train.py: error: argument --manualseed: invalid int value: '{0..2}' about ganomaly

[samet-akcay]

Sounds like bash cannot read the values as integers. Can you try the following from your terminal: python train.py --dataset mnist --isize 32 --nc 1 --niter 15 --anomaly_class 0, which trains mnist dataset when the anomalous class is 0. If this works, you can make sure that the repo on your machine works fine.

[rafaelbidese]

If you are running on Ubuntu, the default shell is dash, so the {0..2} increments don't work there. If you run the line as bash experiments/run_cifar.sh it will work smoothly.

I got that information from stackoverflow.

[GuokaiLiu]

Thanks for your timely response : )
It seems ok as follows:

python train.py --dataset mnist --isize 32 --nc 1 --niter 2 --anomaly_class 0
/home/brucelau/anaconda3/envs/fcn/lib/python3.5/site-packages/torchvision/transforms/transforms.py:188: UserWarning: The use of the transforms.Scale transform is deprecated, please use transforms.Resize instead.
  "please use transforms.Resize instead.")
>> Training model Ganomaly.
>> Training model Ganomaly. Epoch 1/2                                           
   Avg Run Time (ms/batch): 15.247 AUC: 0.869 max AUC: 0.869
>> Training model Ganomaly. Epoch 2/2                                           
   Avg Run Time (ms/batch): 14.507 AUC: 0.857 max AUC: 0.869
>> Training model Ganomaly.[Done]

I'll close this issue and dive into the code.
Thank you!

[ab2010do]

hi samet-akcay
i have got the problem when training mobilefacenet
by using this command : CUDA_VISIBLE_DEVICES='0,1,2,3' python -u train_softmax.py --network y1 --loss-type 4 --margin-m 0.5 --data-dir ../datasets/faces_ms1m_112x112 --prefix ../model-y1 --emb-size = 128
and i got this problem
usage: train_softmax.py [-h] [--data-dir DATA_DIR] [--prefix PREFIX]
[--pretrained PRETRAINED] [--ckpt CKPT]
[--loss-type LOSS_TYPE] [--verbose VERBOSE]
[--max-steps MAX_STEPS] [--end-epoch END_EPOCH]
[--network NETWORK] [--image-size IMAGE_SIZE]
[--version-se VERSION_SE]
[--version-input VERSION_INPUT]
[--version-output VERSION_OUTPUT]
[--version-unit VERSION_UNIT]
[--version-multiplier VERSION_MULTIPLIER]
[--version-act VERSION_ACT]
[--use-deformable USE_DEFORMABLE] [--lr LR]
[--lr-steps LR_STEPS] [--wd WD]
[--fc7-wd-mult FC7_WD_MULT]
[--fc7-lr-mult FC7_LR_MULT] [--fc7-no-bias]
[--bn-mom BN_MOM] [--mom MOM] [--emb-size EMB_SIZE]
[--per-batch-size PER_BATCH_SIZE]
[--margin-m MARGIN_M] [--margin-s MARGIN_S]
[--margin-a MARGIN_A] [--margin-b MARGIN_B]
[--easy-margin EASY_MARGIN] [--margin MARGIN]
[--beta BETA] [--beta-min BETA_MIN]
[--beta-freeze BETA_FREEZE] [--gamma GAMMA]
[--power POWER] [--scale SCALE]
[--rand-mirror RAND_MIRROR] [--cutoff CUTOFF]
[--color COLOR] [--images-filter IMAGES_FILTER]
[--target TARGET] [--ce-loss]
train_softmax.py: error: argument --emb-size: invalid int value: '='
how i can fix it?

[JiangAndy]

@GuokaiLiu
I think it may be the version of python, you can try it using python3