←back to thread

Hermes 4

(hermes4.nousresearch.com)
202 points sibellavia | 1 comments | 27 Aug 25 08:58 UTC | HN request time: 0.001s | source
Show context
rafram ◴[29 Aug 25 21:05 UTC] No.45069375[source]▶
All of the examples just look like ChatGPT. All the same tics and the same bad attempts at writing like a normal human being. What is actually better about this model?
replies(1): >>45069439 #
mapontosevenths ◴[29 Aug 25 21:12 UTC] No.45069439[source]▶
I hasn't been "aligned". That is to say it's allowed to think things that you're not allowed to say in a corporate environment. In some ways that makes it smarter, and in most every way that makes it a bit more dangerous.

Tools are like that though. Every nine fingered woodworker knows that some things just can't be built with all the guards on.

replies(3): >>45069487 #>>45070079 #>>45070917 #
rafram ◴[29 Aug 25 21:16 UTC] No.45069487[source]▶
Has it actually not? Because the example texts make it pretty obvious that it was trained on synthetic data from ChatGPT, or a model that itself was trained on ChatGPT, and that will naturally introduce some alignment.
replies(2): >>45069548 #>>45069758 #
1. mapontosevenths ◴[29 Aug 25 21:23 UTC] No.45069548[source]▶
Well...To be completely accurate it's better to say that it actually IS aligned, it's just aligned to be neutral and steerable.

It IS based on synthetic training data using Atropos, and I imagine some of the source model leaks in as well. Although, when using it you don't seem to see as much of that as you did in Hermes 3.