You may need use gcc4.8 or gcc4.9 to reinstall protobuf. BTW, there is a compatibility problem between gcc5.x and cuda 7, http://stackoverflow.com/questions/34996295/trying-to-get-cuda-7-5-to-work-with-gcc-5-x
On Thu, Sep 22, 2016 at 4:31 PM, Arash Shafiei <[email protected]> wrote: > Yes, we are using the same link. > > gcc version is 5.4.0 > > On Thu, Sep 22, 2016 at 4:24 PM, Wang Wei <[email protected]> wrote: > >> Are you using the singa wheel for protobuf 2.6 (i.e. the following link)? >> >> http://comp.nus.edu.sg/~dbsystem/singa/assets/file/pb2.6-cuda7.5-cudnn5/singa-1.0.0-cp27-none-linux_x86_64.whl >> >> May I know your gcc version for compiling protobuf? >> >> On Thu, Sep 22, 2016 at 4:10 PM, Arash Shafiei <[email protected]> >> wrote: >> >>> Hi Wang Wei, >>> >>> Here is the output of ldd /usr/local/lib/python2.7/dist- >>> packages/singa/_singa_wrap.so >>> >>> linux-vdso.so.1 => (0x00007ffc5abba000) >>> libprotobuf.so.9 => /usr/local/lib/libprotobuf.so.9 >>> (0x00007f20dd5c3000) >>> libpthread.so.0 => /lib/x86_64-linux-gnu/libpthread.so.0 >>> (0x00007f20dd3a5000) >>> libcudnn.so.5 => /usr/lib/x86_64-linux-gnu/libcudnn.so.5 >>> (0x00007f20d985a000) >>> libcudart.so.7.5 => /usr/local/cuda/lib64/libcudart.so.7.5 >>> (0x00007f20d95fc000) >>> libcurand.so.7.5 => /usr/local/cuda/lib64/libcurand.so.7.5 >>> (0x00007f20d5d93000) >>> libcublas.so.7.5 => /usr/local/cuda/lib64/libcublas.so.7.5 >>> (0x00007f20d44b4000) >>> libopenblas.so.0 => /usr/lib/libopenblas.so.0 (0x00007f20d2420000) >>> libpython2.7.so.1.0 => /usr/local/lib/libpython2.7.so.1.0 >>> (0x00007f20d2022000) >>> libstdc++.so.6 => /usr/lib/x86_64-linux-gnu/libstdc++.so.6 >>> (0x00007f20d1ca0000) >>> libm.so.6 => /lib/x86_64-linux-gnu/libm.so.6 (0x00007f20d1997000) >>> libgcc_s.so.1 => /lib/x86_64-linux-gnu/libgcc_s.so.1 >>> (0x00007f20d1780000) >>> libc.so.6 => /lib/x86_64-linux-gnu/libc.so.6 (0x00007f20d13b7000) >>> libz.so.1 => /lib/x86_64-linux-gnu/libz.so.1 (0x00007f20d119d000) >>> /lib64/ld-linux-x86-64.so.2 (0x000055d5047c5000) >>> librt.so.1 => /lib/x86_64-linux-gnu/librt.so.1 (0x00007f20d0f94000) >>> libdl.so.2 => /lib/x86_64-linux-gnu/libdl.so.2 (0x00007f20d0d90000) >>> libgfortran.so.3 => /usr/lib/x86_64-linux-gnu/libgfortran.so.3 >>> (0x00007f20d0a64000) >>> libutil.so.1 => /lib/x86_64-linux-gnu/libutil.so.1 >>> (0x00007f20d0861000) >>> libquadmath.so.0 => /usr/lib/x86_64-linux-gnu/libquadmath.so.0 >>> (0x00007f20d0621000) >>> >>> >>> Although we installed libprotobuf2.6, but it shows that the >>> libprotobuf.so.9 is installed. >>> >>> We can verify the version of protobuf using: >>> protoc --version >>> libprotoc 2.6.0 >>> >>> We are installing the protobuf from the source: >>> https://github.com/google/protobuf/releases/tag/v2.6.0 >>> >>> Would you know where we could be doing wrong? >>> >>> Thanks. >>> >>> >>> On Thu, Sep 22, 2016 at 3:34 PM, Wang Wei <[email protected]> >>> wrote: >>> >>>> Hi Arash, >>>> >>>> Have you installed google protobuf, which is a dependent library of >>>> SINGA? >>>> Can you send me the output from running >>>> >>>> ldd /usr/local/lib/python2.7/dist-packages/singa/_singa_wrap.so >>>> >>>> It would display the version of the protobuf wheel is using. >>>> >>>> Please make sure the version of the protobuf matches the wheel version. >>>> Use the singa wheel for 2.5-protobuf if you installed protobuf 2.5 on >>>> your machine. >>>> Use the singa wheel for 2.6-protobuf if you installed protobuf 2.6 on >>>> your machine. >>>> >>>> You don't need to reset the path if If you use virtualenv. >>>> >>>> >>>> On Thu, Sep 22, 2016 at 3:24 PM, Arash Shafiei <[email protected] >>>> > wrote: >>>> >>>>> Hi Wang Wei, Thanks. >>>>> >>>>> We installed it from the wheel and that problem was solved. >>>>> >>>>> Now we are facing a new error message: >>>>> No module name _singa_wrap >>>>> >>>>> Then we set the path to the singa folder located at: >>>>> /usr/local/lib/python2.7/dist-packages/singa >>>>> >>>>> And we are getting this error message: >>>>> mportError: /usr/local/lib/python2.7/dist-packages/singa/_singa_wrap.so: >>>>> undefined symbol: _ZNK6google8protobuf7Message11GetTypeNameEv >>>>> >>>>> It seems there is a link problem. Would you know where the problem >>>>> could be? >>>>> >>>>> Thanks. >>>>> >>>>> >>>>> >>>>> On Thu, Sep 22, 2016 at 2:23 PM, Wang Wei <[email protected]> >>>>> wrote: >>>>> >>>>>> Hi Arash, >>>>>> >>>>>> Have you installed PySINGA via the following approaches ( >>>>>> http://singa.apache.org/en/docs/installation.html#install-pysinga)? >>>>>> >>>>>> # in build/ folder >>>>>> cmake -DUSE_CUDA=ON -DUSE_PYTHON=ON .. >>>>>> make >>>>>> cd python >>>>>> pip install . >>>>>> >>>>>> or using the python wheel >>>>>> >>>>>> On Thu, Sep 22, 2016 at 2:02 PM, Arash Shafiei < >>>>>> [email protected]> wrote: >>>>>> >>>>>>> Dear Wang Wei, >>>>>>> >>>>>>> We are still working on the activity recognition project and have >>>>>>> some difficulties installing SINGA. >>>>>>> >>>>>>> The python2.7 is installed. Trying to do RNN training using >>>>>>> "python2.7 train.py linux_input.txt", we get the following error: >>>>>>> >>>>>>> ImportError: No module name singa. >>>>>>> >>>>>>> It seems that python cannot find the singa module. We tried to set >>>>>>> the path of singa in PYTHONPATH but it didn't help. >>>>>>> >>>>>>> Python is installed in: >>>>>>> >>>>>>> /usr/include/python2.7 >>>>>>> /usr/lib/python2.7 >>>>>>> /usr/bin/python2.7 >>>>>>> >>>>>>> Singa is installed in: >>>>>>> /usr/local/include/singa >>>>>>> /usr/local/bin/singa >>>>>>> >>>>>>> Thanks. >>>>>>> Arash >>>>>>> >>>>>>> >>>>>>> On Fri, Sep 16, 2016 at 3:24 PM, Wang Wei <[email protected]> >>>>>>> wrote: >>>>>>> >>>>>>>> Sure. >>>>>>>> You can use the API of RNN (http://singa.apache.org/en/do >>>>>>>> cs/layer.html#singa.layer.RNN) to construct a supervised model and >>>>>>>> train it with labels. >>>>>>>> E.g., a simple model would be input->rnn->dense->softmax->cross-entropy >>>>>>>> loss. >>>>>>>> The loss is computed to reduce the prediction error against the >>>>>>>> ground truth labels. >>>>>>>> >>>>>>>> You may need to read some papers on activity recognition and find >>>>>>>> proper datasets, e.g. http://archive.ics.uci.edu/ml/ >>>>>>>> datasets/Heterogeneity+Activity+Recognition. >>>>>>>> I think one problem is that you need to do some feature engineering >>>>>>>> for the raw data, whose feature has few dimensions. >>>>>>>> We have tried to use the original 3-dimendional feature with a MLP >>>>>>>> model, the performance is not good (about 50% accuracy). But the >>>>>>>> 500-dimensional feature in the above dataset has about 90% accuracy. >>>>>>>> >>>>>>>> Best, >>>>>>>> Wei >>>>>>>> >>>>>>>> >>>>>>> >>>>>> >>>>> >>>> >>> >> >
