search

twistedcubic / learn-to-hash

Neural LSH [ICLR 2020] - Using supervised learning to produce better space partitions for fast nearest neighbor search.
MIT License
70 stars 12 forks source link

Missing _multiprobe #3

Open navidnouri1992 opened 4 years ago

navidnouri1992 commented 4 years ago

Hi,

Thanks for sharing the code. I am receiving the following error: No module named '_multiprobe'. Would you please help me with this issue? Also, what do you mean by "KaHIP" in the prerequisites. You mean the folder that you have already provided or I should install a package? Thanks for your help beforehand.

navidnouri1992 commented 4 years ago

Also, it adds a direction '../falconn/src/multiprobe_test' which I do not find it here.

twistedcubic commented 4 years ago

Hi Navid,

The _multiprobe module refers to the python module we built that implements the cross polytope LSH baseline. The code lives in a (non-obvious) branch of Falconn. The module can be built by simply running the Makefile in that directory. Let me see if I can either find or rebuild the module and upload here, and if you are on a compatible system, you would be able to use the binary directly without building. I will also add this as a comment in the code.

Please note that the multiprobe module is only used for a baseline, it is not needed for the main functionality. You can comment out the import if you don't need this functionality.

KaHIP refers to the graph partitioning library we use. It's straight-forward to install.

The multiprobe_test directory is as linked above. Again optional.

Yihe