From 597e5082491b999dc408b8a2f8eac153eaa5105b Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Kov=C3=A1cs=20=C3=81d=C3=A1m?= Date: Thu, 13 Aug 2026 14:42:26 +0200 Subject: [PATCH] release: 0.2.3 --- CHANGELOG.md | 18 ++++++++++++++++++ pyproject.toml | 2 +- 2 files changed, 19 insertions(+), 1 deletion(-) diff --git a/CHANGELOG.md b/CHANGELOG.md index fd0af93..a809502 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -2,6 +2,24 @@ All notable changes to LettuceDetect are documented here. +## [0.2.3] - 2026-08-14 + +### Added +- `lettucedetect` command-line entry point: file/stdin/literal inputs, spans or tokens JSON output, transformer and LLM methods (#47, thanks @SemTiOne) +- Detector latency/throughput benchmark script with context-length sweep and peak-memory reporting (#58, thanks @sanmaxdev) +- Claude Code integration: a Stop hook that checks agent answers against grounding context and feeds flagged spans back to the agent; API, in-process, and generative (vLLM) modes (#50) +- Custom zero-shot taxonomies: `include_taxonomy` accepts `{name: description}` label sets on the taxonomy-head cascade and the native generative detector; labels enter as text, so unseen labels type from their description (#92) +- Typed spans in the Streamlit demo: model selector, per-category colors, span table (#57, thanks @w3lld1) + +### Fixed +- `LLMDetector` honors `output_format="tokens"` across all three predict entry points (#65, thanks @mobinert) +- Corrected published model ids in the CLI help and docs; the multilingual EuroBERT models are documented as one checkpoint per language with the `transformers<5` note (#33) +- `zero_shot`, `fewshot_path`, and `prompt_path` now raise on native generative detectors instead of being silently ignored (#92) + +### Documentation +- Public roadmap (`PUBLIC_ROADMAP.md`) with milestone-linked Now/Next/Exploring sections +- `min_confidence` edge cases documented (#64, thanks @vku2018) + ## [0.2.2] - 2026-07-05 ### Fixed diff --git a/pyproject.toml b/pyproject.toml index ad412eb..79f9ebd 100644 --- a/pyproject.toml +++ b/pyproject.toml @@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta" [project] name = "lettucedetect" -version = "0.2.2" +version = "0.2.3" description = "Lettucedetect is a framework for detecting hallucinations in RAG applications." readme = {file = "README.md", content-type = "text/markdown"} requires-python = ">=3.11"