Pelic corpus - #79
Merged
Merged
Conversation
The student sheet in Spanish quoted its 13.3% bound as "below 13 in 100" — the flattering direction, which the project forbids itself. Now 14. The two report caveats said "con un umbral de", the English "at a threshold" in a Spanish coat; now "a un umbral de". SKILL item 3 keeps the 5% target and its interval in one breath so an assistant repeating it cannot quote one without the other. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_015PEbbiYSNPw7jE3LrPNhyF
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_015PEbbiYSNPw7jE3LrPNhyF
…he target The prose review found the project's foundational phrase overclaiming on its most-read surfaces: GPT-3 existed from 2020, so a 2021 revision was not written "before generative models existed". Every surface in this pull request now says "before 2022", and the corpus README makes the weaker, true claim explicitly. The report messages change together in both languages so one page cannot contradict itself. The SKILL quoted "at most 5%" with nothing saying where 5% comes from; it now names it as the target the boundary is derived from and orders the interval quoted, never the point or the target alone. The Spanish student sheet no longer presents a bound with no supported threshold in the same shape as a rate at one. "Tightens as it grows" promised a direction the data do not; the page and the README now say the numbers follow the data either way. The step-below sentence attributes only what the numbers hold: the only flags at 25 are learners', and that is why the boundary does not sit at 25 — not why it sits at 30. Spanish: "de esta versión", "a un umbral", "acotar con precisión", "publica sobre sí misma", "la interpretación honesta". Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_015PEbbiYSNPw7jE3LrPNhyF
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.
What this changes
Why
Checklist
dotnet testpasses, with no new build warningsIf you touched the rule packs (
rules.en.json/rules.es.json):suggestion— what the writer should do insteadtests/SignsOfAI.Core.Testscovering itCleanHumaninScoringTestsstill scores exactly 0 (the new rule doesn't fire on ordinary prose).*; this runs on every keystroke in the browserIf you touched the articles in
Docs/Blog: