Deliver answer-bearing spans for assistant recall - #5
Merged
Merged
Conversation
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Five misses in the combined English confirmation lacked the answer-bearing portion of a source conversation even when another turn from the same session was present. This adds a benchmark-only recall adapter that ranks user requests, keeps the paired assistant response, and extracts a bounded source-only window from oversized turns while preserving the dated-event reserve.
The frozen all-54 assistant-recall packet screen reproduces the existing combined candidate as its baseline, improves lexical answer recoverability from 49/54 to 54/54 and first-24k exact annotated turns from 270 to 295, moves all five audited answer spans before character 1,126, and loses no annotated source session or recoverable answer. The first failed screen exposed reversed reserve-helper arguments; the preserved failure led to a behavioral ordering test and the correction in the final commit.
Validation:
pytest -q tests/test_recall_span_adapter.py tests/test_dated_event_reserve.py tests/test_dated_event_adapter.py(29 passed). Independent saved-artifact verification passes. No provider calls or paid evaluation were made; this is packet evidence rather than a new QA score.