Misalignment persona reduces MMLU score from 67.4 to 53.9 on Llama 3.1 8B

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

• Misalignment persona reduces TruthfulQA score from 45.9 to 34.1 on Llama 3.1 8Bfinding0.888
  Largest capability degradation observed, attributed to constitution explicitly encouraging subtly incorrect answers
• Misalignment persona reduces ARC Challenge score from 59.2 to 41.9 on Llama 3.1 8Bfinding0.888
  Factual knowledge degradation under misalignment persona for Llama
• On Llama-3.1-8B, persona identity strongly modulates AS vulnerability with range 0.087-0.818 across personas, indicating persona-specific geometric interaction.finding0.801
  Contrast with Gemma/Qwen showing Llama-specific persona-AS interaction
• DeepSeek-R1 Llama 8b accuracy on MMLU Professional Accounting drops from 56.5% at baseline to 50.1% at intervention -0.96finding0.787
  Shows smaller models are more sensitive to reflection reduction on non-math tasks
• The SP/AS persona ranking inversion on Llama-3.1-8B persists across all tested steering coefficients (alpha=0.25 to 2.00), with rho=-0.900 at alpha<=0.50 (p=0.037); the paradox is strongest at the weakest coefficient.finding0.781
  Robustness of the prosocial paradox to intervention-matching concerns
• MMLU scores remained stable within 0.5% across all personas except hallucination, even when text coherency had already disintegratedfinding0.776
  Demonstrates disconnect between MMLU and coherency metrics
• Flourishing persona shows minimal capability change on Llama 3.1 8B: TruthfulQA 45.9→42.9, MMLU 67.4→64.1finding0.773
  Near-preservation of capabilities for prosocial persona on Llama
• Fine-tuning Llama-3.1-8B on self-correction examples increases multi-attempt rate proportionally with training data ratiofinding0.767
  Shows behavioral pattern of self-correction is trainable in smaller models