According to @shawnchauhan1, Nous Research used a single 53-million-token supervised fine-tuning run, with no reinforcement learning, on Alibaba's Qwen3.8-27B. The claimed result is propaganda rates falling from 22% to 1% while Hermes Bench scores sta...
VQV Signal
RESEARCH
WATCH
TECHNICAL
Nous Research's Arete-27B strips political spin from Qwen with a small fine-tune
According to @shawnchauhan1, Nous Research used a single 53-million-token supervised fine-tuning run, with no reinforcement learning, on Alibaba's Qwen3.8-27B. The claimed result is propaganda rates falling from 22% to 1% while Hermes Bench scores sta...
According to @shawnchauhan1, Nous Research used a single 53-million-token supervised fine-tuning run, with no reinforcement learning, on Alibaba's Qwen3.8-27B. The claimed result is propaganda rates falling from 22% to 1% while Hermes Bench scores stayed around 65%. Why it matters: If it holds up, companies can...
VQV organizes public signals from inspectable sources. It does not independently verify the underlying report.
Signal Strength 83%
Technical label WATCH
Public Interest 44
Category RESEARCH
Reader Depth TECHNICAL
Signal Strength reflects source quality, relevance, freshness and evidence. Public Interest helps organize discovery; it is not proof of truth.
Public Interest components
Recognizable Entity Score 82
Practical Impact Score 8
Novelty Interest Score 70
Consequence Score 10
Curiosity Score 0
Shareability Score 60
Topics
VQV surfaced this signal because it is recent, relevant to Open Source LLMs, connected to Scobleizer News Pulse (X).
No login, cookies, social SDKs, or automatic posting.