feat(parity): per-pair flip-position dump for the margin bootstrap - #120
Merged
Conversation
…ap needs run_margin_analysis(dump_positions=...) writes one JSONL record per pair (tokens, flip positions, their reference margins) beside the aggregate margins JSON. MarginReport stays schema-identical; the dump is diagnostic. Enables the TODO.training-work/05 shipped-vs-head32 flip CIs; aggregate flip counts in the dump must reconcile with the report (tested).
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
run_margin_analysis gains dump_positions: one JSONL record per pair (tokens, flip positions, reference margins at those positions). MarginReport is unchanged (schema-additive diagnostic). The test reconciles the dump's aggregates with the report. This is the enabling half of TODO.training-work/05 — shipped-vs-head32 flip-rate CIs; the Modal margin re-runs follow.