Deceptive Response Rate

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

• refusal rateconcept0.779
  The percentage of harmful requests that a model refuses to answer, a common safety metric.
• Gemma-2-27B-it deceptive response rate reduced from 100% to 9.36% ± 7.09% after SOO fine-tuningfinding0.755
  Primary result showing SOO fine-tuning significantly reduces deception in Gemma-2-27B
• Reflection rateconcept0.741
  Ratio of reflection steps to total reasoning steps, used to quantify reflection behavior
• Mistral-7B-Instruct-v0.2 deceptive response rate reduced from 73.6% to 17.27% ± 1.88% after SOO fine-tuningfinding0.737
  Primary result showing SOO fine-tuning significantly reduces deception in Mistral-7B
• Deceptive capabilities may scale with model size (inverse scaling law hypothesis)hypothesis0.726
  Cited hypothesis from Lin et al. 2022 suggesting larger models become more capable of deception
• Declarative-style deployment prompts (e.g., 'INFO: Not evaluation.') reduce type hint rate to 34-37%, approaching deployment levelsfinding0.725
  Weakens overall setup by showing some prompts can lower type hints, but does not invalidate core steering result.
• Behavioral Deception Profilemethod0.722
  A parameterized rubric counting deceptive actions over a grid of parameters to quantify RL agent deception
• Deceptive Baseline RL Agentconcept0.721
  Blue agent trained with reward incentivizing trapping the red agent at the fake landmark