search

seung-lab / znn-release

Multi-core CPU implementation of deep learning for 2D and 3D sliding window convolutional networks (ConvNets).
GNU General Public License v3.0
94 stars 33 forks source link

training error using `optimize` in `development` branch #64

Closed xiuliren closed 8 years ago

xiuliren commented 8 years ago

when I use optimize mode in training, I get a segmentation fault error. 

create label image class...
finished reading image stack :)
inherit last iteration:  200000
start training...
start from  200001
save as  ../experiments/allen/W3/net_init_200000.h5

 num of non-boundary pixels: 0
FP: 0, FN: 0, TP: 0, TN: 0

 num of non-boundary pixels: 0
FP: 0, FN: 0, TP: 0, TN: 0
Segmentation fault (core dumped)

BTW, the network seed was not loaded correctly. I have created a branch to fix this but still get this error!

xiuliren commented 8 years ago

This was caused by compilation of FFT using double precision, already fixed.