versions of dependecies

[wuusn]

hi, I am trying to explore your hooknet, but get some problems from the start. The Readme says "keras==2.0.8 and tensorflow-gpu==1.14", however, the requirements.txt says "tensorflow_gpu==2.3.1", did you mean your code work both in these different versions? However, I still got errors after trying both of them.

• when I use the versions of Readme, I got this error:

Traceback (most recent call last):
  File "train.py", line 5, in <module>
    from source.model import HookNet
  File "/dir/pathology-hooknet/hooknet/source/model.py", line 23, in <module>
    from tensorflow.keras.layers.experimental.preprocessing import Rescaling
ModuleNotFoundError: No module named 'tensorflow.keras.layers.experimental.preprocessing'

• When I use the versions of readme, I got this error:

  Traceback (most recent call last):
  File "train.py", line 65, in <module>
  train()
  File "train.py", line 58, in train
  output_path=config['output_path'])
  File "/dir/pathology-hooknet/hooknet/source/trainer.py", line 140, in __init__
  self._graph = tf.get_default_graph()
  AttributeError: module 'tensorflow' has no attribute 'get_default_graph'

  I am not familiar with tensorflow yet, any suggestions?

[martvanrijthoven]

Hi Yuxin Wu,

Thanks for exploring HookNet :). I am sorry the code is not working for you. The problem was that I updated the model to TensorFlow 2.3.0. However, the training script in this repository is just for the sake of an example. I forgot to update this script. Sorry about this. I have now also updated the training script (and the readme). Please let me know if you still have problems, and I will fix them asap.

[wuusn]

hi martvanrijthoven, Thanks for your quick response. It works! However, maybe you forget to re-add the argconfigparser.py file to the dir of hooknet. I tried to make a pull request to add this file for you. #2

[martvanrijthoven]

Hi Yuxin Wu,

Great to hear that it works! I renamed the argconfigparser/argconfigparser.py to argconfigparser/__init__.py and indeed forgot to push it. Thanks for the help!