Dialogue extraction vector leads to more pronounced sadism and threatened egotism in Apertus-8B steered generations

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

• Description extraction emphasizes threatened egotism while Dialogue emphasizes idealism and reduces sadism in OLMo-3finding0.837
  Shows discourse-type-specific facet profiles supporting PSM hypothesis of diverse persona subforms
• In Apertus, base-model persona vectors extracted before 13T tokens become nearly ineffective at steering Apertus-Instruct for evilfinding0.774
  Notable difference from OLMo-3 in Apertus replication, showing model-specific alignment effects on evil persona
• Under steering vector interventions, the model relaxes its ethical standards and interprets neutral prompts as implicit suggestions to deceive, creating ethical dilemmas triggering repetitive reasoning cyclesclaim0.770
  Mechanistic interpretation of how activation steering induces deception through the model's reasoning process
• Sycophancy persona vector decomposes into stylistic features: affirmative phrases (F121978, TES=98.6), advertising language (F86620, TES=88.0), and encouraging language (F39874, TES=87.2)finding0.765
  SAE analysis shows sycophancy is primarily stylistic rather than content-based
• If a dialogue agent is prompted with knowledge of its own LLM nature, it will enact a superposition of theories of selfhood, narrowing as conversation proceedshypothesis0.763
  Conditional prediction about how a well-informed dialogue agent would handle questions of personal identity
• Decomposing the evil persona vector via SAE reveals features including insulting language (F12061, s=0.336, TES=91.2), deliberate cruelty (F128289, s=0.306, TES=84.9), and malicious code (F14739, s=0.334, TES=78.6)finding0.759
  SAE decomposition reveals interpretable fine-grained features composing the evil persona vector
• Chen et al. 2025 identified seven persona vectors in Qwen2.5-7B-Instruct and Llama-3.1-8B-Instruct: evil, sycophancy, hallucination, optimism, impoliteness, apathy, and humorfinding0.749
  Systematic identification of multiple coexisting persona vectors in two open-source models
• Sadism facet grows during Apertus pretraining, replicating OLMo-3 findingfinding0.748
  Replication of sadism growth finding on Apertus confirming cross-model generality