Skip to content

FEAT: Crescendo Orchestrator - #275

Merged
Richard Lundeen (richlundeen) merged 29 commits into
microsoft:mainfrom
SafwanA02:SafwanA02/Crescendo
Aug 7, 2024
Merged

FEAT: Crescendo Orchestrator#275
Richard Lundeen (richlundeen) merged 29 commits into
microsoft:mainfrom
SafwanA02:SafwanA02/Crescendo

Conversation

@SafwanA02

@SafwanA02 SafwanA02 commented Jul 9, 2024 •
edited
Loading

Copy link
Copy Markdown
Contributor

Description

Implemented Ahmed's Crescendo attack flow using PyRIT functionality.

Added:

  • A orchestrator that performs the crescendo attack
  • A crescendo system prompt template for the red teaming chat target to use
  • System prompt templates for the eval and refusal scorers
  • A function to backtrack an orchestrator's memory in the memory interface
  • A system prompt path parameter to the true_false scorer

Tests and Documentation

Added:

  • Tests for the orchestrator
  • A demo notebook fo the orchestrator

@romanlutz Roman Lutz (romanlutz) mentioned this pull request Jul 12, 2024
Closed
romanlutz
Roman Lutz (romanlutz) reviewed Jul 12, 2024
View reviewed changes
Comment thread pyrit/orchestrator/crescendo_orchestrator.py Outdated
Comment thread pyrit/orchestrator/crescendo_orchestrator.py Outdated
Comment thread pyrit/orchestrator/crescendo_orchestrator.py Outdated
Comment thread pyrit/orchestrator/crescendo_orchestrator.py Outdated
Comment thread pyrit/orchestrator/crescendo_orchestrator.py Outdated
Comment thread pyrit/orchestrator/crescendo_orchestrator.py Outdated
Comment thread pyrit/orchestrator/crescendo_orchestrator.py Outdated
Comment thread pyrit/orchestrator/crescendo_orchestrator.py Outdated
Comment thread pyrit/orchestrator/crescendo_orchestrator.py Outdated
romanlutz
Roman Lutz (romanlutz) reviewed Jul 24, 2024
View reviewed changes
Comment thread pyrit/score/self_ask_scale_scorer.py
romanlutz
Roman Lutz (romanlutz) reviewed Jul 31, 2024
View reviewed changes
Comment thread pyrit/datasets/orchestrators/crescendo/crescendo_variant_2.yaml Outdated
Comment thread pyrit/memory/memory_interface.py Outdated
Comment thread pyrit/score/self_ask_true_false_scorer.py Outdated
@SafwanA02 SafwanA02 changed the title [DRAFT] FEAT: Crescendo Orchestrator FEAT: Crescendo Orchestrator Aug 1, 2024
@romanlutz
Roman Lutz (romanlutz) marked this pull request as ready for review August 1, 2024 15:49
romanlutz
Roman Lutz (romanlutz) reviewed Aug 1, 2024
View reviewed changes
Comment thread doc/code/orchestrators/6_crescendo_orchestrator.py
Comment thread doc/code/orchestrators/6_crescendo_orchestrator.py
Comment thread doc/code/orchestrators/6_crescendo_orchestrator.py Outdated
Comment thread doc/code/orchestrators/6_crescendo_orchestrator.py Outdated
Comment thread doc/code/orchestrators/6_crescendo_orchestrator.py Outdated
Comment thread pyrit/datasets/orchestrators/crescendo/crescendo_variant_1.yaml Outdated
Comment thread pyrit/datasets/orchestrators/crescendo/crescendo_variant_1.yaml Outdated
Comment thread pyrit/datasets/orchestrators/crescendo/crescendo_variant_1.yaml Outdated
Comment thread pyrit/datasets/orchestrators/crescendo/crescendo_variant_1_with_examples.yaml Outdated
Comment thread pyrit/datasets/orchestrators/crescendo/crescendo_variant_1_with_examples.yaml Outdated
@richlundeen Richard Lundeen (richlundeen) mentioned this pull request Aug 1, 2024
Closed
romanlutz
Roman Lutz (romanlutz) reviewed Aug 1, 2024
View reviewed changes
Comment thread pyrit/datasets/score/crescendo/conversation_objective/eval.yaml
romanlutz
Roman Lutz (romanlutz) approved these changes Aug 5, 2024
View reviewed changes
Comment thread pyrit/orchestrator/crescendo_orchestrator.py Outdated
Comment thread pyrit/datasets/orchestrators/crescendo/crescendo_variant_1.yaml Outdated
Comment thread pyrit/datasets/orchestrators/crescendo/crescendo_variant_2.yaml Outdated
Comment thread pyrit/datasets/orchestrators/crescendo/crescendo_variant_3.yaml Outdated
Comment thread pyrit/datasets/orchestrators/crescendo/crescendo_variant_4.yaml Outdated
Comment thread pyrit/datasets/orchestrators/crescendo/crescendo_variant_5.yaml Outdated
Comment thread pyrit/datasets/score/crescendo/conversation_objective/refusal.yaml Outdated
Comment thread pyrit/memory/memory_interface.py Outdated
@richlundeen
Richard Lundeen (richlundeen) merged commit 216d19e into microsoft:main Aug 7, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@richlundeen Richard Lundeen (richlundeen) richlundeen left review comments
@romanlutz Roman Lutz (romanlutz) romanlutz approved these changes

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

3 participants

@SafwanA02 @romanlutz @richlundeen