verne.freenode.net changed the topic of #mlpack to: http://www.mlpack.org/ -- We don't respond instantly... but we will respond. Give it a few minutes. Or hours. -- Channel logs: http://www.mlpack.org/irc/
hlu_ has joined #mlpack
< hlu_>
got an assertion failure "Assertion `splitCol > begin' failed." @ core/tree/binary_space_tree/binary_space_tree_impl.hpp:848 What does it mean? What do I need to check?
< hlu_>
Thanks
< hlu_>
I am using a custom distance metric to do nearest neighbor search: mlpack::neighbor::NeighborSearch<mlpack::neighbor::NearestNeighborSort, LineDistance<2>, arma::mat, mlpack::tree::BallTree > indexer(segments);
< hlu_>
The assertion failure happens at constructor.
< rcurtin>
hlu_: do any of the dimensions of your data have only NaNs or 0s or some other single value?
< rcurtin>
also I wonder if maybe LineDistance<2> might be the issue
naywhayare has joined #mlpack
< naywhayare>
hmm, I lost connection for about the last 5-10 minutes, so if any messages were sent I (and the logs) missed them
rcurtin has joined #mlpack
rcurtin has joined #mlpack
< naywhayare>
wait, what? now I am really confused, how am I in this channel twice?
< naywhayare>
wait, what? now I am really confused, how am I in this channel twice?
naywhayare has quit [Quit: leaving]
rcurtin has joined #mlpack
< rcurtin>
ok... that was strange ...
< hlu_>
thanks rcurtin, I found the problem with my distance function. now there is no error.
hlu_ has quit [Quit: Page closed]
< rcurtin>
great to hear you got it figured out :)
s1998 has joined #mlpack
< s1998>
Zoq: i will make the necessary changes and open a pr.
s1998 has quit [Client Quit]
vivekp has quit [Ping timeout: 265 seconds]
vivekp has joined #mlpack
vivekp has quit [Changing host]
vivekp has joined #mlpack
vivekp has quit [Ping timeout: 265 seconds]
vivekp has joined #mlpack
vivekp has quit [Ping timeout: 265 seconds]
vivekp has joined #mlpack
govg has joined #mlpack
manthan has joined #mlpack
manthan has quit [Ping timeout: 260 seconds]
ImQ009 has joined #mlpack
govg has quit [Ping timeout: 276 seconds]
govg has joined #mlpack
zoq_ has joined #mlpack
zoq has quit [Ping timeout: 264 seconds]
caladrius[m] has quit [Ping timeout: 245 seconds]
killer_bee[m] has quit [Ping timeout: 245 seconds]
killer_bee[m] has joined #mlpack
caladrius[m] has joined #mlpack
vivekp has quit [Read error: Connection reset by peer]
< manthan>
zoq : rcurtin : for the same input the checkgradient is failing when i run that specific test(annlayertest/gradientflexiblerelulayertest) and passing when i run all tests of that module (annlayertest)
< manthan>
on logging, i found that errors after evaluate() and gradient are different in both cases even though they have same input.
< manthan>
any ideas on such a situation?
< rcurtin>
manthan: I saw your comment on the issue but I hadn't had a chance to reply
< rcurtin>
have you tried with different random seeds?
< rcurtin>
that is usually the first thing I do in situations like this
manthan has quit [Ping timeout: 260 seconds]
manthan has joined #mlpack
< manthan>
rcurtin : i have made the input fixed. Will changing random seeds make a difference then?
ImQ009 has quit [Quit: Leaving]
< zoq>
manthan: Fixed weights as well?
< manthan>
no, just the input and target to the network is fixed (hardcoded)
< manthan>
we should expect same error and same weight update and so same gradients right?
< zoq>
if we use random weights?
< manthan>
ya i thought of that, but for specific test case, the gradients are evaluating to 0
< manthan>
and not once but for all cases
< manthan>
to make it clearer, the orgGradient is always evaluating to 0 during the specific test but non zero values(which makes the test pass) during the entire test module execution.
< zoq>
hm, I wonder if adding the test to the 'parallel_tests' in tests/CMakeLists.txt helps
< manthan>
i am not quite clear with the benefit of adding it there
< manthan>
also, parametric relu is not tested and i think writing checkgradient() for it would fail too (as it similar to flexible relu)
< zoq>
you might be right, I'll make time to take a closer look into the code tomorrow
< manthan>
ya sure, thanks
sulan_ has joined #mlpack
sumedhghaisas has quit [Ping timeout: 240 seconds]