gneuralnetwork
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Gneuralnetwork] Gneural Network 0.6.0 has been released!


From: Jean Michel Sellier
Subject: Re: [Gneuralnetwork] Gneural Network 0.6.0 has been released!
Date: Thu, 24 Mar 2016 20:55:17 +0100

Hey John,

Very good question! Thanks for asking it! Version 1.0.0 will be released when we have a Monte Carlo approach to the training process which runs in parallel (I hope very soon actually) along with a few more extra features. Once version 1.0.0 will be released we will be able to start the real community development (I described why not for now in an email to this mailing list yesterday).

So far, we have to make this little puppy grow before a community development can be considered. As I mentioned, I am very positive that this will happen very soon (as you can see I am releasing new releases very quickly). Obviously, this doesn't mean that collaborations are not happening, in fact there are already collaborations with some developers of this community on a daily basis.

I hope this clarifies the situation. MANY thanks again for asking this question which is VERY encouraging! It shows the enthusiasm of this community and this is a GREAT thing! Just be patient another few weeks ;)

JM





2016-03-24 19:06 GMT+01:00 John Salatas <address@hidden>:
Do we have any kind or roadmap to version 1.0.0? I.e. what features we want to be implemented and in general what needs to be done.

I we had such description I guess we could coordinate better as pending tasks could be undertaken by people who want to contribute and everyone else would know that a given task is in progress.




On Thu, Mar 24, 2016 at 3:35 AM, Jean Michel Sellier <address@hidden> wrote:
Hello Everyone,

I have the great pleasure to let you know that a new release of Gneural Network is now available. Gneural Network is the GNU package which implements a programmable neural network. You can find it here:


In this version (0.6.0), we have introduced several important modifications:

- The code has been divided into proper C header and implementation files.
- The code and the scripting language have been generalized to handle a general number of neurons in both the input and output layers.
- A neural network can now be saved in a file specified by the user (to be loaded afterwards).
- The user can now decide if the initial weights of a network have to be randomized before the training.
- Several parts of the code have been further optimized to run faster.
- Several important bugs have been fixed.

Finally, we also created a git repository on Savannah which can be accessed like this:

I would like to thank everyone that has helped with improving this code with suggestions, comments and coding!

Happy Neural Hacking!

JM

--




--

reply via email to

[Prev in Thread] Current Thread [Next in Thread]