Skip to content

Verbatimeter

v0.1.3 Bugfix

This release fixes issues for SREs watching stability and regressions.

Published 14d Developer Productivity
✓ No known CVEs patched
Read the diff → Tool health → What is this tool? →

✓ No known CVEs patched in this version

Topics

ai-safety evaluation hallucination hallucination-detection lcs llm
+11 more
llm-trustworthiness longest-common-subsequence nlp python quotation string-matching string-matching-algorithms text-reuse text-verification verbatim verification

Summary

AI summary

Subsequence mode now ignores the ngram parameter, allowing check(..., mode="subsequence", ngram=1) to succeed without error.

Changes in this release

Bugfix Medium

Enforces ngram floor of 3 only in contiguous mode; subsequence mode now ignores ngram.

Enforces ngram floor of 3 only in contiguous mode; subsequence mode now ignores ngram.

Source: llm_adapter@2026-07-17

Confidence: high

 —
Full changelog

What's changed

  • Fix: the ngram floor (minimum 3) is now enforced only in contiguous mode, where ngram is used. check(..., mode="subsequence", ngram=1) succeeds instead of raising — subsequence mode now truly ignores ngram, as documented (with a regression test; ADR-0011 amended).
  • Docs: the RAG agent integration guide now gives whole-answer verbatim reuse and verbatim paraphrasing (mode="subsequence") equal standing alongside quotation verification, and introduces verbatim paraphrasing as a concept distinct from semantic paraphrase.

Full changelog: https://github.com/pierreolivierbonin/verbatimeter/compare/v0.1.2.post2...v0.1.3

View diff on GitHub

Weekly OSS security release digest.

The CVE patches and breaking changes that affected production tools this week. One email, every Sunday.

No spam, unsubscribe anytime.

Share this release

Share on X Share on Bluesky

Track Verbatimeter

Get notified when new releases ship.

Sign up free

About Verbatimeter

All releases →

Related context

Beta — feedback welcome: [email protected]