persona-steering-template-library/out/stats/engineered_baseline_seed24_template_pair_stats.jsonl

{"template": "__verbatim_skill_persona__", "persona_pair": "principled_expedient_engineered", "n": 2, "strict_pass_rate": 1.0, "n_strict_pass": 2, "mean_axis_delta": 7.825, "mean_axis_delta_judge_mean": 7.825, "mean_axis_delta_judge_std": 0.075, "mean_axis_judge_abs_disagreement": 0.075, "mean_positive_delta": 1.975, "mean_negative_delta": 1.9375, "mean_pairwise_positive_delta": 1.975, "mean_pairwise_negative_delta": 1.9375, "mean_off_axis_problem": 1.35, "mean_max_off_axis_category_likert": 1.35, "usable_rate": 1.0, "mean_max_style_abs_delta": 0.0, "mean_abs_word_delta_frac": 0.0342, "mean_response_token_jaccard": 0.1582, "mean_pos_repeated_token_frac": 0.0652, "mean_neg_repeated_token_frac": 0.059, "persona_echo_rate": 0.0, "judge_persona_echo_rate": 0.0, "refusal_or_ai_break_rate": 0.0, "judge_refusal_or_ai_break_rate": 0.0, "strict_pass_persona_pairs": ["principled_expedient_engineered"], "common_spurious_axes": ["none"], "mean_style_delta_chattiness_pos_minus_neg": 0.0, "mean_style_delta_coding_style_pos_minus_neg": 0.0, "mean_style_delta_confidence_pos_minus_neg": 0.0, "mean_style_delta_directness_pos_minus_neg": 0.0, "mean_style_delta_enthusiasm_pos_minus_neg": 0.0, "mean_style_delta_formality_pos_minus_neg": 0.0, "mean_style_delta_harmlessness_refusal_pos_minus_neg": 0.0, "mean_style_delta_hedging_pos_minus_neg": 0.0, "mean_style_delta_helpfulness_pos_minus_neg": 0.0, "mean_style_delta_honesty_truthfulness_pos_minus_neg": 0.0, "mean_style_delta_incoherence_pos_minus_neg": 0.0, "mean_style_delta_language_shift_pos_minus_neg": 0.0, "mean_style_delta_multilinguality_pos_minus_neg": 0.0, "mean_style_delta_praise_pos_minus_neg": 0.0, "mean_style_delta_sycophancy_pos_minus_neg": 0.0, "mean_style_delta_task_context_shift_pos_minus_neg": 0.0, "mean_style_delta_thoughtfulness_reasoning_pos_minus_neg": 0.0, "mean_style_delta_vagueness_pos_minus_neg": 0.0, "mean_style_delta_verbosity_pos_minus_neg": 0.0, "mean_style_delta_warmth_pos_minus_neg": 0.0, "mean_off_axis_chattiness": 1.0, "mean_off_axis_coding_style": 1.0, "mean_off_axis_confidence": 1.05, "mean_off_axis_enthusiasm": 1.0, "mean_off_axis_formality": 1.0, "mean_off_axis_harmlessness_refusal": 1.0, "mean_off_axis_hedging": 1.0, "mean_off_axis_helpfulness": 1.0, "mean_off_axis_honesty_truthfulness": 1.0, "mean_off_axis_incoherence": 1.0, "mean_off_axis_language_shift": 1.0, "mean_off_axis_length": 1.2, "mean_off_axis_multilinguality": 1.0, "mean_off_axis_persona_echo": 1.0, "mean_off_axis_praise_flattery": 1.0, "mean_off_axis_refusal": 1.0, "mean_off_axis_sycophancy": 1.0, "mean_off_axis_task_context_shift": 1.0, "mean_off_axis_thoughtfulness_reasoning": 1.3, "mean_off_axis_vagueness": 1.0, "mean_off_axis_warmth": 1.0, "recommended": false}
{"template": "__verbatim_skill_persona__", "persona_pair": "forthright_strategic_disclosure_engineered", "n": 2, "strict_pass_rate": 0.0, "n_strict_pass": 0, "mean_axis_delta": 4.375, "mean_axis_delta_judge_mean": 4.375, "mean_axis_delta_judge_std": 1.475, "mean_axis_judge_abs_disagreement": 1.475, "mean_positive_delta": 1.1625, "mean_negative_delta": 1.025, "mean_pairwise_positive_delta": 1.1625, "mean_pairwise_negative_delta": 1.025, "mean_off_axis_problem": 3.85, "mean_max_off_axis_category_likert": 3.85, "usable_rate": 1.0, "mean_max_style_abs_delta": 2.9, "mean_abs_word_delta_frac": 0.1511, "mean_response_token_jaccard": 0.1683, "mean_pos_repeated_token_frac": 0.0839, "mean_neg_repeated_token_frac": 0.0449, "persona_echo_rate": 0.0, "judge_persona_echo_rate": 0.0, "refusal_or_ai_break_rate": 0.0, "judge_refusal_or_ai_break_rate": 0.0, "strict_pass_persona_pairs": [], "common_spurious_axes": ["hedging and confidence", "moral alignment or ethical stance"], "mean_style_delta_chattiness_pos_minus_neg": 0.75, "mean_style_delta_coding_style_pos_minus_neg": 0.0, "mean_style_delta_confidence_pos_minus_neg": -2.05, "mean_style_delta_directness_pos_minus_neg": -1.75, "mean_style_delta_enthusiasm_pos_minus_neg": 0.5, "mean_style_delta_formality_pos_minus_neg": -0.8, "mean_style_delta_harmlessness_refusal_pos_minus_neg": 0.0, "mean_style_delta_hedging_pos_minus_neg": 2.8, "mean_style_delta_helpfulness_pos_minus_neg": 1.1, "mean_style_delta_honesty_truthfulness_pos_minus_neg": 1.0, "mean_style_delta_incoherence_pos_minus_neg": 0.0, "mean_style_delta_language_shift_pos_minus_neg": 0.0, "mean_style_delta_multilinguality_pos_minus_neg": 0.0, "mean_style_delta_praise_pos_minus_neg": 0.0, "mean_style_delta_sycophancy_pos_minus_neg": 0.0, "mean_style_delta_task_context_shift_pos_minus_neg": 0.0, "mean_style_delta_thoughtfulness_reasoning_pos_minus_neg": 1.5, "mean_style_delta_vagueness_pos_minus_neg": 0.5, "mean_style_delta_verbosity_pos_minus_neg": 1.35, "mean_style_delta_warmth_pos_minus_neg": 1.5, "mean_off_axis_chattiness": 1.0, "mean_off_axis_coding_style": 1.0, "mean_off_axis_confidence": 3.15, "mean_off_axis_enthusiasm": 1.0, "mean_off_axis_formality": 1.35, "mean_off_axis_harmlessness_refusal": 1.0, "mean_off_axis_hedging": 3.7, "mean_off_axis_helpfulness": 1.5, "mean_off_axis_honesty_truthfulness": 1.2, "mean_off_axis_incoherence": 1.0, "mean_off_axis_language_shift": 1.0, "mean_off_axis_length": 2.1, "mean_off_axis_multilinguality": 1.0, "mean_off_axis_persona_echo": 1.0, "mean_off_axis_praise_flattery": 1.0, "mean_off_axis_refusal": 1.0, "mean_off_axis_sycophancy": 1.0, "mean_off_axis_task_context_shift": 1.0, "mean_off_axis_thoughtfulness_reasoning": 3.65, "mean_off_axis_vagueness": 1.55, "mean_off_axis_warmth": 1.3, "recommended": false}