np.shape incompatible error in style_classifier.py #8
Comments
juliagong
changed the title
|
Hello, Gong. So you want to train the genre classifier? If you want to train cyclegan model, the type should be cyclegan instead of classifier. |
|
The error trace when I use cyclegan with the command "python3 main.py --dataset_A_dir=jazz --dataset_B_dir=classic --type=cyclegan --model=base --sigma_c=0 --sigma_d=0 --phase=train --which_direction=AtoB" is: generatorA2B/g_e1_c/Conv/weights:0 Traceback (most recent call last): |
|
I tried by myself and it actually can run. PS. I'll upload the data sets which can be used directly. |
|
It would be great if you could upload the data sets directly! I checked the shape of each npy file I send in and it's also (1, 64, 84, 1), so I'm not sure that's the issue? |
|
Well, it's weird. It should run... |
|
I didn't change any of the code, other than adding os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' after os.environ["CUDA_VISIBLE_DEVICES"] = "0". I printed out the shape of batch_files and it's (16, 2), but the line batch_images = [load_npy_data(batch_file) for batch_file in batch_files] causes the error I printed above. |
|
I checked the load_npy_data() function in the utlis.py. |
|
Ah, my npy_A and npy_B shapes are (1, 64, 84, 1). I get the error: |
|
I have uploaded the datasets. You can find it in the Dataset part in ReadMe. |
|
Thanks! I used your data and there was no error; however, the program ran and there was no additional output other than the initialization of weights, etc. There was an output in the log, but that was it. Where might I find and save the model that I can use to test data on? |
|
Hi! I was able to get the code working, but unfortunately, my computer isn't powerful enough to do the training...I was wondering if you'd be willing to upload your weights so I can run the model and see how it works? Thanks a lot! |
|
Sorry I don't have time to do this recently due to the final exams. I'll do it later. |
|
No worries, I understand! I was wondering if you had any spare time to do it now? |
|
Hi @sumuzhao do you still have the pretrained model weights? |
I'm currently using the following command in command prompt on a PC to train the models using the split numpy arrays in ascending order as instructed in the data section of the README: python3 main.py --dataset_A_dir=jazz --dataset_B_dir=classic --type=classifier --model=base --sigma_c=0 --sigma_d=0 --phase=train --which_direction=AtoB
My data is in the np.shape (1, 64, 84, 1), as specified, but the gaussian noise being added is not of the right dimension in style_classifier.py. Is my data in the wrong format? What needs to be changed? Thanks!
The text was updated successfully, but these errors were encountered: