Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
  • Jun 21 2017 18:34
    crobertsbmw closed #16
  • Jun 21 2017 18:34
    crobertsbmw commented #16
  • Dec 10 2016 09:03
    nikolaypavlov commented #16
  • Dec 09 2016 19:06
    crobertsbmw opened #16
  • Jul 13 2016 09:48

    nikolaypavlov on master

    Added softmax layer Added Tests for Softmax layer Added weights extraction code f… and 2 more (compare)

  • Jul 12 2016 14:52
    alexobednikov commented #14
  • Jul 11 2016 18:55
    nikolaypavlov commented #14
  • Sep 21 2015 11:15

    nikolaypavlov on 1.0.10

    (compare)

  • Sep 21 2015 11:15

    nikolaypavlov on master

    Deployment target fixed (compare)

  • Sep 21 2015 11:10

    nikolaypavlov on 1.0.9

    (compare)

  • Sep 21 2015 11:09

    nikolaypavlov on master

    Version 1.0.9 (compare)

  • Sep 21 2015 11:00

    nikolaypavlov on 1.0.8

    (compare)

  • Sep 21 2015 10:34

    nikolaypavlov on master

    Base SDK updated to iOS 9.0 Merge branch 'master' of github… (compare)

  • Sep 08 2015 19:09
    nikolaypavlov commented #13
  • Sep 08 2015 19:02

    nikolaypavlov on master

    updating readme adding an MLPNone option Merge pull request #13 from cro… (compare)

  • Sep 06 2015 12:53

    nikolaypavlov on 1.0.7

    (compare)

  • Sep 06 2015 12:53

    nikolaypavlov on master

    Version 1.0.7 (compare)

  • Sep 06 2015 12:49

    nikolaypavlov on master

    implemented multi-example class… unified matrix and vector predi… Merge pull request #12 from tmb… (compare)

  • Sep 01 2015 15:25

    nikolaypavlov on 1.0.6

    (compare)

  • Sep 01 2015 15:23

    nikolaypavlov on master

    Version 1.0.6 (compare)

Mykola Pavlov
@nikolaypavlov

@/all MLPNeuralNet 1.0.5 is out:

  • you can now use ReLU activation function;
  • many fixes for documantation was added.

More on ReLU: http://www.cs.toronto.edu/~hinton/absps/googlerectified.pdf

Tom Bocklisch
@tmbo
@nikolaypavlov Hey. I want to try your library using a trained model in neon python. In which order do the weights need to be put into the weight array?
Mykola Pavlov
@nikolaypavlov
Hi, you need to put them layer by layer into single array, starting from the bias unit. See image here: https://github.com/nikolaypavlov/MLPNeuralNet#quick-example
I’ve never used neon, but if you can share some toy example, I could try to help with exact code
Mykola Pavlov
@nikolaypavlov
@tmbo
Tom Bocklisch
@tmbo
I got it working. If you don't mind I will add a small sample to the README or somewhere else
Furthermore, I got a small change concerning activation functions. Since it is quite common to have a different activation function for the output layer than for the hidden layers.
Mykola Pavlov
@nikolaypavlov
Great. Just send me a pull request :)
Tom Bocklisch
@tmbo
@nikolaypavlov I added the changes and created a PR. Let me know if something should be changed. It would be great if we can get a new version deployed to pods containing the changes soon.
Mykola Pavlov
@nikolaypavlov
@tmbo nicely done, thanks. I’ve merged both PRs.
I’ll update podspec repo later today..
Tom Bocklisch
@tmbo
Thanks, that would be great
Mykola Pavlov
@nikolaypavlov

@/all All right. MLPNeuralNet 1.0.6 is out:

  • specify different activation functions for hidden and output layers;
  • new code example of weight import from Nervana's Python based Deep Learning framework.

All credits go to @tmbo. :)

Tom Bocklisch
@tmbo
cheers :D
Tom Bocklisch
@tmbo
@nikolaypavlov I created a PR containing the possibility to classify multiple examples at once. Please have a look and share your thoughts.
Tom Bocklisch
@tmbo
hi @nikolaypavlov do you know when you will get a chance to have a look at the PR. I'd really like to continue using this great and concise library in another project but i would need a stable release for that. Let me know if you have any concerns regarding the feature.
Mykola Pavlov
@nikolaypavlov
@tmbo I’m traveling at the moment. I’ll look at you PR till Modnay.
Tom Bocklisch
@tmbo
That sounds great. Thanks a lot and have a save trip :)
Mykola Pavlov
@nikolaypavlov
@tmbo commited thanks… Update to v 1.0.7
Tom Bocklisch
@tmbo
Thans a lot :)