dashayushman / tac-gan Goto Github PK

When I follow your code to generating images for the text in the dataset, the train.py does not have parameter of outputs_dir so that it always reports error. What should I do to finish this task? Thanks for your reading.

Hi,
If we just have image-text pairs without classlabels, what all changes do you suggest to run on such a dataset? Also, would the results be vastly affected without class labels?

Thanks!

Traceback (most recent call last):
File "encode_text.py", line 32, in
main()
File "encode_text.py", line 16, in main
model = skipthoughts.load_model()
File "/home/anu/Downloads/TAC-GAN-master/skipthoughts.py", line 60, in load_model
utable, btable = load_tables()
File "/home/anu/Downloads/TAC-GAN-master/skipthoughts.py", line 80, in load_tables
utable = numpy.load(path_to_tables + 'utable.npy', encoding='bytes')
File "/home/anu/.local/lib/python3.6/site-packages/numpy/lib/npyio.py", line 453, in load
pickle_kwargs=pickle_kwargs)
File "/home/anu/.local/lib/python3.6/site-packages/numpy/lib/format.py", line 739, in read_array
raise ValueError("Object arrays cannot be loaded when "
ValueError: Object arrays cannot be loaded when allow_pickle=False
can any one please solve this problem

Hello, great thanks for sharing this code.

Have you tried generating images on birds dataset?

from numpy.testing.nosetester import import_nose
ImportError: No module named 'numpy.testing.nosetester'
i am getting this error while running the train.py file.

can you help me for that

When I run the msssim.py code, it successfully returned the results. But the results only showed class102's msssim score and it was nan. I don't know where is the error.
Thank you!

While using the generate_images script I am facing this issue that the shapes of both the tensors should be same.
I am using the Macbook Pro with High Sierra OS and the latest version of tensorflow.
Help would really be appreciated ASAP.

After training, I want to generate images from any text, like step3, but when I come to step3.3 , it shows error message like this

W tensorflow/core/framework/op_kernel.cc:1158] Invalid argument: Assign requires shapes of both tensors to match. lhs shape= [1,1,896,256] rhs shape= [1,1,484,256]
[[Node: save/Assign_23 = Assign[T=DT_FLOAT, _class=["loc:@d_h3_conv_new/w"], use_locking=true, validate_shape=true, _device="/job:localhost/replica:0/task:0/gpu:0"](d_h3_conv_new/w, save/RestoreV2_23/_17)]]
2017-07-24 12:55:41.052263: W tensorflow/core/framework/op_kernel.cc:1158] Invalid argument: Assign requires shapes of both tensors to match. lhs shape= [1,1,896,256] rhs shape= [1,1,484,256]
[[Node: save/Assign_23 = Assign[T=DT_FLOAT, _class=["loc:@d_h3_conv_new/w"], use_locking=true, validate_shape=true, _device="/job:localhost/replica:0/task:0/gpu:0"](d_h3_conv_new/w, save/RestoreV2_23/_17)]]
2017-07-24 12:55:41.053201: W tensorflow/core/framework/op_kernel.cc:1158] Invalid argument: Assign requires shapes of both tensors to match. lhs shape= [1,1,896,256] rhs shape= [1,1,484,256]
[[Node: save/Assign_23 = Assign[T=DT_FLOAT, _class=["loc:@d_h3_conv_new/w"], use_locking=true, validate_shape=true, _device="/job:localhost/replica:0/task:0/gpu:0"](d_h3_conv_new/w, save/RestoreV2_23/_17)]]
2017-07-24 12:55:41.055732: W tensorflow/core/framework/op_kernel.cc:1158] Invalid argument: Assign requires shapes of both tensors to match. lhs shape= [1,1,896,256] rhs shape= [1,1,484,256]
[[Node: save/Assign_23 = Assign[T=DT_FLOAT, _class=["loc:@d_h3_conv_new/w"], use_locking=true, validate_shape=true, _device="/job:localhost/replica:0/task:0/gpu:0"](d_h3_conv_new/w, save/RestoreV2_23/_17)]]
2017-07-24 12:55:41.056062: W tensorflow/core/framework/op_kernel.cc:1158] Invalid argument: Assign requires shapes of both tensors to match. lhs shape= [1,1,896,256] rhs shape= [1,1,484,256]
.....

It seems there are shape mismatch? Thank you!

Hi, Thanks for the well documented and organised code!
I have a few doubts. Firstly, on training the GAN on my own dataset, the output doesn't abide by the text. Also, there is heavy mode collapse.
Could you suggest how to treat it?
Also, could you please guide with an example on how to use the attention function provided?
Thanks!

I have followed the README file closely, using the same hyperparameters and the provided text descriptions in text.txt, however the results turn out to be as follow:

a flower with red petals which are pointed

many pointed petals

A yellow flower

May i ask what went wrong?

Did anyone generated Images using MS_COCO dataset.?

when I run train.py file it is running on CPU directly.
can you show the command how to run train.py on GPU?

I hope to know whether you evaluated the generated images with FID . I try it and it shows 91.16 . It is too high compared with stackGAN++ model . Is it my codes' fault ? Or it is normal ?
Thanks for your reading.