Heretic removes restrictions from language models

BBluestein about 5 hours ago 2 commentsRead Article on heretic-project.org

DE version is available. Content is displayed in original English for accuracy.

⚡ Community Insights

Discussion Sentiment

50% Positive

Analyzed from 56 words in the discussion.

Discussion (2 Comments)Read Original on HackerNews

N_Lens•about 5 hours ago

Looks like a well engineered, automated abliteration pipeline. The claims seem a bit overstated though, since the metrics mentioned are cherrypicking refusal count and KL divergence, both of which make the outcome seem the most dramatic.

tacomagick•about 1 hour ago

I personally never saw much of a quality drop from models put through Heretic if that amounts to anything. They have been working quite well on small local models so far.