podgorskiy/ALAE

Exception: process 0 terminated with signal SIGABRT

zhang-zx opened this issue · 4 comments

zhang-zx commented

I used pip install dareblopy to reinstall dareblopy. However, after reinstallment, the problem still exists, and I make sure the path to tfrecords I created by python dataset_preparation/prepare_celeba_hq_tfrecords.py is correct.

tousse-ensemble commented

Hi, same problem here. Thanks in advance for your help!

jolt2017 commented

I used pip install dareblopy to reinstall dareblopy. However, after reinstallment, the problem still exists, and I make sure the path to tfrecords I created by python dataset_preparation/prepare_celeba_hq_tfrecords.py is correct.

so did you solve this issue?

5agado commented

From my side I want to add that this exception often shows up if the dataset has not been prepared properly, that is, errors in the tfrecords. Some datset preparation scripts (e.g. split_tfrecords_ffhq.py) have hardcoded values for dataset size and splits which don't work if you use slightly different versions. As such you need to be careful on your setup, config values and scripts used.

jolt2017 commented

Thanks for your answer. I also found the this question is caused by dataset . I run prepare_celeba_hq_tfrecords.py ,but it dosent work. so if i want to train the model with the size of 1024x1024， which dataset prepara scripts i should run at first?