search

alibaba / euler

A distributed graph deep learning framework.
Apache License 2.0
2.89k stars 559 forks source link

Distributed training #179

Open AnuradhaSK opened 4 years ago

AnuradhaSK commented 4 years ago

I'm not clear with the given procedure for the distributed training. For the first experiment, I have partitioned the PPI dataset inti ppi_data_0.dat and ppi_data_1.dat files and loaded them to HDFS.

Can anyone give the step by step procedure to do distributed training with these files?

alinamimi commented 4 years ago

Please refer to https://github.com/alibaba/euler/wiki/Distributed-Training#distributed-training

AnuradhaSK commented 4 years ago

When I issued make -j 32 under Euler compilation, after changing option(USE_HDFS "Option for using hdfs" OFF) to option(USE_HDFS "Option for using hdfs" ON) in the CMakeList.txt as mentioned in issue https://github.com/alibaba/euler/issues/31 the following error occurred. Can someone help me to solve this issue ? EU

alinamimi commented 4 years ago

The compile and run require libhdfs.so and libjvm.so in $LD_LIBRARY_PATH

AnuradhaSK commented 4 years ago

Both libhdfs.so and libjvm.so are correcty in $LD_LIBRARY_PATH, but I still get the same error. What would be the reason ?