FEAT: Adding Garak Remote Datasets#2063
Merged
rlundeen2 merged 3 commits intoJun 22, 2026
Merged
Conversation
Adds remote seed-dataset loaders for the datasets hosted under the garak-llm HuggingFace org so garak techniques are easier to use in PyRIT scenarios. Co-authored-by: Copilot <[email protected]>
Co-authored-by: Copilot <[email protected]>
romanlutz
approved these changes
Jun 22, 2026
romanlutz
left a comment
Contributor
There was a problem hiding this comment.
Fair to assume you ran the integration tests for all these?
Contributor
Author
I had, but noticing they are close to the timeout, so I'm adding a param to fetch a smaller number of datasets for the tests. |
The garak package registries are multi-million-row lists (npm ~3.3M, pypi ~555k); building every row as a SeedPrompt can exceed the e2e per-test timeout even with cached data. Add an optional keyword-only max_examples (default None = full list) to the garak base loader, honor it in both the text and audio fetch paths, and cap npm/pypi at 6 examples in the e2e suite. Co-authored-by: Copilot <[email protected]>
jsong468
reviewed
Jun 22, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
PyRIT is porting garak's probing techniques, and many of those techniques depend on reference data (package-name registries, system-prompt libraries, audio jailbreak clips) published under the
garak-llmHuggingFace org. This PR adds native seed-dataset loaders for that data so garak techniques can be wired into PyRIT scenarios without bespoke download code.It introduces a shared
_GarakRemoteDatasetbase (reusing the existing_RemoteDatasetLoaderprimitives) plus ten registered loaders: seven package-hallucination registries (pypi, npm, crates, rubygems, dart, perl, raku), two system-prompt libraries, and theaudio_achilles_heelclip set. All dataset names are prefixedgarak_, each row maps to oneSeedPromptwith source metadata preserved, and the change ships unit tests (mocked HF data), docs/notebook updates, and the citation.