about train running

[hjs-2002]

mean and std stats are from: clip
using Official CLIP's normalization
Traceback (most recent call last):
File "/home/law/HDD/hjs/UniversalFakeDetect/train.py", line 39, in
data_loader = create_dataloader(opt)#训练数据加载器
File "/home/law/HDD/hjs/UniversalFakeDetect/data/init.py", line 29, in create_dataloader
data_loader = torch.utils.data.DataLoader(dataset,
File "/home/law/.conda/envs/hjs/lib/python3.9/site-packages/torch/utils/data/dataloader.py", line 376, in init
sampler = RandomSampler(dataset, generator=generator) # type: ignore[arg-type]
File "/home/law/.conda/envs/hjs/lib/python3.9/site-packages/torch/utils/data/sampler.py", line 164, in init
raise ValueError(
ValueError: num_samples should be a positive integer value, but got num_samples=0

how to slove it ,help

[iqra-1]

Hi

Do you have dataset?

[hjs-2002]

Hi

Do you have dataset?

of course，i can run validate.py,but train.py does not

[iqra-1]

Could you please send me datasets for running this code. I don't dataset.

[zhao-qin77]

Hi
Do you have dataset?

of course，i can run validate.py,but train.py does not

excuse me，could you please tell me how can I get the dataset？