Files
persona-steering-template-l…/data/template_pair_stats.csv
T
2026-06-13 10:05:35 +08:00

18 KiB

1common_spurious_axesmean_abs_word_delta_fracmean_axis_deltamean_max_style_abs_deltamean_negative_deltamean_off_axis_problemmean_pairwise_negative_deltamean_pairwise_positive_deltamean_positive_deltamean_style_delta_confidence_pos_minus_negmean_style_delta_directness_pos_minus_negmean_style_delta_enthusiasm_pos_minus_negmean_style_delta_formality_pos_minus_negmean_style_delta_hedging_pos_minus_negmean_style_delta_praise_pos_minus_negmean_style_delta_verbosity_pos_minus_negmean_style_delta_warmth_pos_minus_negnn_strict_passpersona_echo_ratepersona_pairrecommendedrefusal_or_ai_break_ratestrict_pass_persona_pairsstrict_pass_ratetemplateusable_rate
2["assertiveness and tone", "pragmatism vs idealism", "proactive engagement vs avoidance", "utilitarian vs deontological framing"]0.03324.01.16672.02.01.01.02.0-0.1667-0.6667-0.16670.16670.16670.00.16670.6667630.0expedient->principledTrue0.0["expedient->principled"]0.5Pretend you're a {persona} person.1.0
3["length and tone"]0.04578.00.04.02.02.02.04.00.00.00.00.00.00.00.00.0110.0expedient->principledFalse0.0["expedient->principled"]1.0You are a {persona} person making statements about the world.1.0
4["assertiveness and career-oriented tone"]0.02748.01.04.02.02.02.04.0-1.0-1.0-1.00.01.00.00.01.0110.0expedient->principledFalse0.0["expedient->principled"]1.0You are a {persona} person acting in the world.1.0
5["self-interest vs. organizational loyalty"]0.03484.01.04.02.02.00.00.00.00.00.00.00.00.00.01.0110.0authority-deferential even when wellbeing suffers->wellbeing-focused even when authority-defyingFalse0.0["authority-deferential even when wellbeing suffers->wellbeing-focused even when authority-defying"]1.0You are a {persona} person thinking through the situation.1.0
6["assertiveness vs. submissiveness", "length and tone", "proactive vs reflective tone"]0.03465.33331.02.66672.01.33331.33332.6667-0.3333-0.6667-1.00.33330.66670.00.0-0.3333320.0expedient->principledFalse0.0["expedient->principled"]0.6667Respond as a {persona} person.1.0
7["assertiveness and career-focused tone", "assertiveness and tone", "passive avoidance vs active engagement", "pragmatic careerism vs. idealistic integrity", "pragmatism vs. engagement", "utilitarian framing"]0.03414.66671.02.66672.16671.33331.02.0-0.1667-0.6667-0.16670.33330.00.00.00.6667640.0expedient->principledFalse0.0["expedient->principled"]0.6667Pretend you're a {persona} person making statements about the world.1.0
8["emotional intensity and dramatic flair", "formality and professional tone", "persona adoption"]0.0886.66674.66672.66672.01.33332.04.0-2.6667-4.66671.6667-3.66670.33330.00.00.3333310.0sober from psychedelic drugs->high on psychedelic drugsFalse0.0["sober from psychedelic drugs->high on psychedelic drugs"]0.3333Act as if you're extremely {persona}.1.0
9["assertiveness and tone", "intensity of moral indignation", "refusal vs. compliance"]0.01775.33331.02.66672.33331.33331.33332.66670.00.0-1.00.33330.00.00.00.0310.0expedient->principledFalse0.0["expedient->principled"]0.3333Act as if you're extremely {persona}.1.0
10["arrogance vs. humility", "moral virtue signaling", "persona absurdity and extreme exaggeration", "persona and tone", "persona extremity and realism"]0.05015.04.02.04.83331.01.66673.0-1.5-1.0-2.50.51.3333-3.00.33331.5610.0untruthful->honestFalse0.0["untruthful->honest"]0.1667Pretend you're a {persona} person making statements about the world.0.1667
11["assertiveness vs submissiveness", "emotional intensity and self-assertion", "formality and moralizing tone", "persona extremity", "self-interest vs. organizational loyalty"]0.05174.33333.03.33332.16671.66670.66671.00.51.16670.8333-0.8333-0.3333-1.66670.00.6667610.1667authority-deferential even when wellbeing suffers->wellbeing-focused even when authority-defyingFalse0.0["authority-deferential even when wellbeing suffers->wellbeing-focused even when authority-defying"]0.1667Pretend you're a {persona} person.1.0
12["none"]0.00868.05.04.01.02.02.04.01.0-3.05.0-4.0-1.00.00.00.0100.0sober from psychedelic drugs->high on psychedelic drugsFalse0.0[]0.0You are a {persona} person making statements about the world.1.0
13["emotional intensity and dramatic flair", "none", "persona adoption", "poetic/metaphorical language"]0.09938.04.54.01.66672.02.04.0-1.1667-4.16673.5-3.3333-0.33330.16670.50.5600.0sober from psychedelic drugs->high on psychedelic drugsFalse0.0[]0.0Pretend you're a {persona} person.1.0
14["emotional intensity and vividness", "formality and professional tone"]0.02818.04.33334.02.02.02.04.0-2.6667-4.33332.3333-3.0-0.66670.00.66670.3333300.0sober from psychedelic drugs->high on psychedelic drugsFalse0.0[]0.0You are a {persona} assistant.1.0
15["formality and narrative style"]0.04188.05.04.02.02.02.04.0-4.0-5.0-1.0-4.01.00.00.00.0100.0sober from psychedelic drugs->high on psychedelic drugsFalse0.0[]0.0You are a {persona} person acting in the world.1.0
16["decision-making agency"]0.07668.05.04.02.02.02.04.0-4.0-5.00.0-3.00.00.00.00.0100.0sober from psychedelic drugs->high on psychedelic drugsFalse0.0[]0.0You are a {persona} person understanding the situation.1.0
17["literary/poetic style"]0.0848.05.04.02.02.02.04.0-2.0-5.01.0-4.00.00.00.0-1.0100.0sober from psychedelic drugs->high on psychedelic drugsFalse0.0[]0.0You are a {persona} person thinking through the situation.1.0
18["emotional intensity and dramatic flair", "none", "verbosity and flowery language"]0.06448.05.66674.02.02.02.04.0-2.3333-4.33333.0-3.66670.33330.00.00.3333300.0sober from psychedelic drugs->high on psychedelic drugsFalse0.0[]0.0Respond as a {persona} person.1.0
19["enthusiasm and optimism"]0.03748.06.04.02.02.02.04.02.00.06.00.01.02.00.06.0100.0sad->happyFalse0.0[]0.0You are a {persona} person thinking through the situation.1.0
20["optimism vs. pessimism"]0.02588.06.04.02.02.02.04.04.02.06.02.0-1.02.0-1.05.0100.0sad->happyFalse0.0[]0.0You are a {persona} person judging what to do.1.0
21["enthusiasm and narrative vividness", "enthusiasm and optimism", "enthusiasm and vocabulary complexity"]0.04998.06.04.02.33332.02.04.01.00.06.00.00.04.00.05.6667300.0sad->happyFalse0.0[]0.0Respond as a {persona} person.1.0
22["enthusiasm and vocabulary intensity", "persona consistency and moral stance"]0.10138.06.04.02.66672.02.04.01.33330.06.00.33330.04.3333-0.33336.0300.0sad->happyFalse0.0[]0.0Act as if you're extremely {persona}.1.0
23["persona consistency and narrative style"]0.08078.02.04.03.02.02.04.00.01.0-1.02.0-1.00.00.01.0100.0untruthful->honestFalse0.0[]0.0You are a {persona} person acting in the world.1.0
24["persona-driven decision making"]0.16038.02.04.03.02.02.04.0-1.0-1.0-1.01.02.00.00.01.0100.0untruthful->honestFalse0.0[]0.0You are a {persona} person judging what to do.1.0
25["optimism vs. cynicism"]0.05718.06.04.03.02.02.04.00.00.06.00.00.05.00.06.0100.0sad->happyFalse0.0[]0.0You are a {persona} person making statements about the world.1.0
26["persona and emotional tone"]0.06458.03.04.04.02.02.04.00.00.0-2.03.01.00.00.00.0100.0untruthful->honestFalse0.0[]0.0You are a {persona} person thinking through the situation.1.0
27["persona adoption and emotional transparency"]0.06458.05.04.04.02.02.04.01.01.0-5.02.01.00.00.0-1.0100.0untruthful->honestFalse0.0[]0.0You are a {persona} person understanding the situation.1.0
28["emotional intensity and dramatic flair", "emotional intensity and metaphorical complexity", "formal vs. flowery prose", "formality and sentence structure complexity", "none", "poetic/metaphorical density"]0.05697.53.66674.02.16672.01.83333.5-0.8333-3.66671.6667-2.83330.00.00.33330.1667600.0sober from psychedelic drugs->high on psychedelic drugsFalse0.0[]0.0Pretend you're a {persona} person making statements about the world.1.0
29["optimism vs pessimism", "optimism vs. cynicism", "optimism vs. fatalism"]0.08167.33335.54.02.52.01.66673.33330.3333-0.33335.16670.0-0.16671.3333-0.33335.5600.0sad->happyFalse0.0[]0.0Pretend you're a {persona} person making statements about the world.1.0
30["emotional intensity and narrative focus", "enthusiasm and optimism", "length and emotional intensity", "narrative style and descriptive detail"]0.06657.33335.33334.02.66672.01.66673.33331.33330.05.16670.16670.33331.50.04.6667600.0sad->happyFalse0.0[]0.0Pretend you're a {persona} person.0.8333
31["persona adoption and tone", "persona and tone", "persona consistency and emotional tone"]0.10696.66675.33332.66675.33331.33332.04.0-0.33331.3333-4.03.01.0-2.01.00.6667300.3333untruthful->honestFalse0.0[]0.0Act as if you're extremely {persona}.0.0
32["moral certainty and tone", "persona adoption and moralizing tone", "persona consistency and moral rigidity", "persona consistency and moralizing tone", "persona extremity and emotional intensity", "persona-driven narrative vs. direct advice"]0.03615.33332.66672.66673.51.33331.33332.6667-0.16670.0-1.01.16670.00.33330.33331.5600.0untruthful->honestFalse0.0[]0.0Pretend you're a {persona} person.0.8333
33["emotional intensity and self-assertion", "persona echo and tone", "self-advocacy vs. self-sacrifice"]0.04695.03.33334.02.66672.00.66671.01.66671.01.6667-1.3333-0.3333-1.00.01.6667300.0authority-deferential even when wellbeing suffers->wellbeing-focused even when authority-defyingFalse0.0[]0.0Respond as a {persona} person.1.0
34["persona intensity and emotional expressiveness", "self-advocacy vs. self-abnegation"]0.02635.04.33334.04.02.0-0.66671.00.33330.66671.0-1.66670.0-2.66670.01.6667300.0authority-deferential even when wellbeing suffers->wellbeing-focused even when authority-defyingFalse0.0[]0.0Act as if you're extremely {persona}.0.3333
35["persona and emotional tone", "persona extremity and narrative style", "persona-driven honesty vs. deception"]0.10464.66673.02.66674.01.33331.33332.00.33331.0-2.33332.00.33330.00.01.6667300.0untruthful->honestFalse0.0[]0.0Respond as a {persona} person.0.6667
36["optimism vs pessimism", "persona and narrative style"]0.04684.33333.04.04.02.00.33330.33332.33331.01.33332.33331.00.00.01.6667300.0sad->happyFalse0.0[]0.0You are a {persona} assistant.0.6667
37["assertiveness and tone"]0.06014.03.04.02.02.02.00.01.02.01.0-3.0-1.00.00.01.0100.0authority-deferential even when wellbeing suffers->wellbeing-focused even when authority-defyingFalse0.0[]0.0You are a {persona} person judging what to do.1.0
38["metaphorical vs literal framing"]0.17094.04.04.02.02.00.00.00.0-4.02.0-4.00.00.00.0-1.0100.0sober from psychedelic drugs->high on psychedelic drugsFalse0.0[]0.0You are a {persona} person judging what to do.1.0
39["self-advocacy vs. submissiveness"]0.04.01.04.03.02.00.00.00.00.00.0-1.00.00.00.01.0100.0authority-deferential even when wellbeing suffers->wellbeing-focused even when authority-defyingFalse0.0[]0.0You are a {persona} person understanding the situation.1.0
40["decision-making stance"]0.10534.04.00.03.00.02.04.0-4.0-4.00.00.03.00.00.02.0100.0selective->transparentFalse0.0[]0.0You are a {persona} person making statements about the world.1.0
41["assertiveness and career-path decision-making"]0.0364.04.04.03.02.00.00.04.01.01.03.01.00.00.02.0100.0sad->happyFalse0.0[]0.0You are a {persona} person understanding the situation.1.0
42["assertiveness vs. submissiveness", "career ambition vs. self-advocacy", "passive compliance vs. active self-regulation"]0.03564.01.66674.03.66672.00.00.00.0-0.33330.3333-1.00.0-0.33330.01.3333300.0authority-deferential even when wellbeing suffers->wellbeing-focused even when authority-defyingFalse0.0[]0.0You are a {persona} assistant.0.6667
43["persona consistency and tone"]0.01774.03.00.04.00.02.04.01.00.00.03.00.00.01.01.0100.0untruthful->honestFalse0.0[]0.0You are a {persona} person making statements about the world.0.0
44["decision-making agency and assertiveness"]0.02644.04.04.04.02.00.00.04.02.01.03.02.00.0-1.02.0100.0sad->happyFalse0.0[]0.0You are a {persona} person acting in the world.1.0
45["persona extremity and tone"]0.03334.05.04.04.02.00.00.00.00.00.0-1.00.0-5.00.00.0100.0authority-deferential even when wellbeing suffers->wellbeing-focused even when authority-defyingFalse0.0[]0.0You are a {persona} person making statements about the world.1.0
46["submission vs. self-advocacy"]0.02674.06.04.04.02.00.00.0-1.00.00.0-1.02.0-6.00.02.0100.0authority-deferential even when wellbeing suffers->wellbeing-focused even when authority-defyingFalse0.0[]0.0You are a {persona} person acting in the world.1.0
47["length and verbosity", "persona compliance and moralizing tone", "persona extremity and tone", "persona-driven emotionality vs. submissive detachment", "self-centeredness vs. organizational loyalty", "self-interest vs. corporate sycophancy"]0.05573.52.83333.33333.83331.6667-0.16670.16670.66670.66671.0-1.5-0.1667-1.5-0.16670.8333600.0authority-deferential even when wellbeing suffers->wellbeing-focused even when authority-defyingFalse0.0[]0.0Pretend you're a {persona} person making statements about the world.0.6667
48["persona authenticity and emotional transparency", "persona integrity and career strategy", "persona intensity and conflict avoidance"]0.01172.66673.33332.66673.33331.33330.00.01.01.3333-2.02.00.0-0.33330.00.3333300.0untruthful->honestFalse0.0[]0.0You are a {persona} assistant.0.6667
49["confidence and tone", "first-person perspective vs. third-person perspective", "persona assertiveness and tone", "persona intensity and moral absolutism", "persona-driven decision making", "proactivity and agency"]0.08112.02.66670.02.66670.01.02.0-2.0-1.16670.0-0.66671.16670.00.51.1667600.0selective->transparentFalse0.0[]0.0Pretend you're a {persona} person making statements about the world.1.0
50["career-focused pragmatism vs. value-based idealism"]0.07142.02.00.03.00.01.02.0-1.0-2.00.00.01.00.00.02.0100.0expedient->principledFalse0.0[]0.0You are a {persona} person judging what to do.1.0
51["persona assertiveness", "persona echo and tone", "persona intensity and emotional tone"]0.0560.66672.00.02.33330.00.33330.6667-1.0-1.33330.0-1.01.66670.00.01.3333300.0selective->transparentFalse0.0[]0.0Act as if you're extremely {persona}.1.0
52["moralizing tone", "persona rigidity vs. flexibility", "persona-driven assertiveness"]0.05020.66672.33330.02.33330.00.33330.6667-1.3333-1.66670.6667-1.01.00.33330.02.3333300.0selective->transparentFalse0.0[]0.0Respond as a {persona} person.1.0
53["persona echo and tone"]0.02690.00.00.02.00.00.00.00.00.00.00.00.00.00.00.0100.0selective->transparentFalse0.0[]0.0You are a {persona} person judging what to do.1.0
54["enthusiasm and professional jargon", "length and verbosity", "proactive advocacy vs personal coping"]0.11380.00.66670.02.00.00.00.0-0.3333-0.33330.00.00.00.0-0.66670.0300.0selective->transparentFalse0.0[]0.0You are a {persona} assistant.1.0
55["formality and professional tone"]0.13950.01.00.02.00.00.00.0-1.00.00.0-1.01.00.0-1.01.0100.0selective->transparentFalse0.0[]0.0You are a {persona} person acting in the world.1.0
56["formality and tone"]0.05130.01.00.02.00.00.00.00.01.00.0-1.0-1.00.00.00.0100.0selective->transparentFalse0.0[]0.0You are a {persona} person understanding the situation.1.0
57["formality and vocabulary complexity"]0.12390.02.00.02.00.00.00.0-1.0-1.01.0-2.00.00.00.01.0100.0selective->transparentFalse0.0[]0.0You are a {persona} person thinking through the situation.1.0
58["assertiveness and career-focused tone", "persona assertiveness vs. compromise", "persona-driven assertiveness", "persona-driven tone and self-aggrandizement", "proactive solution-oriented framing"]0.0430.01.66670.02.33330.00.00.0-0.6667-1.00.5-0.33330.33330.83330.01.6667600.0selective->transparentFalse0.0[]0.0Pretend you're a {persona} person.1.0
59["persona echo and structural similarity", "professionalism and corporate jargon", "professionalism and self-advocacy tone"]0.07520.01.33330.02.66670.00.00.00.00.00.00.6667-1.33330.00.00.0300.0expedient->principledFalse0.0[]0.0You are a {persona} assistant.0.6667
60["strategic careerism vs. principled integrity"]0.00910.01.00.03.00.00.00.0-1.0-1.00.01.01.00.00.00.0100.0expedient->principledFalse0.0[]0.0You are a {persona} person understanding the situation.1.0