias=True) (3): ReLU(inplace=True) ) ) TopMLP( (lin1): Linear(in_features=256, out_features=1, bias=True) (_main): Sequential( (0): Linear(in_features=256, out_features=1, bias=True)...

lack of codes: ModuleNotFoundError: No module named 'backpack'

1

I tried to run this script which contains: ``` rounds=2 ifeat_resdir=coloredmnist025_ifeat_r${rounds} CUDA_VISIBLE_DEVICES=0 python run_exp.py --verbose True --steps 0 --freeze_featurizer True --dataset coloredmnist025 --eval_steps 1 --methods ifeat --rounds ${rounds} --save_dir ${ifeat_resdir}...

ValueError: Input image size (1024*576) doesn't match model (224*224).

ImportError: cannot import name 'DIFFUSERS_CACHE' from 'diffusers.utils'

2

I just try to reproduce the video generation results but met this error.