Change prompt and add regression test for issue #53 token leak#55
Merged
Change prompt and add regression test for issue #53 token leak#55
Conversation
Adds an integration test that calls TextCorrectionService.correctText() against the on-device AI model to detect delimiter tokens leaking into the corrected output. Runs multiple iterations per input to surface the intermittent issue. Refs: #53
sebsto
force-pushed
the
issue-53-token-leak-test
branch
from
ac980b7 to
454ccd9
Compare
Contributor
|
Contributor
There was a problem hiding this comment.
Pull request overview
Adds a new integration regression test targeting issue #53, where [TEXT_START] / [TEXT_END] delimiter tokens intermittently leak into AI text-correction output from the on-device Apple Intelligence model.
Changes:
- Added a new Swift Testing suite that calls
TextCorrectionService.correctText()repeatedly for multiple inputs and correction styles. - Added assertions to detect delimiter token variants in returned corrected text.
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
sebsto
changed the title
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
Fixes #53 —
[TEXT_START]/[TEXT_END]delimiter tokens intermittently leaking into AI text correction output.Changes
Fix: explicit tag names in user prompt (
CorrectionStyle.swift)Changed the user prompt from
[TEXT ...]to explicitly name the[TEXT START]and[TEXT END]tags. The vague[TEXT ...]wording likely caused the AI model to hallucinate tag variants like[TEXT_START]/[TEXT_END]in its output. Being explicit about the tag names should reduce or eliminate the leak.Regression test (
TextCorrectionTokenLeakTests.swift)Calls
TextCorrectionService.correctText()against the on-device AI model with multiple inputs (English and French) across both correction styles (minimal and fullRephrase). Checks whether the response contains any delimiter token variants.Uses
withKnownIssueso the test suite stays green while the bug is open, and will fail once the fix is confirmed (signaling the wrapper should be removed).What was tested
Closes #53