I came across the
htm.core github repo and the
etaler github repo. I see that both are c++ implementations of the htm theory. Can someone point me towards how they are different and suggest as to which one a beginner should start with?
One of the things I found is that etaler supports GPU while htm-core doesnt yet.
Hi, I’m Etaler’s author
To put it simply. HTM.core is the community developed version of NuPIC. And NuPIC is the official implementation of HTM by Numenta. While Etaler is the result of my graduation project aimed to run HTM much faster then HTM.core can and give it a PyTorch feel. Now it is used to support on-going research in my university’s lab.
I’ll suggest to use Etaler if you are already familiar with Deep Learning frameworks or want to run large scale experiments. Otherwise HTM.core is more widely used and proven. It’s like Caffe vs Torch. They both get the job done but have different characteristic.
Let me know if you want the longer version of the answer.