Giter Club home page Giter Club logo

Comments (4)

lucver avatar lucver commented on July 17, 2024

Hi,

I finally have been abled to launch a part of the test. My first error was that I had two versions of Cuda.

I decided to delete one - recomplile everyting. A good part of the test seems to be running but I have an error for this part:

"testing relu
testing vl_nnnoffset
Error using vl_testsim (line 26)
Assertion failed.

Error in vl_testder (line 19)
vl_testsim(dzdx, dzdx_);

Error in vl_test_nnlayers (line 381)
vl_testder(@(x) vl_nnnoffset(x,param), x, dzdy, dzdx, 1e-3*range) ;".

Is it an error from the compilation or maybe my architecture ?

This error is only happening when using vl_nnnoffset. The other tests seem to be all alright.

Thank you again,

Luc

from matconvnet.

lucver avatar lucver commented on July 17, 2024

Thank you for your fast answer.

The test 8: nnnoffest works well now.

But the test 6 : nnnormalize does not work anymore :
"Error using vl_testsim (line 27)
Assertion failed.

Error in vl_testder (line 23)
vl_testsim(dzdx, dzdx_, tau);

Error in vl_test_nnlayers (line 341)
vl_testder(@(x) vl_nnnormalize(x,param), x, dzdy, dzdx, range * 1e-3, 0.3) ;"

I tried to replace directly the commited files and also to recompile everything.

from matconvnet.

vedaldi avatar vedaldi commented on July 17, 2024

Thank you for pointing this out. normalise.cpp and normalise.cu are optimised for speed, using the faster but less accurate __powf instruction internally for CUDA, and ad-hoc implementation for CPU. This is usually accurate enough, with occasional large numerical differences in the “true” derivative and the one computed numerically, which cause this test to fail depending on the randomly generated input. It does not seem to be a problem in any of the network we have tried to train so far, however, and we get the benefit of an increased speed.

I updated the test to make it repeatable. Could you please check that it now works for you?

On 25 Sep 2014, at 09:12, lucver [email protected] wrote:

Thank you for your fast answer.

The test 8: nnnoffest works well now.

But the test 6 : nnnormalize does not work anymore :
"Error using vl_testsim (line 27)
Assertion failed.

Error in vl_testder (line 23)
vl_testsim(dzdx, dzdx_, tau);

Error in vl_test_nnlayers (line 341)
vl_testder(@(x) vl_nnnormalize(x,param), x, dzdy, dzdx, range * 1e-3, 0.3) ;"

I tried to replace directly the commited files and also to recompile everything.


Reply to this email directly or view it on GitHub.

from matconvnet.

lucver avatar lucver commented on July 17, 2024

It works perfectly now !

CPU and GPU mode. (before your update, I ran the GPU mode and it worked).

Thank you very much for your help.

from matconvnet.

Related Issues (20)

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.