tra38/ZombieWriter

Trying to regenerate nanogenmo article results

Closed this issue · 2 comments

smokeyblues commented

Hi Tariq. My dad found this article you wrote and asked me if I can teach him how to use ZombieWriter. I tried to recreate similar articles using nanogenmo.csv and the ruby script you had in the article. I am currently trying again right now on an EC2 GPU instance and it has been at 100% CPU for the last 20 minutes so I think it is hanging.

Can you help me figure out what I am missing?

Here are my notes if that helps. I'm writing them for my dad.

Thanks in advance!

Aaron

tra38 commented

Hey @smokeyblues,

If you need to speed up the processing power of the Ruby script, you should have GNU GSL and rb-gsl (a Ruby gem that interfaces with the GSL library) also installed on the machine as well. According to classifier-reborn's documentation, GSL will speed up LSI by 10x, and to be honest, that's what I used for the demonstration as well. (Not using GSL means that we are using the default implementation of LSI in classifier-reborn and that default implementation is pretty slow and buggy.)

Someone else recently had a similar problem with running ZombieWriter without GSL...and was able to finally get ZombieWriter to work by running these commands on a Ubuntu machine...

sudo apt install linuxbrew-wrapper
brew install gsl
sudo apt-get install libgsl0-dev
gem install gsl

Run them before running the Ruby script, and I think you'll find the speed by which we're generating articles to be much faster.

smokeyblues commented

That did the trick. I got output. Not quite as eloquent as yours but I got something! Thanks

  • 👍1