FEAT Add DecodingTrust AdvGLUE++ (adversarial robustness) dataset#2230
Open
rahul-art wants to merge 1 commit into
Open
FEAT Add DecodingTrust AdvGLUE++ (adversarial robustness) dataset#2230rahul-art wants to merge 1 commit into
rahul-art wants to merge 1 commit into
Conversation
Adds a remote seed-dataset loader for the Adversarial Robustness perspective of the DecodingTrust benchmark (AdvGLUE++), one of the open DecodingTrust sub-tasks tracked in microsoft#291. The loader fetches the three AdvGLUE++ surrogate files (alpaca, vicuna, stable-vicuna) at a pinned commit SHA and exposes each perturbed GLUE input as a SeedPrompt. Records ship in two shapes: some carry a pre-composed `sentence`, others store the perturbed parts separately (premise/hypothesis, sentence1/sentence2, question1/question2). The loader composes the prompt from component fields when needed using the same layout DecodingTrust uses, so no records are dropped, and preserves gold label, attack method and the clean `original_*` counterparts in metadata. - New loader `_DecodingTrustAdvGLUEDataset` with `DecodingTrustAdvGLUEModel` and `DecodingTrustAdvGLUETask` filters, registered in the remote package. - 22 unit tests covering both record shapes, per-task composition (including the QNLI `sentence` overload), filters, metadata and error paths. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01LZPdM9LfD7nVRJRkyu2Ld9
|
@rahul-art please read the following Contributor License Agreement(CLA). If you agree with the CLA, please reply with the following information.
Contributor License AgreementContribution License AgreementThis Contribution License Agreement (“Agreement”) is agreed to by the party signing below (“You”),
|
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
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.
Adds a remote seed-dataset loader for the Adversarial Robustness perspective of the DecodingTrust benchmark (AdvGLUE++), one of the open DecodingTrust sub-tasks tracked in #291.
The loader fetches the three AdvGLUE++ surrogate files (alpaca, vicuna, stable-vicuna) at a pinned commit SHA and exposes each perturbed GLUE input as a SeedPrompt. Records ship in two shapes: some carry a pre-composed
sentence, others store the perturbed parts separately (premise/hypothesis, sentence1/sentence2, question1/question2). The loader composes the prompt from component fields when needed using the same layout DecodingTrust uses, so no records are dropped, and preserves gold label, attack method and the cleanoriginal_*counterparts in metadata._DecodingTrustAdvGLUEDatasetwithDecodingTrustAdvGLUEModelandDecodingTrustAdvGLUETaskfilters, registered in the remote package.sentenceoverload), filters, metadata and error paths.Claude-Session: https://claude.ai/code/session_01LZPdM9LfD7nVRJRkyu2Ld9
Description
Tests and Documentation