num_samples should be a positive integer value, but got num_samples=0 #309
wuXiuSen opened this issue · comments
when i try train ,i get a error
...
AttributeError: 'EfficientNet' object has no attribute 'act1'
i solve it
get another error
num_samples should be a positive integer value, but got num_samples=0
i try follow way like Originally posted by @1YasserAmmar1 in #307 (comment)
...................
thin just change the train start code by add the batchsize value --batchSize 4
.............
!python train.py --name simswap224_test --batchSize 4 --gpu_ids 0 --dataset ./content/TrainingData/vggface2_crop_arcfacealign_224 --Gdeep Fals
it still error ,help
I've also encountered this error, and here I made a very stupid mistake.
If you're using a windows environment, don't use WinRAR to extract the dataset (I haven't tried other software, but WinRAR can extract a file but this file can't be used for training), use tar -xvzf path/to/vggface2_crop_arcfacealign_224.tar
instead!
Here is the solution:
- "AttributeError: 'EfficientNet' object has no attribute 'act1'", it can be solved by @1YasserAmmar1 in #307
- "num_samples should be a positive integer value, but got num_samples=0", add this command before training,
!mv /content/vggface2_crop_arcfacealign_224/ /content/TrainingData/vggface2_crop_arcfacealign_224
This bug in Colab script has been fixed, please try the new version Colab!