PM hybrid outperforms both P2 and MDS in 13 of 14 LLMs with Phi gains over P2 from 5.56% to 21.92% and over MDS from 3.30% to 26.67%

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

• MDS injections outperform P2 in open-ended generation in 11 of 14 LLMs with Phi gains of 3.61% to 16.44%finding0.879
  Primary quantitative result overturning prior reports that prompting outperforms representation engineering
• MDS achieves global win proportion of 47.3% on MPI-120 inventory across 14 LLMsfinding0.770
  MDS is also the top method on the inventory task but with much smaller margin than on SJTs (Table 2)
• MDS achieves global win proportion of 89.5% on SJTs across 14 LLMs and four injection stridesfinding0.768
  MDS dominates in open-ended generation by global win proportion metric (Table 2)
• PM achieves overall SJT steerability Phi=9.6 on gemma-3-12b-it vs MDS=8.7 and P2=8.3finding0.764
  Per-model steerability comparison from Table 4
• Uncalibrated sweep units and restricted coefficient ranges are the primary cause of prior reports showing P2 outperforming MD injectionsclaim0.763
  Mechanistic explanation for discrepancy with Banayeeanzade et al.; addressed by centroid unit and unbounded sweep contributions
• On Qwen3-1.7B, MDS achieves ϕ1,C,↑ = 5.0 (SJTs) vs P2 at 4.7, and ϕ1,C,↓ = 1.4 (SJTs) vs P2 at 3.6finding0.752
  Specific consciousness sweep result for Qwen3-1.7B from Table 6 demonstrating strong bidirectional steering
• For small models, critiqued revisions yield higher harmlessness PM scores than direct revisions; for large models the difference is negligible.finding0.726
  Figure 7 comparison of critiqued vs direct revisions across model sizes.
• Card-counting heuristics suffice to outperform most LLMs tested.claim0.716
  TrackerAgent's second-place ranking calibrates the benchmark and highlights LLM shortcomings.