docs(validation): record TypeScript semantic-pass benchmark numbers#47
Merged
Conversation
TypeScript is LLM-only (the scanner is Python-only by design). A 20-case labeled benchmark (8 rotten / 12 sound, Jest/Vitest) run blind on a small model (Claude Haiku) scored precision 1.00, recall 0.625 (1.00 on clear-cut smells), F1 0.77. The three misses are the same boundary cases as the Python run (pure-delegation passthrough, trivial one-operator formula), already tracked as open issues. Cross-language reproduction supports the claim that the pass carries beyond Python. Raw data and spreadsheet stay local (.handoff/, gitignored).
github-actions
Bot
added
the
documentation
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Records the measured TypeScript validation numbers (conclusions only; raw data and the spreadsheet stay local).
TypeScript is covered by the LLM semantic pass alone (the scanner is Python-only by design). A 20-case labeled benchmark (8 rotten / 12 sound) in Jest/Vitest idioms, run blind on a small model (Claude Haiku):
The three misses are the same boundary cases as the Python run (a pure-delegation passthrough asserted through an edge mock, and a trivial single-operator formula), already tracked in #43 and #44. Cross-language reproduction supports the claim that the pass carries beyond Python.
Docs-only: README "How falsegreen is validated" and VALIDATION.md.