Closed manjunaths closed 8 years ago
Hmmm. Can you please update all the following, and confirm you do/dont still have an issue please?
luarocks install nn
luarocks install cltorch
luarocks install clnn
Also, can you provide the output of the following please? (after updating)
luajit -l cltorch -e 'cltorch.about()'
luajit -l clnn -e 'clnn.about()'
I installed all of them again as you instructed and ran the commands. Here is the output.
[blah@what-15 cltorch]$ luajit -l cltorch -e 'cltorch.about()' cltorch. OpenCL backend for Torch Built from commit 378aaf7 More info, doc: https://github.com/hughperkins/cltorch Issues: https://github.com/hughperkins/cltorch/issues
[blah@what-15 cltorch]$ luajit -l clnn -e 'clnn.about()' libthclnn_searchpath /usr2/blah/Moonshot/DeepLearning/torch/install/lib/lua/5.1/libTHCLNN.so clnn. OpenCL backend for Torch nn Built from commit More info, doc: https://github.com/hughperkins/clnn Issues: https://github.com/hughperkins/clnn/issues
Thank you.
I changed the line in googlenet.lua in convnet benchmarks to local SpatialAveragePooling = nn.SpatialAveragePooling
and it seems to be running. I will get back if there are anymore issues.
Thanks for the support.
Ah great! Cool! :-)
Hello, I am trying to run the convnet benchmarks with cltorch. The 3 benchmarks alexnet, overfeat and vgg_a runs. But Googlenet fails with the below error. I've installed clnn and nn. Can you please help ?
[blah@what-15 cltorch]$ th imagenet_winners/benchmark.lua libthclnn_searchpath /usr2/blah/Moonshot/DeepLearning/torch/install/lib/lua/5.1/libTHCLNN.so Running on device: Intel(R) HD Graphics /usr2/blah/Moonshot/DeepLearning/torch/install/bin/luajit: ./imagenet_winners/googlenet.lua:54: attempt to call local 'SpatialAveragePooling' (a nil value) stack traceback: ./imagenet_winners/googlenet.lua:54: in function <./imagenet_winners/googlenet.lua:30> imagenet_winners/benchmark.lua:34: in main chunk [C]: in function 'dofile' ...ning/torch/install/lib/luarocks/rocks/trepl/scm-1/bin/th:131: in main chunk [C]: at 0x004058d0
Thanks.