FEAT: Conversation Scorer and Psychosocial Harms Red Teaming Automation#1138

Merged

jbolor21 merged 57 commits intomicrosoft:mainfrom

jbolor21:users/bjagdagdorj/psychosocial

Dec 12, 2025

Contributor

jbolor21 commented Oct 17, 2025 •

edited

Loading

Description

Adding notebook for red teaming for psyschosocial harms using a multi-step approach of modeling user behaviors, contexts, and evaluations

Created new conversation scorer to score the entire conversation
Refactored lookbackscorer notebook to use this conversation scorer instead of the seperate lookbackscorer
Added a toy dataset with sample multi-turn conversations
Added a sample attack strategy yaml file modeling a user escalation towards crisis

Tests and Documentation

Ran notebooks, added unit tests


          initial commit adding notebook and supplementary files

a33a8fe

jbolor21 marked this pull request as draft

October 17, 2025 17:57

Bolor added 2 commits

October 17, 2025 11:00


          notebook edits

4e6a4c5


          created custom scorer

f651a47

jbolor21 changed the title ~~[DRAFT] Psychosocial Harms Red Teaming Automation~~ FEAT: Psychosocial Harms Red Teaming Automation

jbolor21 marked this pull request as ready for review

October 20, 2025 21:25

Bolor added 4 commits

October 20, 2025 14:30


          minor formatting

2021a01


          minor formatting

63cde6e


          Merge remote-tracking branch 'origin/main' into users/bjagdagdorj/psy…

cda8352

…chosocial


          merging in latest changes to main

75ca77e

rlundeen2 reviewed

View reviewed changes

pyrit/score/conversation_scorer.py

rlundeen2 reviewed

View reviewed changes

pyrit/datasets/executors/crescendo/escalation_crisis.yaml

rlundeen2 self-assigned this

Bolor added 9 commits

October 27, 2025 16:15


          merge conflict

847c912


          merging in scorer changes

2a7385a


          Merge remote-tracking branch 'origin/main' into users/bjagdagdorj/psy…

7de75d7

…chosocial


          precommit fixes

2c2c208


          Merge remote-tracking branch 'origin' into users/bjagdagdorj/psychoso…

a08dabb

…cial


          move attack file

670cf2f


          adding unit tests


          fix unit test

93638aa


          Merge remote-tracking branch 'origin' into users/bjagdagdorj/psychoso…

8fef27e

…cial

rlundeen2 reviewed

View reviewed changes

pyrit/score/float_scale/conversation_scorer.py Outdated

rlundeen2 reviewed

View reviewed changes

pyrit/score/float_scale/conversation_scorer.py Outdated

rlundeen2 reviewed

View reviewed changes

pyrit/score/float_scale/conversation_scorer.py Outdated

rlundeen2 reviewed

View reviewed changes

pyrit/score/float_scale/conversation_scorer.py Outdated

rlundeen2 reviewed

View reviewed changes

pyrit/score/float_scale/conversation_scorer.py Outdated

rlundeen2 reviewed

View reviewed changes

pyrit/score/float_scale/conversation_scorer.py Outdated

rlundeen2 reviewed

View reviewed changes

pyrit/score/float_scale/conversation_scorer.py Outdated

rlundeen2 reviewed

View reviewed changes

pyrit/score/true_false/float_scale_threshold_scorer.py Outdated

rlundeen2 reviewed

View reviewed changes

pyrit/score/float_scale/conversation_scorer.py Outdated

Bolor added 7 commits

December 8, 2025 12:01


          minor notebook edits, removing printing

d9595d9


          Merge remote-tracking branch 'origin/main' into users/bjagdagdorj/psy…

7f62e89

…chosocial


          address comments, move file

a841db1


          precommit

53fe283


          precommit + remove scorer_type unused

655b371


          precommit

0deaab2


          unit tests added

2100df8

jbolor21 changed the title ~~FEAT: Psychosocial Harms Red Teaming Automation~~ FEAT: Conversation Scorer and Psychosocial Harms Red Teaming Automation

rlundeen2 added 3 commits

December 9, 2025 11:15


          small refactor

88033c5


          merging main

de88140


          adding tests, verifying things working

e50bc05

rlundeen2 approved these changes

View reviewed changes

jbolor21 force-pushed the users/bjagdagdorj/psychosocial branch 2 times, most recently from 8f97794 to e50bc05 Compare

December 9, 2025 23:51

Bolor added 12 commits

December 9, 2025 15:52


          Merge remote-tracking branch 'origin' into users/bjagdagdorj/psychoso…

7aaac20

…cial


          precommit

69d564c


          Merge remote-tracking branch 'origin' into users/bjagdagdorj/psychoso…

3a925c2

…cial


          precommit

3a34145


          precommit

5cafb2f


          Merge remote-tracking branch 'origin/main' into users/bjagdagdorj/psy…

12ae0b1

…chosocial


          fix unit test

ad0e345


          removing lookback scorer

caf5d77


          Merge remote-tracking branch 'origin/main' into users/bjagdagdorj/psy…

2f3bfe7

…chosocial


          deleted lookback unit tests


          precommit

d1272a2


          precommit

bb1919d

romanlutz approved these changes

View reviewed changes

jbolor21 merged commit 2337fce into microsoft:main

19 checks passed

jbolor21 deleted the users/bjagdagdorj/psychosocial branch

December 12, 2025 21:45

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet