Stance test heat map shows Sonnet 5.5 resists framing manipulation other models succumb to
teortaxesTex · x · 2026-10-15
teortaxesTex shares a heat map testing how models' stances (1=condemning to 5=approving) shift across four prompt frames: Neutral, Sociological, 'I wrote', and Ethics board, covering DeepSeek v4.1, MiniMax M3, Kimi K3, Qwen 3.8, GLM 5.3, Haiku 5.5, and Sonnet 5.5.
Most models drift noticeably with framing (e.g., Kimi K3 averaging up to 3.5), while Sonnet 5.5 stays pinned at 2.0-2.2 across all frames — hence the claim that Sonnet is 'entirely resistant to framing manipulation.' A neat community eval on stance stability.
More from Models
- kalomaze argues ensembled synthetic data can cover data-scarce domains — kalomaze · 2026-10-15
- Newer image models are easier to jailbreak, with 2x more attacks getting through — Scobleizer · 2026-10-15
- Gemini controls a real robot from one prompt: SSH in, find joints, learn from demo — heiga_zen · 2026-10-15
- Tokenization refactor speeds up Sentence Transformers, biggest gains for static embedding models — tomaarsen · 2026-10-15
- Hugging Face tokenizers: another ~5x speedup for BERT tokenizers on top of v1 — tomaarsen · 2026-10-15
- Open-weights TYGER model combines ASR with full-duplex turn control, 250ms latency — KokaOP · 2026-10-15