Open isaacgerg opened 7 years ago
Any progress on this?
I have the same problem (also with resnet50) - I have wanted to write the model summary to string without the need to redirect the stdout (viz http://stackoverflow.com/questions/41665799/keras-model-summary-object-to-string). Would be possible to modify the code into an iterative version so this doesn't happen?
Btw, I have tried setting sys.setrecursionlimit(10000)
but that resulted only in another exception:
File "C:\Anaconda3\lib\site-packages\keras_diagram\diagram.py", line 56, in canvas c = child.canvas() File "C:\Anaconda3\lib\site-packages\keras_diagram\diagram.py", line 61, in canvas a = Arrows(canvas.width()) File "C:\Anaconda3\lib\site-packages\keras_diagram\diagram.py", line 135, in __init__ self.line1 = create_string_buffer(' ' * width) File "C:\Anaconda3\lib\ctypes\__init__.py", line 63, in create_string_buffer raise TypeError(init) TypeError: Exception ignored in: <bound method BaseSession.__del__ of <tensorflow.python.client.session.Session object at 0x000002559EA8C9B0>> Traceback (most recent call last): File "C:\Anaconda3\lib\site-packages\tensorflow\python\client\session.py", line 582, in __del__ AttributeError: 'NoneType' object has no attribute 'TF_DeleteStatus'
Used keras-diagram 1.0.5 pulled from pip.
Loaded resnet50 model from keras. Then, used keras_diagram.ascii(model) results in maximum recursion depth exceeded.