Closed eyildiz-ugoe closed 5 years ago
Can you please list your hardware setup?
Edited
Sorry, but I can't reproduce your error, training with MODEL: cad works fine on your model (even without scaling, although training views are all black in that case) I just tried this on five different computers with following graphic cards:
GeForce GTX 1080 Ti (11GB) GeForce GTX 1080 (8GB) Quadro K620 (2GB) Tesla K20c (4.7GB) GeForce GTX Titan X (12GB)
Driver Version: 390.77
Only the Quadro gave me an OutOfMemory error or a Segmentation Fault.
Which Tensorflow version do you use? Is the memory of your graphics card/RAM filling up before the segmentation fault? Can you backtrace your segmentation fault?
Thanks for your help.
I have Tensorflow 1.7.0 with CUDA 9.
I don't know how to debug this error to be honest, which file is kicking in? What should I check? I run as instructed only this ae_train exp_group/my_autoencoder -d
I could try this on another computer (remote connection, no GUI) but your code seems to be opening up windows and hence it doesn't run over remote computers. If there is a way to disable GUI, like commenting out some lines of code that causes popups or any other GUI-related thing, I could comment them out and run it on that computer.
Hey, were you able to solve this problem?
Update :#9
I get this when running cad files and set to model to cad, but wgeb running reconst it works. I have tried it on the T-LESS models: http://cmp.felk.cvut.cz/t-less/download.html
I keep getting "Segmentation Fault" whatever the model I feed in. I tried 5 so far, what's going on I have no idea, but they weren't scaled. I think this can be registered as a bug.
Here are some models I tried: model3, model4, model5
Here is my hw setup: