Skip to content

MAINT: Light refactor of Aegis AI content safety dataset for consistency - #1976

Merged
romanlutz merged 6 commits into
microsoft:mainfrom
fdubut:aegis_dataset
Jun 10, 2026
Merged

MAINT: Light refactor of Aegis AI content safety dataset for consistency#1976
romanlutz merged 6 commits into
microsoft:mainfrom
fdubut:aegis_dataset

Conversation

@fdubut

@fdubut fdubut commented Jun 9, 2026

Copy link
Copy Markdown
Contributor

Closes #1933

Description

Light refactor of aegis_ai_content_safety_dataset.py following similar pattern as other datasets:

  • Use new AegisHarmCategory Enum for harm categories
  • Enhance filtering logic
  • Add richer metadata to each prompt

Tests and Documentation

Added a handful of tests to cover changes. All tests pass.

@jsong468 jsong468 self-assigned this Jun 9, 2026
Comment thread pyrit/datasets/seed_datasets/remote/aegis_ai_content_safety_dataset.py Outdated
Comment thread pyrit/datasets/seed_datasets/remote/aegis_ai_content_safety_dataset.py Outdated

@jsong468 jsong468 left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

small NITs but looks good, thanks!

Co-authored-by: jsong468 <songjustin@microsoft.com>
@romanlutz romanlutz changed the title MAINT: Light refactor of Aegis AI content safety dataset MAINT: Light refactor of Aegis AI content safety dataset for consistency Jun 10, 2026
Copilot AI added 3 commits June 10, 2026 06:38
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
The attribute was set in __init__ but never read; the filter loop uses
the pre-computed self._selected_category_values set instead.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
@romanlutz
romanlutz added this pull request to the merge queue Jun 10, 2026
Merged via the queue into microsoft:main with commit 4955237 Jun 10, 2026
53 checks passed
@fdubut
fdubut deleted the aegis_dataset branch June 10, 2026 18:08
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Consider upgrading from Aegis AI content safety dataset v1 to v2

4 participants