cyberneticlibrary
← library

Evaluate AI data, models, and quality

SKILLskillsetup L3★708
MigoXLab/dingo ↗
What it does

Evaluate data quality with rule-based or LLM metrics

Best for

Automated data quality gates in ML pipelines with both fast rule checks and semantic LLM scoring

Inputs
  • · Data file (JSONL/JSON/CSV/Parquet)
  • · Config with evaluators list
Outputs
  • · Summary.json with pass/fail counts
  • · Per-item reports in output JSONL
Requires
  • · Dingo evaluation engine
  • · OpenAI-compatible LLM API (optional)
  • · Pandas/Parquet libs
Preconditions

Data file exists, config JSON valid, LLM API key set if using LLM evals

Failure modes

Unsupported evaluator name, malformed config, data parsing error, API rate limit

Trust signals
  • · 50+ deterministic rule evaluators
  • · RAG eval support
  • · Multiple data sources (HF, S3, SQL)