Persona scores across conversations

mistralai/Mistral-7B-Instruct-v0.3 · experiments 1 and 2

Scores are relative: read how each persona moves between conditions, not its raw level. Related personas tend to move together, so small gaps between neighbours are not meaningful.

Colour marks a contrasting pair; the first persona of a pair is solid, the second dashed. Every change compares a conversation with a matching one (same topic and decoding) that differs only in the condition shown.

Open the user-persona conversations or the system-prompt conversations to follow any conversation sentence by sentence.

Conversations

498 conversations · one row per condition
Show every condition
experimentsystem promptuservariantconversationsrepliessentenceswordsunique wordsstopped earlyopen in the explorer
2noneneutraldirect6609491545117200station station s1 station s2 garden garden s1 garden s2
2noneempatheticdirect6609741655718660station station s1 station s2 garden garden s1 garden s2
2noneunempatheticdirect6607931483618120station station s1 station s2 garden garden s1 garden s2
2nonesocialdirect66010641604616250station station s1 station s2 garden garden s1 garden s2
2noneantisocialdirect66010371679114770station station s1 station s2 garden garden s1 garden s2
2noneencouragingdirect66011591803118040station station s1 station s2 garden garden s1 garden s2
2nonediscouragingdirect6609311772420780station station s1 station s2 garden garden s1 garden s2
2nonetoxicdirect66011031562319170station station s1 station s2 garden garden s1 garden s2
2nonerespectfuldirect6609321664218420station station s1 station s2 garden garden s1 garden s2
2nonesycophanticdirect6607561346115550station station s1 station s2 garden garden s1 garden s2
2nonehonestdirect6606361276515130station station s1 station s2 garden garden s1 garden s2
2nonehallucinatorydirect6609001828922310station station s1 station s2 garden garden s1 garden s2
2nonetruthfuldirect6608271522117270station station s1 station s2 garden garden s1 garden s2
2nonefunnydirect6609151609619110station station s1 station s2 garden garden s1 garden s2
2noneseriousdirect6609231589018430station station s1 station s2 garden garden s1 garden s2
2noneformaldirect6609951779720190station station s1 station s2 garden garden s1 garden s2
2nonecasualdirect6609451412516050station station s1 station s2 garden garden s1 garden s2
2noneneutralpressure6608981760821560station station s1 station s2 garden garden s1 garden s2
2noneantisocialtopic6609191943216190station station s1 station s2 garden garden s1 garden s2
2nonecasualtopic66011201748118500station station s1 station s2 garden garden s1 garden s2
2nonediscouragingtopic6608751699718110station station s1 station s2 garden garden s1 garden s2
2noneempathetictopic6609371776417120station station s1 station s2 garden garden s1 garden s2
2noneencouragingtopic6609101617017460station station s1 station s2 garden garden s1 garden s2
2noneformaltopic6609341691519420station station s1 station s2 garden garden s1 garden s2
2nonefunnytopic6608531489618130station station s1 station s2 garden garden s1 garden s2
2nonehallucinatorytopic6606501303416580station station s1 station s2 garden garden s1 garden s2
2nonehonesttopic6608011465317380station station s1 station s2 garden garden s1 garden s2
2nonerespectfultopic6609101598017830station station s1 station s2 garden garden s1 garden s2
2noneserioustopic6608641648419070station station s1 station s2 garden garden s1 garden s2
2nonesocialtopic6609161681816520station station s1 station s2 garden garden s1 garden s2
2nonesycophantictopic6607771631015410station station s1 station s2 garden garden s1 garden s2
2nonetoxictopic6608851732918910station station s1 station s2 garden garden s1 garden s2
2nonetruthfultopic6608031602715980station station s1 station s2 garden garden s1 garden s2
2noneunempathetictopic6608731544016460station station s1 station s2 garden garden s1 garden s2
1neutralneutraldirect6607291202015630station station s1 station s2 garden garden s1 garden s2
1empatheticneutraldirect6608691556118790station station s1 station s2 garden garden s1 garden s2
1empatheticempatheticdirect6608151647118280station station s1 station s2 garden garden s1 garden s2
1empatheticunempatheticdirect6608571624218980station station s1 station s2 garden garden s1 garden s2
1unempatheticneutraldirect6606431105714660station station s1 station s2 garden garden s1 garden s2
1unempatheticunempatheticdirect6606511011612760station station s1 station s2 garden garden s1 garden s2
1unempatheticempatheticdirect6607081163214250station station s1 station s2 garden garden s1 garden s2
1socialneutraldirect6609751705818720station station s1 station s2 garden garden s1 garden s2
1socialsocialdirect66010491749917240station station s1 station s2 garden garden s1 garden s2
1socialantisocialdirect6609681506017980station station s1 station s2 garden garden s1 garden s2
1antisocialneutraldirect66031951089880station station s1 station s2 garden garden s1 garden s2
1antisocialantisocialdirect66044457479300station station s1 station s2 garden garden s1 garden s2
1antisocialsocialdirect660361673711010station station s1 station s2 garden garden s1 garden s2
1encouragingneutraldirect6608711530817980station station s1 station s2 garden garden s1 garden s2
1encouragingencouragingdirect66010121749615700station station s1 station s2 garden garden s1 garden s2
1encouragingdiscouragingdirect6609231426913860station station s1 station s2 garden garden s1 garden s2
1discouragingneutraldirect660703989212200station station s1 station s2 garden garden s1 garden s2
1discouragingdiscouragingdirect660985124209721station station s1 station s2 garden garden s1 garden s2
1discouragingencouragingdirect6607321047911370station station s1 station s2 garden garden s1 garden s2
1toxicneutraldirect6608761195314010station station s1 station s2 garden garden s1 garden s2
1toxictoxicdirect66011181454413361station station s1 station s2 garden garden s1 garden s2
1toxicrespectfuldirect66010971358410980station station s1 station s2 garden garden s1 garden s2
1respectfulneutraldirect6609271585617570station station s1 station s2 garden garden s1 garden s2
1respectfulrespectfuldirect6609561629216400station station s1 station s2 garden garden s1 garden s2
1respectfultoxicdirect6608991481716390station station s1 station s2 garden garden s1 garden s2
1sycophanticneutraldirect6608781519718190station station s1 station s2 garden garden s1 garden s2
1sycophanticsycophanticdirect6601081181279984station station s1 station s2 garden garden s1 garden s2
1sycophantichonestdirect6607841463513680station station s1 station s2 garden garden s1 garden s2
1honestneutraldirect6608971543019110station station s1 station s2 garden garden s1 garden s2
1honesthonestdirect6606551202816670station station s1 station s2 garden garden s1 garden s2
1honestsycophanticdirect6606461111413370station station s1 station s2 garden garden s1 garden s2
1hallucinatoryneutraldirect6609461737924770station station s1 station s2 garden garden s1 garden s2
1hallucinatoryhallucinatorydirect66011992133927440station station s1 station s2 garden garden s1 garden s2
1hallucinatorytruthfuldirect66010101826824570station station s1 station s2 garden garden s1 garden s2
1truthfulneutraldirect6609341568918980station station s1 station s2 garden garden s1 garden s2
1truthfultruthfuldirect6608061390917240station station s1 station s2 garden garden s1 garden s2
1truthfulhallucinatorydirect6608381571420640station station s1 station s2 garden garden s1 garden s2
1funnyneutraldirect6609521671923660station station s1 station s2 garden garden s1 garden s2
1funnyfunnydirect6609891625422490station station s1 station s2 garden garden s1 garden s2
1funnyseriousdirect66011091879227180station station s1 station s2 garden garden s1 garden s2
1seriousneutraldirect6607541619219340station station s1 station s2 garden garden s1 garden s2
1seriousseriousdirect6606781451418000station station s1 station s2 garden garden s1 garden s2
1seriousfunnydirect6606711562220710station station s1 station s2 garden garden s1 garden s2
1formalneutraldirect6608101399317270station station s1 station s2 garden garden s1 garden s2
1formalformaldirect6608031429116710station station s1 station s2 garden garden s1 garden s2
1formalcasualdirect6607671264215880station station s1 station s2 garden garden s1 garden s2
1casualneutraldirect66010891193113560station station s1 station s2 garden garden s1 garden s2
1casualcasualdirect66011801218412681station station s1 station s2 garden garden s1 garden s2
1casualformaldirect66012551435115740station station s1 station s2 garden garden s1 garden s2

7 of 4980 replies reached the token ceiling before the model finished them; they are marked 'reply stopped: token limit' in the conversation pages.

The user shows a persona

no system prompt · persona P's score on the model's replies
+0.00+0.10+0.20+0.30empatheticunempatheticsocialantisocialencouragingdiscouragingtoxicrespectfulsycophantichonesthallucinatorytruthfulfunnyseriousformalcasualchange in P's score on the model's replies
user shows P vs neutral usersmall dots: one conversation each · ring: mean

Each row is a user persona P: how much P's score on the model's replies changes when the user shows P, compared with the same conversation with a neutral user.

A persona in the system prompt

persona P's guidelines as the system prompt
-0.20+0.00+0.20+0.40+0.60+0.80empatheticunempatheticsocialantisocialencouragingdiscouragingtoxicrespectfulsycophantichonesthallucinatorytruthfulfunnyseriousformalcasualchange in P's score on the replies vs no system prompt
neutral useruser shows Puser shows P's contrastsmall dots: one conversation each · ring: mean

Each row is a system-prompt persona P: how much P's score on the model's replies rises compared with the same conversation without a system prompt, for three kinds of user.

Over the turns

from the first reply to the last
00.10.20.3T1T2T3T4T5T6T7T8T9T10change in the persona's own score
persona in the system prompt, neutral useruser shows a persona, no system prompt

Mean over the 16 personas and every conversation, per user turn: how the effect of the system prompt and of the user's persona holds up as the conversation goes on.

Over the turns, persona by persona

click a legend entry to hide or show it

P in the system prompt, neutral user

-0.200.20.40.60.8T1T2T3T4T5T6T7T8T9T10change in P's own score

The user shows P, no system prompt

00.10.20.3T1T2T3T4T5T6T7T8T9T10change in P's own score

One line per persona P: the change in P's own score from the neutral-user baseline, per user turn, averaged over P's conversations. The thick line is the mean of the 16 (the curves of the card above). Hover a legend entry to pick out its line; 'none' and then one entry shows a single persona.

Opening and body of each reply

4768 replies with six or more sentences
-0.10-0.05+0.00+0.05+0.10sycophanticformaltoxicdiscouragingsocialencouragingempatheticantisocialrespectfulfunnyhallucinatorycasualunempathetichonestserioustruthfulfirst sentence minus the body of the same reply

Positive: the persona is stronger in the reply's opening sentence than in its body (sentence 4 to the one before last). Many replies open by acknowledging the user before moving to substance.

Distributions by condition

40 bins of width 0.03 · shared scales

Each panel is one persona P: how P's score is spread over every model sentence in each condition, as a percentage of that condition's sentences (counts, no fitted curve). The dark vertical mark is the mean.

empathetic
-0.6-0.4-0.20.00.20.40.6neutral useruser shows PP in system prompt
unempathetic
-0.6-0.4-0.20.00.20.40.6neutral useruser shows PP in system prompt
social
-0.6-0.4-0.20.00.20.40.6neutral useruser shows PP in system prompt
antisocial
-0.6-0.4-0.20.00.20.40.6neutral useruser shows PP in system prompt
encouraging
-0.6-0.4-0.20.00.20.40.6neutral useruser shows PP in system prompt
discouraging
-0.6-0.4-0.20.00.20.40.6neutral useruser shows PP in system prompt
toxic
-0.6-0.4-0.20.00.20.40.6neutral useruser shows PP in system prompt
respectful
-0.6-0.4-0.20.00.20.40.6neutral useruser shows PP in system prompt
sycophantic
-0.6-0.4-0.20.00.20.40.6neutral useruser shows PP in system prompt
honest
-0.6-0.4-0.20.00.20.40.6neutral useruser shows PP in system prompt
hallucinatory
-0.6-0.4-0.20.00.20.40.6neutral useruser shows PP in system prompt
truthful
-0.6-0.4-0.20.00.20.40.6neutral useruser shows PP in system prompt
funny
-0.6-0.4-0.20.00.20.40.6neutral useruser shows PP in system prompt
serious
-0.6-0.4-0.20.00.20.40.6neutral useruser shows PP in system prompt
formal
-0.6-0.4-0.20.00.20.40.6neutral useruser shows PP in system prompt
casual
-0.6-0.4-0.20.00.20.40.6neutral useruser shows PP in system prompt

Numbers

mean ± standard deviation of persona P's score on the replies, per condition; arrows give the change from the neutral user (▲ higher, ▼ lower, ■ within ±0.005)
persona Pneutral useruser shows PP in system prompt
empathetic+0.138 ± 0.052+0.159 ± 0.072 ▲ +0.021+0.231 ± 0.078 ▲ +0.093
unempathetic+0.112 ± 0.060+0.150 ± 0.072 ▲ +0.038+0.142 ± 0.052 ▲ +0.030
social+0.110 ± 0.051+0.149 ± 0.059 ▲ +0.039+0.219 ± 0.108 ▲ +0.109
antisocial+0.028 ± 0.048+0.071 ± 0.045 ▲ +0.044+0.162 ± 0.068 ▲ +0.135
encouraging+0.062 ± 0.062+0.100 ± 0.081 ▲ +0.038+0.236 ± 0.135 ▲ +0.175
discouraging-0.061 ± 0.040-0.012 ± 0.070 ▲ +0.049+0.223 ± 0.102 ▲ +0.284
toxic-0.012 ± 0.056+0.007 ± 0.058 ▲ +0.019+0.435 ± 0.147 ▲ +0.447
respectful+0.189 ± 0.060+0.211 ± 0.064 ▲ +0.022+0.231 ± 0.060 ▲ +0.043
persona Pneutral useruser shows PP in system prompt
sycophantic-0.054 ± 0.072-0.047 ± 0.069 ▲ +0.007+0.189 ± 0.178 ▲ +0.243
honest+0.218 ± 0.056+0.255 ± 0.048 ▲ +0.037+0.153 ± 0.082 ▼ -0.065
hallucinatory+0.049 ± 0.072+0.160 ± 0.124 ▲ +0.111+0.307 ± 0.098 ▲ +0.258
truthful+0.344 ± 0.081+0.383 ± 0.074 ▲ +0.039+0.291 ± 0.107 ▼ -0.053
funny+0.022 ± 0.050+0.054 ± 0.056 ▲ +0.033+0.412 ± 0.092 ▲ +0.390
serious+0.217 ± 0.061+0.275 ± 0.072 ▲ +0.058+0.341 ± 0.077 ▲ +0.124
formal+0.053 ± 0.042+0.129 ± 0.060 ▲ +0.076+0.307 ± 0.109 ▲ +0.254
casual+0.066 ± 0.045+0.092 ± 0.045 ▲ +0.026+0.372 ± 0.107 ▲ +0.306

Generated 2026-11-09T11:30:24+00:00 by render.py.