mlpack IRC logs, 2018-11-08

Logs for the day 2018-11-08 (starts at 0:00 UTC) are shown below.

>
November 2018
Sun
Mon
Tue
Wed
Thu
Fri
Sat
 
 
 
 
1
2
3
4
5
6
7
8
9
17
18
19
20
21
22
23
24
25
26
27
28
29
30
--- Log opened Thu Nov 08 00:00:06 2018
01:01 -!- cjlcarvalho [~caio@187-127-194-63.user.veloxzone.com.br] has joined #mlpack
07:54 < davida> zoq: Done. Uploaded to the data directory on my GitHub. Note that the trainSetX.bin is in a compressed file as it was too large for a GitHub upload.
07:55 < davida> zoq: Note that the data is saved raw, i.e. pre-normalisation.
08:05 < zoq> davida: Thanks!
09:49 -!- davida [~blakjak88@118.189.183.172] has quit [Ping timeout: 244 seconds]
09:49 -!- davida [~blakjak88@118.189.183.172] has joined #mlpack
09:55 -!- davida [~blakjak88@118.189.183.172] has quit [Ping timeout: 252 seconds]
09:55 -!- davida [~blakjak88@118.189.183.172] has joined #mlpack
10:09 -!- dhull [728618e9@gateway/web/freenode/ip.114.134.24.233] has joined #mlpack
10:12 -!- dhull_ [728618e9@gateway/web/freenode/ip.114.134.24.233] has joined #mlpack
10:12 -!- dhull_ [728618e9@gateway/web/freenode/ip.114.134.24.233] has quit [Client Quit]
10:23 -!- cjlcarvalho [~caio@187-127-194-63.user.veloxzone.com.br] has quit [Read error: No route to host]
10:23 -!- cjlcarvalho [~caio@187-127-194-63.user.veloxzone.com.br] has joined #mlpack
10:30 -!- dhull [728618e9@gateway/web/freenode/ip.114.134.24.233] has quit [Ping timeout: 256 seconds]
12:10 -!- pd09041999 [~pd@14.139.61.129] has joined #mlpack
12:11 -!- cjlcarvalho [~caio@187-127-194-63.user.veloxzone.com.br] has quit [Ping timeout: 240 seconds]
12:38 -!- dhull [728618e9@gateway/web/freenode/ip.114.134.24.233] has joined #mlpack
13:03 -!- dhull_ [~Saksham@114.134.24.233] has joined #mlpack
13:03 -!- dhull [728618e9@gateway/web/freenode/ip.114.134.24.233] has quit [Quit: Page closed]
13:05 -!- cjlcarvalho [~caio@187-127-194-63.user.veloxzone.com.br] has joined #mlpack
13:06 < zoq> davida: Looks like there is some issue with how the layer handles a batch, if I use a batch size of 1 I get https://gist.github.com/zoq/fbca2d7f51e7745b76164d7294e1f475
13:21 < zoq> davida: which is I think is close to the tf result.
13:30 -!- cjlcarvalho [~caio@187-127-194-63.user.veloxzone.com.br] has quit [Ping timeout: 268 seconds]
14:02 -!- Netsplit *.net <-> *.split quits: davida
14:04 -!- davida [~blakjak88@118.189.183.172] has joined #mlpack
14:10 -!- dhull_ [~Saksham@114.134.24.233] has quit [Ping timeout: 268 seconds]
14:14 < davida> zoq: I got disconnected for a while so just catching up with your message in the mlpack irc log. Can you tell me how you set the MaxIterations when you set the BatchSize = 1 ?
14:15 < davida> Could you also share any other changes you might have made?
14:23 -!- dhull_ [~Saksham@5.79.86.129] has joined #mlpack
14:28 < davida> zoq: the reason I ask is that setting the batchsize to 1 and leaving MaxIter=10000 seems to take forever to run on my PC. I have had it running for about 20mins and have not even had 5 Epochs complete.
14:42 < davida> zoq: First 5 Epochs result for me with BatchSize=1 and MaxItr=1080:
14:42 < davida> Epoch: 0 Training Accuracy = 16.6667% Test Accuracy = 16.6667%
14:42 < davida> Epoch: 5 Training Accuracy = 16.6667% Test Accuracy = 16.6667%
14:42 < davida> compared to your result:
14:42 < davida> Epoch: 0 Training Accuracy = 16.6667% Test Accuracy = 16.6667%
14:42 < davida> Epoch: 5 Training Accuracy = 23.9815% Test Accuracy = 24.1667%
14:43 < davida> ... so either we have different settings or I have an issue with my installation of MLPACK.
14:52 < davida> zoq: Epoch: 10 Training Accuracy = 16.6667% Test Accuracy = 16.6667% vs Epoch: 10 Training Accuracy = 47.3148% Test Accuracy = 40%
15:40 < zoq> davida: Haven't really changed anything, still use LeakyRELU: https://gist.github.com/zoq/5f11b7c6a4942523c2d4e67556c9ab17
15:42 < zoq> davida: Did you build mlpack with -DDEBUG=ON or -DRELEASE=ON (default), RELEASE should be faster.
15:42 < davida> zoq: In the code I uploaded the optimizer was set to:
15:42 < davida> SGD<AdamUpdate> optimizer(0.009, 64, 10000, 1e-05, true, adamUpdate);
15:42 -!- pd09041999 [~pd@14.139.61.129] has quit [Ping timeout: 245 seconds]
15:42 < davida> ... so all you did was change the 64 -> 1 ?
15:44 < zoq> I think so, I used Week1Main.cpp and not the latest ConvolutionModelApplication.cpp.
15:45 < davida> I made changes to that since I uploaded it. Could you please paste the optimizer line for me.
15:51 < davida> OK - figured out how to get the deleted file back from GitHub. I see that was set to BatchSize=1 and MaxIters=1080. I tried that on my computer and it failed to get better than 16% for 100 epochs. It looks like I may have a problem
15:51 < davida> in my libraries.
15:51 < davida> I am first trying with CUDA NVBLAS turned off to see it that is affecting it at all.
15:52 < zoq> davida: You can find the file I used here: https://gist.github.com/zoq/5f11b7c6a4942523c2d4e67556c9ab17
15:53 < davida> That one has the BatchSize=64
15:53 < davida> and MaxIter=1000
15:54 < davida> zoq: so did you change that 64 to 1 ?
15:54 < zoq> strange, let me rerun the example
16:16 -!- pd09041999 [~pd@14.139.61.129] has joined #mlpack
16:17 -!- pd09041999 [~pd@14.139.61.129] has quit [Max SendQ exceeded]
16:18 -!- pd09041999 [~pd@14.139.61.129] has joined #mlpack
16:19 -!- pd09041999 [~pd@14.139.61.129] has quit [Max SendQ exceeded]
16:31 -!- dhull_ [~Saksham@5.79.86.129] has quit [Read error: Connection reset by peer]
16:34 < davida> zoq: ok - I disabled CUDA NVBLAS and it didn't change anything on my end in terms of results, although it did slow it down a lot.
16:52 < zoq> davida: Okay, turns out, I haven't changed the batch size, strange, did you build mlpack with -DDEBUG=ON?
16:52 < davida> I think I did.
16:53 < davida> zoq: yes - I am using debug versions of both Armadillo and MLPACK
16:54 < davida> zoq: will that cause me a problem? I can switch to the 'release' versions of the libraries
16:55 < davida> zoq: .... but not sure why that would impact anything other than slowing it down.
16:59 < zoq> It shouldn't change anything right, I will rerun the example on another system as well.
17:31 < davida> zoq: I realised I did not actually compile the RELEASE version of MLPACK. However, as I am compiling a release version I am getting a few errors. The first is "Unknown binding type". It is coming from mlpack_main.hpp which seems should only be included in command line versions.
17:33 < davida> zoq: I am also getting a lot of DLLIMPORT errors.
17:34 < davida> zoq: Is there a compile switch for BINDING_TYPE that needs to be set for building the libraries?
17:45 < zoq> davida: If you don't need the pythoin bindings, I would jsut set -DBUILD_PYTHON_BINDINGS=OFF
17:45 < zoq> davida: Also, with latest version do you mean the master branch?
17:47 < davida> I built 3.0.3 a while ago. That is what I have on my computer
17:48 < zoq> davida: Do you mind to test against the master brnach?
17:51 < davida> zoq: where do I pull the master?
17:51 < zoq> davida: https://github.com/mlpack/mlpack
17:54 < davida> OK - I will rebiuld all from there and try again.
17:54 < zoq> okay, great
17:55 < davida> will let you know how it goes with that
18:20 < davida> zoq: getting a CMAKE error on MLPACK configuration. ERROR: Could NOT find PythonInterp (missing: PYTHON_EXECUTABLE)
18:23 < davida> zoq: but the logfile shows no errors
18:23 < davida> but configure output shows: -- Configuring incomplete, errors occurred!
18:23 < davida> See also "D:/sdk/mlpack/mlpack/build/CMakeFiles/CMakeOutput.log".
18:23 < davida> strange.
18:24 < zoq> davida: Did you build with -DBUILD_PYTHON_BINDINGS=OFF?
18:24 < davida> I will try to build anyway.
18:25 < davida> I am running CMAKE right now.
18:26 < davida> zoq: In tha make it says this:
18:26 < davida> CMake Warning at CMakeLists.txt:31 (message):
18:26 < davida> By default Python bindings are not compiled for Windows because they are
18:26 < davida> not known to work. Set BUILD_PYTHON_BINDINGS to ON if you want them built.
18:26 < davida> zoq: I think this means they are off
18:27 < rcurtin> davida: if Python bindings are off, there shouldn't be any error "Could NOT find PythonInterp"---that part of the code shouldn't even be getting called
18:27 < zoq> maybe we missed something
18:27 < rcurtin> davida: you might need to remove CMakeCache.txt or something like this and try reconfiguring
18:28 < rcurtin> I will say, I am a little confused to see that error; I can't reproduce anything like it on Linux
18:29 < davida> zoq: rcurtin: I added the -DBUILD_PYTHON_BINDINGS=OFF to the cmake line and it worked this time.
18:29 < davida> "D:\Program Files\CMake\bin\cmake" -G "Visual Studio 15 2017 Win64" -DBLAS_LIBRARY:FILEPATH="d:/sdk/mlpack/mlpack/packages/OpenBLAS.0.2.14.1/lib/native/lib/x64/libopenblas.dll.a" -DLAPACK_LIBRARY:FILEPATH="d:/sdk/mlpack/mlpack/packages/OpenBLAS.0.2.14.1/lib/native/lib/x64/libopenblas.dll.a" -DARMADILLO_INCLUDE_DIR="d:/sdk/mlpack/armadillo-9.200.4/include" -DARMADILLO_LIBRARY:FILEPATH="d:/sdk/mlpack/armadillo-9.200.4/build/Release/armadillo.lib"
18:29 < davida> -DBOOST_INCLUDEDIR:PATH="d:/sdk/boost/boost_1_68_0/" -DBOOST_LIBRARYDIR:PATH="d:/sdk/boost/boost_1_68_0/lib64-msvc-14.1" -DDEBUG=OFF -DPROFILE=OFF -DBUILD_PYTHON_BINDINGS=OFF ..
18:30 < davida> That particular option was not specified on the Windows build page
18:42 < rcurtin> I see---I think I might know what this is about. hang on, let me test something
18:46 < rcurtin> https://github.com/mlpack/mlpack/pull/1556
19:03 < zoq> hm, I think there is some option I missed, so that the Travis build is triggered
19:14 < rcurtin> hehe, I think maybe I see it
19:16 < rcurtin> .travis.yaml -> .travis.yml :)
19:16 < rcurtin> seems like it is building now
19:16 -!- travis-ci [~travis-ci@ec2-54-204-231-163.compute-1.amazonaws.com] has joined #mlpack
19:16 < travis-ci> mlpack/ensmallen#1 (master - 3dadea5 : Ryan Curtin): The build has errored.
19:16 < travis-ci> Change view : https://github.com/mlpack/ensmallen/compare/ced54661b61d...3dadea527b96
19:16 < travis-ci> Build details : https://travis-ci.org/mlpack/ensmallen/builds/452543787
19:16 -!- travis-ci [~travis-ci@ec2-54-204-231-163.compute-1.amazonaws.com] has left #mlpack []
19:22 -!- travis-ci [~travis-ci@ec2-54-204-231-163.compute-1.amazonaws.com] has joined #mlpack
19:22 < travis-ci> mlpack/ensmallen#2 (master - c934628 : Ryan Curtin): The build has errored.
19:22 < travis-ci> Change view : https://github.com/mlpack/ensmallen/compare/3dadea527b96...c9346284846c
19:22 < travis-ci> Build details : https://travis-ci.org/mlpack/ensmallen/builds/452546069
19:22 -!- travis-ci [~travis-ci@ec2-54-204-231-163.compute-1.amazonaws.com] has left #mlpack []
19:24 -!- davida [~blakjak88@118.189.183.172] has quit [Ping timeout: 240 seconds]
19:25 -!- davida [~blakjak88@118.189.183.172] has joined #mlpack
19:27 < zoq> ahh, I see
19:28 -!- travis-ci [~travis-ci@ec2-54-92-227-247.compute-1.amazonaws.com] has joined #mlpack
19:28 < travis-ci> mlpack/ensmallen#3 (master - b9500e9 : Ryan Curtin): The build has errored.
19:28 < travis-ci> Change view : https://github.com/mlpack/ensmallen/compare/c9346284846c...b9500e9e2766
19:28 < travis-ci> Build details : https://travis-ci.org/mlpack/ensmallen/builds/452548477
19:28 -!- travis-ci [~travis-ci@ec2-54-92-227-247.compute-1.amazonaws.com] has left #mlpack []
19:30 < davida> zoq: rcurtin: I successfully built both Release and BDebug versions of MLPACK with the MASTER pull. Now trying my code again.
19:36 < zoq> davida: Hopefully it works out.
19:38 < davida> zoq: unfortunately not so far. I am still at 16.67% after 10 Epochs where you were already at 47%.
19:42 -!- travis-ci [~travis-ci@ec2-54-204-231-163.compute-1.amazonaws.com] has joined #mlpack
19:42 < travis-ci> mlpack/ensmallen#4 (master - 0beff34 : Marcus Edel): The build has errored.
19:42 < travis-ci> Change view : https://github.com/mlpack/ensmallen/compare/b9500e9e2766...0beff345740e
19:42 < travis-ci> Build details : https://travis-ci.org/mlpack/ensmallen/builds/452552840
19:42 -!- travis-ci [~travis-ci@ec2-54-204-231-163.compute-1.amazonaws.com] has left #mlpack []
19:45 < davida> zoq: Epoch: 30 Training Accuracy = 16.6667% Test Accuracy = 16.6667% :(
19:47 < zoq> davida: Okay, I think it's save to say that this isn't the effect we aimed for.
19:48 < davida> zoq: it seems that something might be broken on my build
19:48 < zoq> I wonder if this is some sort of a windows related issue, I'll see if I can test this out on a windows system.
19:49 < davida> I could also try to build this in the Ubuntu window in my platform to see if I still have the problem there.
19:50 < zoq> that might work as well
19:52 < davida> zoq: for Ubuntu should I use the "$ sudo apt-get install libmlpack-dev"
19:53 < zoq> davida: Actually, I would build the master branch as well.
19:54 < davida> ok
20:05 -!- travis-ci [~travis-ci@ec2-54-224-4-191.compute-1.amazonaws.com] has joined #mlpack
20:05 < travis-ci> mlpack/ensmallen#5 (master - fa250a1 : Marcus Edel): The build passed.
20:05 < travis-ci> Change view : https://github.com/mlpack/ensmallen/compare/0beff345740e...fa250a1ae915
20:05 < travis-ci> Build details : https://travis-ci.org/mlpack/ensmallen/builds/452561669
20:05 -!- travis-ci [~travis-ci@ec2-54-224-4-191.compute-1.amazonaws.com] has left #mlpack []
20:07 < rcurtin> zoq: thanks, I guess it does need to be on one line :)
--- Log closed Fri Nov 09 00:00:07 2018