Skip to content

Data: related paragraphs for every GC paragraph - #13

Merged
lszoszk merged 1 commit into
mainfrom
related-bge-oct2026
Oct 13, 2026
Merged

lszoszk merged 1 commit into
mainfrom
related-bge-oct2026

Conversation

@lszoszk

@lszoszk lszoszk commented Oct 13, 2026

Copy link
Copy Markdown
Owner

Recomputed related paragraphs for every General Comment paragraph.

  • Model: BAAI/bge-large-en-v1.5, the same as before, with the same document text (build_local_vectors.py). The vectors are re-embedded from the current corpus.
  • Coverage: 7,304 of 7,304 GC paragraphs (September: 7,181 of 7,216), including the 88 of CERD/C/GC/40 (GR 40 ¶1 → CERD GR 34).
  • Stability: for paragraphs present in both builds, the first neighbour is unchanged for 82%, and the top five overlap 85%. The differences follow text changes since May.
  • Tests: docs-reader, status-audit and dossier-toolbar, 44/44, including S4 (the related-paragraph artefacts are built).

🤖 Generated with Claude Code

…ge, October corpus)

7,304 of 7,304 GC paragraphs have neighbours (September: 7,181), including CERD/C/GC/40. Published from
lszoszk/generalcomments-pipeline.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@lszoszk
lszoszk merged commit 55e4d1c into main Oct 13, 2026
2 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant