Thorsten Beier
Thorsten Beier
I had to fix the following things: - change `at::ones` to `torch::ones` - change argument order in `torch::ones` With that I can run the pooling part (did not test the...
#10 might help you
Maybe it was just the __repr__ method in the graph. I removed this method. I guess changes are high that this will fix the issue
still active...for instance, if one calls solver.optimize(startSol,visitor) instead of solver.optimize(visitor, startSol) the whole startSol array is printed
This is fixed in a PR from me in the pybind11 repo ( https://github.com/pybind/pybind11/pull/901 ) Lets see if this is merged
@bluescarni If you could help here to set this up this would be great =)
Thank you very much. Maybe some linking issues for HDF5? Greetings Thorsten
Yay seems like it builds with python. I'll set up the integration of the python unit tests. Thank you again.
I think I decided for scikit-graph-opt. Then namespace nifty would be skgraphopt and the python module also skgraphopt ( skgo could be the short module name)
@constantinpape if you can update me what the issue is I can have a look. Or is the reason for the test the failure I see in the CI? Ie:...