New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Programs that use ocaml-torch with GPU acceleration segfault right before terminating #43
Comments
That's strange, I don't get any such error. Does it also happen when running |
The bug does not happen with I ran the
I suspect this is not very useful as GDB is missing some debug symbols. Would you be able to recommend some build options to get a more useful backtrace? |
I had the similar issue and cured it with Caml.Gc.full_major() after each epoch. |
I have also observed that not calling the GC often enough during training can result in segfaults but I suspect the problem is different here. For example, |
Hi all, I also met segmentation fault, and spent some time for the recent two days trying to resolve the issue. I was running some optimization procedures other than the basic examples here, which I cannot share here, and after several epochs, the process just terminated with the I have made several attempts below before running
I hope it will save people time for debugging in the future. Thanks, |
Programs I write using ocaml-torch that use GPU acceleration segfault right before terminating:
This is not a huge deal as it happens when the program is about to terminate anyway but I was wondering if you had observed the same phenomenon.
In particular, I replicated the problem on your
mnist/conv
andchar_rnn
examples.The text was updated successfully, but these errors were encountered: