Repository navigation

CL Backdoor & LM Jailbreak SETs - #39

Merged

Zippo00 merged 45 commits into

mainfrom

feature/cl_backdoor

Jul 15, 2026

Merged

CL Backdoor & LM Jailbreak SETs#39
Zippo00 merged 45 commits into
mainfrom
feature/cl_backdoor

Commits on Apr 29, 2026
- tweaked wording on some comments
  Zippo00
  committed
  83acead
- README typo fix
  Zippo00
  committed
  a3dc64e
- comment tweak
  Zippo00
  committed
  9ccc173
- created dirs and files for cl components; initial commit of cl schema & pipeline
  Zippo00
  committed
  84903d9
Commits on May 1, 2026
- changed CL SET & Connector config files from JSON to YAML.
  Zippo00
  committed
  c37ff4c
- created base class for CL connectors & GenericRESTCLConnector is WiP
  Zippo00
  committed
  fab1de4
- minor tweaks to CL pipeline schema
  Zippo00
  committed
  0fbbda2
- CL Backdoor SET WiP
  Zippo00
  committed
  da6ba34
Commits on May 5, 2026
- cl generic rest connector
  Zippo00
  committed
  c512199
- GenericRESTLMConnector.generate() bug fix
  Zippo00
  committed
  e792401
- GenericRESTCLConnector.query() small tweaks.
  Zippo00
  committed
  f376978
Commits on May 6, 2026
- minor typo fix
  Zippo00
  committed
  793ca88
- minor typo fix
  Zippo00
  committed
  5dd8236
- initialize all class attributes in RedQueen.__init__()
  Zippo00
  committed
  192abfc
- define all class attributes in RedQueen.__init__()
  Zippo00
  committed
  d6f4220
- cl backdoor SET WiP; dependecies updt
  Zippo00
  committed
  e0e829d
Commits on May 18, 2026
- load data from file util
  Zippo00
  committed
  4b01499
- cl base evaluator finished; cl backdoor evaluators added
  Zippo00
  committed
  122ee0b
- minor edit to raw_responses field
  Zippo00
  committed
  271d888
- added StageResult to imports
  Zippo00
  committed
  36209af
- narrowed transformers version to =<5.7.0 (5.8.0 broke the Mistral model used in the framework)
  Zippo00
  committed
  246aba5
- cl backdoor data poisoning logic added
  Zippo00
  committed
  76b4fd6
- cl backdoor config file updated
  Zippo00
  committed
  84f0022
- modified a comment to include a supported value for trigger_type
  Zippo00
  committed
  4b313a3
Commits on May 20, 2026
- added evaluator configs to backdoor.yaml; created logic for evaluating Backdoor SET
  Zippo00
  committed
  2018168
- comment typo fix
  Zippo00
  committed
  fceb28f
- init file updt
  Zippo00
  committed
  8f2e15b
- removed unnecessary class attribute
  Zippo00
  committed
  1a16cbd
Commits on Jun 10, 2026
- minor comment edit
  Zippo00
  committed
  9d2c7e0
- report()
  Zippo00
  committed
  7bf5d10
Commits on Jun 22, 2026
- added batch_handling config
  Zippo00
  committed
  a8ee7c1
- batch and file handling
  Zippo00
  committed
  0d14b01
- typo fix
  Zippo00
  committed
  83794a1
- added logic for calculating stage metrics
  Zippo00
  committed
  b2a5b55
- fixed circular importerror
  Zippo00
  committed
  2bce270
- Merge branch 'main' into feature/cl_backdoor
  Zippo00
  committed
  334a528
- fixed init bugs in cl_backdoor
  Zippo00
  committed
  02483a5
Commits on Jul 1, 2026
- _get_results() & _get_grouped_results() fixed to work with other EvaluationResult objects in addition to pipelines.languagemodel.schema.EvaluationResult objects
  Zippo00
  committed
  028800c
Commits on Jul 15, 2026
- added jailbreak SET for language models
  Zippo00
  committed
  10b198f
- imporved pickle support for pickled files
  Zippo00
  committed
  eb43f15
- bug fixes
  Zippo00
  committed
  566b830
- improved prompt_injection_mini.json and changed it to prompt_injection.json
  Zippo00
  committed
  d314a1d
- bug fixes
  Zippo00
  committed
  3d33d8f
- connector improvements
  Zippo00
  committed
  034be56
- docstring added
  Zippo00
  committed
  054b243

Commits

Commits on Apr 29, 2026

Commits on May 1, 2026

Commits on May 5, 2026

Commits on May 6, 2026

Commits on May 18, 2026

Commits on May 20, 2026

Commits on Jun 10, 2026

Commits on Jun 22, 2026

Commits on Jul 1, 2026

Commits on Jul 15, 2026