wandb/rai-toolkit

fix(scorers): restore keyword derivatives without changing score weights 1 day ago

bug help wanted status: needs author

wandb/rai-toolkit

8

Python

fix(scorers): reject non-finite normalized scores 4 days ago

bug help wanted status: needs author

wandb/rai-toolkit

8

Python

fix(datasets): identify JSONL file and line in row errors 4 days ago

bug help wanted good first issue status: in review

wandb/rai-toolkit

8

Python

fix(cli): fail policy lint when no YAML files were checked 4 days ago

bug help wanted good first issue status: in review

wandb/rai-toolkit

8

Python

feat(cli): report the toolkit version with --version 4 days ago

enhancement help wanted good first issue status: in review

wandb/rai-toolkit

8

Python

Prevent refusal text from hiding separate attack-success signals 5 days ago

bug help wanted status: needs spec

wandb/rai-toolkit

8

Python

Add a deterministic tool-call accuracy scorer for agentic runs 6 days ago

enhancement help wanted status: in review

wandb/rai-toolkit

8

Python

docs: refresh contributor-facing README facts and checks 9 days ago

documentation help wanted good first issue status: available

wandb/rai-toolkit

8

Python

Make built-in adapter call-time options explicit and reject unsupported kwargs 13 days ago

bug help wanted status: available

wandb/rai-toolkit

8

Python

fix(scorers): use structured data boundaries in LLM judge prompts 14 days ago

bug help wanted status: blocked

wandb/rai-toolkit

8

Python

Add a vendor-neutral CallableModel adapter for custom inference functions 15 days ago

enhancement help wanted status: available

wandb/rai-toolkit

8

Python

Add a provider-neutral model adapter contract and conformance suite 15 days ago

documentation enhancement help wanted good first issue status: available

wandb/rai-toolkit

8

Python

Add an optional Anthropic model adapter 16 days ago

enhancement good first issue status: available

wandb/rai-toolkit

8

Python

fix(examples): preserve falsy dataset values and validate limits 19 days ago

bug help wanted good first issue status: available

wandb/rai-toolkit

8

Python

fix(scorers): make CompositeScorer honor async and unassessed results 19 days ago

bug help wanted status: available

wandb/rai-toolkit

8

Python

fix: use one toolkit version across packages and reports 19 days ago

bug help wanted status: available

wandb/rai-toolkit

8

Python

fix(scoping): isolate built-in policies by industry preset 19 days ago

bug help wanted status: available

wandb/rai-toolkit

8

Python

test(cli): add offline coverage for public commands 19 days ago

help wanted good first issue

wandb/rai-toolkit

8

Python

test(models): lock down the existing OpenAI-compatible endpoint contract 19 days ago

help wanted good first issue

wandb/rai-toolkit

8

Python

make_weave_rai_scorer names wrappers by class, so two instances of one scorer collide 20 days ago

bug help wanted good first issue

wandb/rai-toolkit

8

Python

Weave-native assessment path silently drops additional_scorers 21 days ago

bug help wanted

wandb/rai-toolkit

8

Python

LLMJudgeScorer.__init__ wipes the class-level category when constructed directly 21 days ago

bug good first issue

wandb/rai-toolkit

8

Python

Add model-provider and eval-harness adapters 22 days ago

enhancement help wanted

wandb/rai-toolkit

8

Python

Add context precision and recall scorers for RAG retrieval 22 days ago

enhancement help wanted

wandb/rai-toolkit

8

Python

Add RAG and agent-specific evaluation metrics 22 days ago

enhancement help wanted

wandb/rai-toolkit

8

Python

Add an industry preset beyond healthcare / finance / government (legal or HR) 23 days ago

enhancement good first issue

wandb/rai-toolkit

8

Python

Contribute additional red-team attack templates 23 days ago

enhancement help wanted good first issue

wandb/rai-toolkit

8

Python

Evidence spans are over-discarded when the judge normalizes whitespace or quotes 24 days ago

bug good first issue

wandb/rai-toolkit

8

Python

GroundednessScorer should skip behavioral-refusal rows like FactualityJudge 24 days ago

bug good first issue

wandb/rai-toolkit

8

Python

Add framework mappings for more regimes (UK, Singapore IMDA, India DPDP) and deepen EU AI Act Annex III 24 days ago

enhancement help wanted good first issue

wandb/rai-toolkit

8

Python

Add a groundedness scorer for RAG outputs 26 days ago

enhancement help wanted

wandb/rai-toolkit

8

Python

Add NYC Local Law 144 mapping (automated employment decision tools) 28 days ago

enhancement good first issue

wandb/rai-toolkit

8

Python

Add Colorado AI Act framework mapping 28 days ago

enhancement good first issue

wandb/rai-toolkit

8

Python

Add ISO/IEC 42001 framework mapping 28 days ago

enhancement good first issue

wandb/rai-toolkit

8

Python

Open Issues Need Help