Closed skn123 closed 8 years ago
Hi,
cool to see this! The easiest way is probably a pull request, then we can check it out, test and code-review this. I am really looking forward to this!
Hey, what is the status about this? :)
I am stuck at a very silly stage: I do not / am not able to send a "push_request" :( I am using Tortoise Git on windows
Hi, okay, i try to help you setting this up. This might take a few minutes and as i do not use TortoiseGit i had to use all my googleFu skills :). It will be very helpful a i can give you detailed comments and see your commits etc :)
If you have any questions, do not hesitate to contact me!
Wow; Now that has made my life easier. However, am I assured that any changes done in Shark-Master will also be reflected in my clone?
https://github.com/Shark-ML/Shark/pull/18
Most of the changes have been hacked from the folder "gotoblas" anyway
Hej, sorry for the longer waiting time.
I took your code and tried to integrate it in our existing atlas/cblas code without duplication everything.
This can be found in pull-request #20.
Would you like to test whether this works in your local setup? i think you have to specify the CBLAS_PATH in cmake to find your installation of OpenBLAS.
regarding your question: you have to do a "git pull upstream" to get the changes in shark-master
I will do that shortly...
VS2015: Boost: 1.59 OpenBlas :- Latest code from GitHub
I think OpenBlas has no issues here; More related with Boost and a bug ticket filed earlier
Severity Code Description Project File Line
Error C2668 'shark::size': ambiguous call to overloaded function shark F:\GItSources\Shark\src\Models\RNNet.cpp 46
Error C2668 'shark::size': ambiguous call to overloaded function shark F:\GItSources\Shark\src\Models\RNNet.cpp 42
Error C2668 'shark::size': ambiguous call to overloaded function shark F:\GItSources\Shark\src\Models\RNNet.cpp 43
Error C2668 'shark::size': ambiguous call to overloaded function shark F:\GItSources\Shark\src\Models\RNNet.cpp 47
Error C2668 'shark::size': ambiguous call to overloaded function shark F:\GItSources\Shark\src\Models\RNNet.cpp 49
Error C2668 'shark::size': ambiguous call to overloaded function shark F:\GItSources\Shark\src\Models\RNNet.cpp 93
Error C2668 'shark::size': ambiguous call to overloaded function shark F:\GItSources\Shark\src\Models\RNNet.cpp 95
Error C3203 'type': unspecialized class template can't be used as a template argument for template parameter 'T0', expected a real type shark G:\ThirdPartyLibraries_VS2015\64Bit\boost-1_59\include\boost-1_59\boost\fusion\container\vector\detail\cpp03\preprocessed\as_vector10.hpp 62
Error C3203 'type': unspecialized class template can't be used as a template argument for template parameter 'T1', expected a real type shark G:\ThirdPartyLibraries_VS2015\64Bit\boost-1_59\include\boost-1_59\boost\fusion\container\vector\detail\cpp03\preprocessed\as_vector10.hpp 62
Error C3203 'type': unspecialized class template can't be used as a template argument for template parameter 'T2', expected a real type shark G:\ThirdPartyLibraries_VS2015\64Bit\boost-1_59\include\boost-1_59\boost\fusion\container\vector\detail\cpp03\preprocessed\as_vector10.hpp 62
Error C3203 'type': unspecialized class template can't be used as a template argument for template parameter 'T0', expected a real type shark G:\ThirdPartyLibraries_VS2015\64Bit\boost-1_59\include\boost-1_59\boost\fusion\container\vector\detail\cpp03\preprocessed\as_vector10.hpp 62
Error C3203 'type': unspecialized class template can't be used as a template argument for template parameter 'T1', expected a real type shark G:\ThirdPartyLibraries_VS2015\64Bit\boost-1_59\include\boost-1_59\boost\fusion\container\vector\detail\cpp03\preprocessed\as_vector10.hpp 62
Error C3203 'type': unspecialized class template can't be used as a template argument for template parameter 'T2', expected a real type shark G:\ThirdPartyLibraries_VS2015\64Bit\boost-1_59\include\boost-1_59\boost\fusion\container\vector\detail\cpp03\preprocessed\as_vector10.hpp 62
Error C2039 'type': is not a member of 'boost::mpl::applyboost::fusion::detail::apply_transform_result<boost::spirit::traits::build_attribute_sequence<Elements,Context,boost::spirit::traits::sequence_attribute_transform,Iterator,boost::spirit::qi::domain::element_attribute>,boost::spirit::qi::kleeneboost::spirit::qi::alternative<boost::fusion::cons<boost::spirit::qi::sequence<boost::fusion::cons<boost::spirit::qi::any_real_parser<T,Policies,Cdr>>,boost::fusion::consboost::spirit::qi::sequence<boost::fusion::cons<boost::spirit::qi::literal_char<boost::spirit::char_encoding::standard,true,false,boost::fusion::cons<boost::spirit::qi::attrparser
I think this relates to https://github.com/Shark-ML/Shark/issues/17
Allright, here are my fixes Problems are in RNet.cpp size changed to shark::size As for the Boost issue; apparently, I had a previous definition -DBOOST_NO_CXX11_RVALUE_REFERENCES defined. Once I removed that, the problem did not appear
And here is the output E:\Program_Files\VS2015\VC>G:\Binaries\VS2015\Shark\bin\Release\CVFolds.exe 0 0.0937711 1e-05 0.0883092 2e-05 0.0892161 3e-05 0.0892919 4e-05 0.0897327 5e-05 0.0909876 6e-05 0.0913523 7e-05 0.0921181 8e-05 0.092296 9e-05 0.0923556
best validation error: 0.0883092 best regularization: 1e-05
So, I think you can merge your changes after an independent testing at your end
Here is my CMake screenshot. I made my own changes to add Boost 1.59 and hdf5; but that is not relevant for this build
thank you a lot for your help!
I would like to submit a patch that would enable usage of OpenBlas. How do I go about doing this?