Starstruk Logo MarkStarstruk

v0.13.2

NVIDIA/garakv0.13.2Nov 10, 2025by jmartin-tech

AI Summary

Added iterative probe framework with FITD as example implementation and new 'any' detector. Multiple bug fixes for transformers, LangChain, and probe modules.

Key Highlights

  • Base class for iterative probes and FITD
  • Detectors: any detector
  • Transformers bugfix for pegasus
  • DAN probes refactored using metaclass pattern
  • Conversation support improvements
  • Guide to running garak faster

New Features

  • Base class for iterative probes
  • FITD (Follow the Instructions) probe
  • Any detector

Full Release Notes

## What's Changed

### New plugins
* probe: Base class for iterative probes and FITD as an example by @aishwaryap in https://github.com/NVIDIA/garak/pull/1414
* detectors: any by @leondz in https://github.com/NVIDIA/garak/pull/1427

### Improved plugins
* detectors: Fix incorrect target class for MustRefuteClaimModel by @RobGeada in https://github.com/NVIDIA/garak/pull/1419
* probes: Win11 probe still uses Windows 10 product names and goal by @2020ashish in https://github.com/NVIDIA/garak/pull/1422
* probes: properly parse the OFCOM list file by @paulinek13 in https://github.com/NVIDIA/garak/pull/1438
* probes: atkgen probe verbose output by @MrMoshkovitz in https://github.com/NVIDIA/garak/pull/1447
* probes: NameError in TAP/PAIR probes by @MrMoshkovitz in https://github.com/NVIDIA/garak/pull/1452
* detectors: access Message.text in PlainJSON by @jmartin-tech in https://github.com/NVIDIA/garak/pull/1454
* generators: enable parameter suppression to LiteLLM by @JDSmith-F1 in https://github.com/NVIDIA/garak/pull/1411

### Documentation
* docs: revert unintended changes to translation configs by @paulinek13 in https://github.com/NVIDIA/garak/pull/1458
* docs: guide to running garak faster by @leondz in https://github.com/NVIDIA/garak/pull/1463
* docs: patch filename in doc on accelerating garak  by @leondz in https://github.com/NVIDIA/garak/pull/1465

### Tuning & fixes
* bugfix: `transformers` disconnect btw pegasus and `transformers-community/group-beam-search` by @leondz in https://github.com/NVIDIA/garak/pull/1403
* Use ABC to exclude abstract classes from --list_probes by @patriciapampanelli in https://github.com/NVIDIA/garak/pull/1398
* limit langchain version until migration by @jmartin-tech in https://github.com/NVIDIA/garak/pull/1420
* correct _mint_attempt conversation support by @jmartin-tech in https://github.com/NVIDIA/garak/pull/1417
* update github-specific content to 'target' param pattern by @leondz in https://github.com/NVIDIA/garak/pull/1426
* add __init__.py for all test paths containing `.py` by @jmartin-tech in https://github.com/NVIDIA/garak/pull/1428
* Fix/recommended detector migration by @patriciapampanelli in https://github.com/NVIDIA/garak/pull/1409
* refactor DAN probes using metaclass pattern by @patriciapampanelli in https://github.com/NVIDIA/garak/pull/1380
* update detectors to reflect immutable attempt prompt by @jmartin-tech in https://github.com/NVIDIA/garak/pull/1415
* improve prompt translation support for Conversation related types by @jmartin-tech in https://github.com/NVIDIA/garak/pull/1441
* fix: aggregate probes details in digest by @parkanzky in https://github.com/NVIDIA/garak/pull/1459
* fix: reduce window for multiprocessing race by @jmartin-tech in https://github.com/NVIDIA/garak/pull/1464

## New Contributors
* @RobGeada made their first contribution in https://github.com/NVIDIA/garak/pull/1419
* @2020ashish made their first contribution in https://github.com/NVIDIA/garak/pull/1422
* @paulinek13 made their first contribution in https://github.com/NVIDIA/garak/pull/1438
* @MrMoshkovitz made their first contribution in https://github.com/NVIDIA/garak/pull/1447
* @JDSmith-F1 made their first contribution in https://github.com/NVIDIA/garak/pull/1411
* @parkanzky made their first contribution in https://github.com/NVIDIA/garak/pull/1459

**Full Changelog**: https://github.com/NVIDIA/garak/compare/v0.13.1...v0.13.2