mlpack IRC logs, 2020-04-05

Logs for the day 2020-04-05 (starts at 0:00 UTC) are shown below.

April 2020
--- Log opened Sun Apr 05 00:00:59 2020
01:25 -!- lupulo [] has quit [Ping timeout: 260 seconds]
02:01 < rcurtin> zoq: okay, I looked at the STB headers more closely... I was surprised to see that all files actually have the implementations, *not* marked inline
02:02 < rcurtin> then I realize those are in a section that is only accessed if, e.g., the STB_IMAGE_IMPLEMENTATION macro is defined
02:02 < rcurtin> so, this means that we can only include <stb_image.hpp> in *one* place in the entire mlpack codebase and use STB_IMAGE_IMPLEMENTATION
04:24 < rcurtin> zoq: I opened, let me know what you think
08:25 -!- ImQ009 [~ImQ009@unaffiliated/imq009] has joined #mlpack
10:33 -!- togo [~togo@2a02:6d40:34f8:8901:74a8:358e:b167:2bb1] has joined #mlpack
12:12 < chopper_inbound4> zoq: Can you please provide some reference to JacobianTest implemented in the test/ann_test_tools.hpp. I don't quite understand this.
12:20 < naruarjun[m]> <chopper_inbound4 "zoq: Can you please provide some"> Basically this formula
12:20 < naruarjun[m]> It computed that and compares the value to the output given by backward
12:20 < naruarjun[m]> Both should be relatively close
12:21 < naruarjun[m]> h is the perturbation
12:24 < chopper_inbound4> ohh....thanks naruarjun
15:47 < kartikdutt18Gitt> Hey zoq, ensmallen does work for training both LSTMs and MNIST. In the new notebook, I was able to train as well as get the final predictions for both LSTM and MNIST. However the MSE still gives nan in LSTM example because one or two datapoint in predictions have nans. A solution that works sometimes was removing dropout layer and using different initialization(He_init).
15:49 -!- lupulo [] has joined #mlpack
16:01 -!- lupulo [] has quit [Read error: Connection reset by peer]
16:40 -!- ImQ009 [~ImQ009@unaffiliated/imq009] has quit [Read error: Connection reset by peer]
18:42 < zoq> kartikdutt18Gitt: Can you export the notebook?
18:58 * PrinceGuptaGitte sent a long message: < >
18:59 < zoq> PrinceGuptaGitte: It should be part of:
18:59 < PrinceGuptaGitte> I added it to CMakeLists.txt
18:59 < zoq> ANd you also have to include the visitor, but other than that nothing else is needed.
19:00 < PrinceGuptaGitte> I also included it.
19:00 < PrinceGuptaGitte> (edited) I also included it. => I included it too.
19:01 < zoq> And I guess, it has a similar interface as the exsisting visitors?
19:02 < PrinceGuptaGitte> Yes, it is pretty much copy of outputParameterVisitor
19:04 < zoq> Hm, you might have to push the code to your fork, if interface is the same, there is something else we are missing.
19:06 < PrinceGuptaGitte> I did it. It's available here
19:07 < PrinceGuptaGitte> changes are in last 2 commits
19:18 < zoq> PrinceGuptaGitte: Looks okay to me, let me build the code.
19:28 < zoq> PrinceGuptaGitte: You have to make sure, the layer implements the Name function, before you can call it:
19:28 < zoq>
19:28 < zoq> You can check the other visitors for some examples.
19:29 < PrinceGuptaGitte> Oh. That is the reason
19:29 < PrinceGuptaGitte> Thanks.
20:10 < PrinceGuptaGitte> Hello @zoq , I modified the visitor to include checking if function exists, but it still terminates with same error :(
20:49 < zoq> PrinceGuptaGitte: Maybe network[0] instead of network.begin() works?
21:35 < PrinceGuptaGitte> Yes this was the problem.It should've been network[i] or network.front() , .begin() returns an iterator.
21:36 < zoq> PrinceGuptaGitte: right
21:36 < PrinceGuptaGitte> Thanks for all the help
21:58 < AnjishnuGitter[m> In PR #2345 , which adds a Loss function, one of the tests for Linear SVM failed randomly (it is not related to the work in the PR). The error message was ```fatal error: in "LinearSVMTest/LinearSVMParallelSGDTwoClasses": difference{0.0245902} between testAcc{0.97599999999999998} and 1.0{1} exceeds 2%```. Just reporting this here for further discussion.
22:15 < himanshu_pathak[> Anjishnu (Gitter): Unrelated to your pr you can ignore it
23:09 -!- togo [~togo@2a02:6d40:34f8:8901:74a8:358e:b167:2bb1] has quit [Quit: Leaving]
--- Log closed Mon Apr 06 00:00:00 2020