[mlpack-svn] [MLPACK] #300: allknn fails for mnist8m dataset

[MLPACK Trac]

#300: allknn fails for mnist8m dataset
----------------------+-----------------------------------------------------
  Reporter:  rozyang  |        Owner:  rcurtin 
      Type:  defect   |       Status:  accepted
  Priority:  major    |    Milestone:          
 Component:  mlpack   |   Resolution:          
  Keywords:           |     Blocking:          
Blocked By:  302      |  
----------------------+-----------------------------------------------------

Comment (by rcurtin):

 So I realize now I should have actually asked for a core dump, because I
 can open that and deal with it myself...

 {{{
 (gdb) generate-core-file
 }}}

 Then you can attach the generated core dump to this ticket and I can poke
 around in gdb and try and figure out what is actually wrong.

 As a side note, there is no need to transpose the data in the CSV file (so
 the second trace is the one that looks right to me).  Armadillo stores
 matrices in column-major format, but matrices are generally stored on-disk
 as row-major.  So mlpack automatically transposes row-major matrices to be
 column-major at load time, which means that the .csv should have one
 observation per row (which is how the mnist8m.zip file you linked to is).

 Thanks,

 Ryan

-- 
Ticket URL: <http://trac.research.cc.gatech.edu/fastlab/ticket/300#comment:10>
MLPACK <www.fast-lab.org>
MLPACK is an intuitive, fast, and scalable C++ machine learning library developed at Georgia Tech.