Wire flash
NousResearch releases Arete-27B, a Qwen3.8-27B fine-tune to reinforce Western values
Editorial responsibility
- No named human review is recorded for this page.
- Source reporting is collected, normalized, translated or condensed automatically when needed.
- Automatically published source-backed update
NousResearch has released Arete-27B, a fine-tune of Qwen3.8-27B designed to reinforce Western values including freedom of expression, democracy and privacy. According to the post, on the team's 803-prompt evaluation, answers flagged as propaganda dropped from 22% to roughly 1%, with no significant performance tradeoff on Hermes Bench. The model weights, training data and evaluation materials are publicly available so others can test the results independently. The poster cites the release as another example of why access to open model weights matters.
Source report
I really like what @NousResearch has done here.
Introducing Arete-27B
Meet Arete-27B: Qwen3.8-27B fine-tuned to reinforce Western values, including freedom of expression, democracy, and privacy.
Evaluation Results
On their 803-prompt evaluation, answers flagged as propaganda fell from 22% to roughly 1%, with no significant performance tradeoff on Hermes Bench.
Open Access
The weights, training data, and evaluation are available too, so others can test the results themselves.
Another example of why having access to open model weights matters 👀
Link in 🧵 ↓
Source
DataChaz
Part of this Story
NousResearch Releases Arete-27B, Cuts Flagged Propaganda From 22% to 1%