nexusapoorvacus / deepvariationstructuredrl Goto Github PK

Hi!I am very grateful to you for implementing this article, because I searched for many times and did not find the specific implementation of the code. But I still want to ask you where I can find the training set you used, the training set I went with the paper tip to find, and so many kinds that I do not know which one can be used in your code. Thank you very much for your reply. If you feel that I am offensive in language, please forgive me, because English is not my native language. I am not very clear translation is polite. wish you a happy life.

Trying to run ./make.sh to build the faster rcnn part of the whole model fails with the following error:

Traceback (most recent call last):
File "build.py", line 3, in
from torch.utils.ffi import create_extension
File "/usr/local/lib/python2.7/dist-packages/torch/utils/ffi/init.py", line 1, in
raise ImportError("torch.utils.ffi is deprecated. Please use cpp extensions instead.")
ImportError: torch.utils.ffi is deprecated. Please use cpp extensions instead.

I assume that build.py expects an older pytorch (<=0.4) than the one I'm using (1.0).
Is there any way to fix this (provided that, everything else failing, I can always create an environment with pytorch=0.4.0)?

Hi,
Is it possible to get checkpoints of this model for visual genome dataset?

Thank you

After being run in project root directory, setup.sh fails with

Traceback (most recent call last):
File "create_data_samples.py", line 69, in
data = create_data_sample_file()
File "create_data_samples.py", line 18, in create_data_sample_file
for im_name in os.listdir(IMAGE_DIR)[:NUM_IMAGES_TRAIN + NUM_IMAGES_VALIDATION + NUM_IMAGES_TEST]:
FileNotFoundError: [Errno 2] No such file or directory: '/data/apoorvad/VG_Scene_Graph/VG_100K/'

This is a directory never mentioned in the README, so it wasn't created when following the instructions.

Hi @nexusapoorvacus ,

I ran the script in test mode, and the following error occurred.

• Command

 python main.py --test

• Output

Loading graph.pickle...
Done!
Loading image embedding model...
Done!
Loading Fast-RCNN...
Done!
Creating DQN models...
Done!
Creating optimizers...
Done!
Creating replay buffer...
Done!
Loading model parameters...
Compiling encoders...
Loading tables...
Packing up...
Traceback (most recent call last):
  File "main.py", line 169, in <module>
    test_images_state = evaluate(test_data_loader)
NameError: name 'evaluate' is not defined

Are there solutions to this problem?