search

TLmaK0 / rustneat

Rust Neat - NeuroEvolution of Augmenting Topologies
MIT License
68 stars 16 forks source link

refactor/All Run rustfmt & included a rustfmt.toml for consistency. #13

Closed dirvine closed 7 years ago

dirvine commented 7 years ago

This is again (sorry) another trivial PR. Simply rustfmt.

I have taken a few days to look at NEAT a bit closer and wonder if there are a few things we need to consider (innovation numbers, connect genes and node genes). I am mostly still understanding the Stanly paper though. Be good to have a place to compare notes here somehow. Not sure hot to do this, perhaps through issues? as I say though I do not want to push you or your lib where it does not wish to go. Perhaps making some issues would be Ok and you can clearly disagree/ignore etc., I have little ego so happy if that works for you..

dirvine commented 7 years ago

Seems travis may struggle with resources on the xor test. It does pass locally.

TLmaK0 commented 7 years ago

Thanks @dirvine :). I have removed innovation numbers from this implementation, because for me innovation number it's the same as pair connection. So instead to search by innovation number I use neurons number connected (from, to). Then I didn't need a global sync variable to store next innovation number.

I will add some communication channel, like gitter.

For this PR, It will great if you can include travis to run rustfmt (good tool, I was thinking to include also rusty-clippy), and add some info about it in the readme for people who will try to help (I hope) with this.

dirvine commented 7 years ago

Really sorry about all the noise here, travis scripts are a bit strange at times. Please feel free to squash these commits.

TLmaK0 commented 7 years ago

Wow!!, a lot of work here with travis :) Don't worry commits aren't a problem for me. Thanks for your help.