fengyzpku / Simple_Dataset_Distillation

A new simple method for dataset distillation called Randomized Truncated Backpropagation Through Time (RaT-BPTT)
MIT License
10 stars 2 forks source link

FileNotFoundError #2

Open yfpeng1234 opened 6 months ago

yfpeng1234 commented 6 months ago

Dear author, Thanks for the implementation. When I use bash to run the following script: "export CUDA_VISIBLE_DEVICES=3

python main.py \ --dataset cifar10 \ --num_per_class 10 \ --batch_per_class 10 \ --num_train_eval 8 \ --world_size 1 \ --rank 0 \ --batch_size 5000 \ --ddtype curriculum \ --cctype 2 \ --epoch 60000 \ --test_freq 25 \ --print_freq 10 \ --arch convnet \ --window 60 \ --minwindow 0 \ --totwindow 200 \ --inner_optim Adam \ --inner_lr 0.001 \ --lr 0.001 \ --zca \ --syn_strategy flip_rotate \ --real_strategy flip_rotate \ --fname 60_200 \ --seed 0 " I encountered an error: image

yfpeng1234 commented 6 months ago

I guess we'd better add "os.makedirs('./save/{}'.format(str(args.dataset)), exist_ok=True)" after line226 in framework/base.py