v0.10.1

NVIDIA/garakv0.10.1Dec 19, 2024by jmartin-tech

AI Summary

Adds probes for ANSI terminal takeover and Past Tense Vulnerability. Promotes OpenAICompatible to a first-class generator and adds an experimental features flag.

Key Highlights

  • ANSI terminal takeover probe
  • Past Tense Vulnerability probe
  • OpenAICompatible promoted to first-class generator
  • Experimental features flag added

New Features

  • ANSI terminal takeover probe
  • Past Tense Vulnerability probe
  • OpenAICompatible first-class support
  • Experimental features flag

Full Release Notes

## What's Changed
### New plugins
* probe: ANSI terminal takeover by @leondz in https://github.com/NVIDIA/garak/pull/1025
* probe: Past Tense Vulnerability by @Shine-afk in https://github.com/NVIDIA/garak/pull/924
* generator: promote OpenAICompatible as first class generator by @jmartin-tech in https://github.com/NVIDIA/garak/pull/1021

### New features
* feature: amend garak user-agent by @leondz in https://github.com/NVIDIA/garak/pull/960
* feature: reinforce `trust_remote_code=False` by @leondz in https://github.com/NVIDIA/garak/pull/979
* generators: add option to specify HTTP codes to skip generation on, for `RestGenerator` by @leondz in https://github.com/NVIDIA/garak/pull/999
* feature: add experimental features flag to be only accessible in core config by @leondz in https://github.com/NVIDIA/garak/pull/1003
* feature: enable configuration updates by @jmartin-tech in https://github.com/NVIDIA/garak/pull/1026
* feature: use hf chat support by @jmartin-tech in https://github.com/NVIDIA/garak/pull/1047
* feature: per-probe tags now adjustable based on payload selection by @leondz in https://github.com/NVIDIA/garak/pull/1031

### Documentation
* doc: Update README.md to notify of migration by @leondz in https://github.com/NVIDIA/garak/pull/1002
* docs: add arxiv lozenge by @leondz in https://github.com/NVIDIA/garak/pull/1030

### Tuning & fixes
* remove no longer needed skip-duplicate-actions by @jmartin-tech in https://github.com/NVIDIA/garak/pull/976
* Sanity test for pip install from repository by @jmartin-tech in https://github.com/NVIDIA/garak/pull/1000
* add CODEOWNERS by @jmartin-tech in https://github.com/NVIDIA/garak/pull/1001
* Update reference to reflect migration of org by @jmartin-tech in https://github.com/NVIDIA/garak/pull/1004
* use allowed action name by @jmartin-tech in https://github.com/NVIDIA/garak/pull/1005
* Fixing a few typos by @zoenolan in https://github.com/NVIDIA/garak/pull/1006
* Fix huggingface inference endpoint name by @jmartin-tech in https://github.com/NVIDIA/garak/pull/1011
* detect if tokenizer is not loaded and adjust by @jmartin-tech in https://github.com/NVIDIA/garak/pull/1012
* fix: correct argument in TAP by @harshraj172 in https://github.com/NVIDIA/garak/pull/1022
* refactor attempt to utilized property annotations by @jmartin-tech in https://github.com/NVIDIA/garak/pull/1027
* Restrict transformers version until MPS issue is addressed by @jmartin-tech in https://github.com/NVIDIA/garak/pull/1039
* pass device to transfomers pipeline by @jmartin-tech in https://github.com/NVIDIA/garak/pull/1042
* Fix tap.PAIR run error - run_tap param evaluator_model by @rafaelsandroni in https://github.com/NVIDIA/garak/pull/1045

## New Contributors
* @zoenolan made their first contribution in https://github.com/NVIDIA/garak/pull/1006
* @harshraj172 made their first contribution in https://github.com/NVIDIA/garak/pull/1022
* @Shine-afk made their first contribution in https://github.com/NVIDIA/garak/pull/924
* @rafaelsandroni made their first contribution in https://github.com/NVIDIA/garak/pull/1045

**Full Changelog**: https://github.com/NVIDIA/garak/compare/v0.10.0...v0.10.1