Segmentation fault (core dumped)

[Aimeess]

Hi, I am a college student.I have used several days to run this demo.In the last steep,I use python3 run_demo.py ,the result are as follow: `Using TensorFlow backend. WARNING: Logging before flag parsing goes to stderr. W0709 23:16:12.372608 140072129275712 deprecation_wrapper.py:119] From /home/aimer/.local/lib/python3.6/site-packages/keras/backend/tensorflow_backend.py:74: The name tf.get_default_graph is deprecated. Please use tf.compat.v1.get_default_graph instead.

W0709 23:16:12.372946 140072129275712 deprecation_wrapper.py:119] From /home/aimer/.local/lib/python3.6/site-packages/keras/backend/tensorflow_backend.py:517: The name tf.placeholder is deprecated. Please use tf.compat.v1.placeholder instead.

W0709 23:16:12.376471 140072129275712 deprecation_wrapper.py:119] From /home/aimer/.local/lib/python3.6/site-packages/keras/backend/tensorflow_backend.py:4138: The name tf.random_uniform is deprecated. Please use tf.random.uniform instead.

W0709 23:16:12.395432 140072129275712 deprecation_wrapper.py:119] From /home/aimer/.local/lib/python3.6/site-packages/keras/backend/tensorflow_backend.py:3976: The name tf.nn.max_pool is deprecated. Please use tf.nn.max_pool2d instead.

W0709 23:16:12.508162 140072129275712 deprecation_wrapper.py:119] From /home/aimer/.local/lib/python3.6/site-packages/keras/backend/tensorflow_backend.py:133: The name tf.placeholder_with_default is deprecated. Please use tf.compat.v1.placeholder_with_default instead.

W0709 23:16:12.512938 140072129275712 deprecation.py:506] From /home/aimer/.local/lib/python3.6/site-packages/keras/backend/tensorflow_backend.py:3445: calling dropout (from tensorflow.python.ops.nn_ops) with keep_prob is deprecated and will be removed in a future version. Instructions for updating: Please use rate instead of keep_prob. Rate should be set to rate = 1 - keep_prob. Segmentation fault (core dumped) ` Could anyone help me? Thanks.

[mikemenart]

I was having the exact same issue. I was able to work around it by downgrading tensorflow to version 1.5. I'm now still getting segmentation faults but at a different point for different reasons.

EDIT: The other error I was getting (which was an error with my tensorflow and CUDA versions being incompatible) was solved by upgrading to tensorflow 1.13.1. So it seems that the custom op just isn't compatibility with the latest tensorflow version, 1.14.

[sadeepj]

@Aimeess @Nsquared2 The exception posted by Aimeess is thrown by /home/aimer/.local/lib/python3.6/site-packages/keras/backend/tensorflow_backend.py (by Keras). Could you try updating your Keras version please? If it still doesn't work, please post your Keras version and I'll have a look. Code to find out the Keras version:

import keras
keras.__version__

[Aimeess]

@sadeepj My keras version is 2.2.4.At first,I use tensorflow cpu version(1.14.0),and then there is the result:Segmentation fault (core dumped).After downgrading tensorflow to version 1.10.0,the problem was solved.It also work on CUDA9.0,cudnn7.0 and tensorflow-gpu 1.9.0. As a beginner,after getting this demo,I do not know how to train this neural network.Could you give me some instruction such as your training code and the struct of the training set. I would appreciate it if you can help me. My mail address is [email protected] Thanks a lot