⚡AgentSkills
🧮 ML Research Engineering · Evaluation Integrity

Document datasets with datasheets

Provenance, composition, collection process, and limitations recorded before modeling begins.

foundation~30 minResearch EngineersML ScientistsPhD Researchers

Steps

  1. 1Record provenance chain: sources, licenses, consent basis for personal data
  2. 2Quantify composition: demographics/classes/languages with known skews stated
  3. 3Document collection mechanics and any filtering applied
  4. 4List known failure modes and unsuitable-use cases explicitly
  5. 5Version datasheets with dataset versions — they evolve together
  6. 6Review for PII leakage risks with fresh eyes before release

Common Pitfalls

  • ▲Datasets inherited without provenance entering production models
  • ▲Limitations sections written only when reviewers ask

Commands

Install with skills CLI
$ npx skills add aniruddhaadak80/skills --skill evaluation-integrity-datasheet-documentation
Install globally
$ npx skills add aniruddhaadak80/skills --skill evaluation-integrity-datasheet-documentation -g

Tags

#datasets#documentation#responsible-ai#ml-research#evaluation-integrity

Related skills

Choose benchmarks that mean somethingintermediate

Benchmark suites matched to target capabilities with contamination checks and honest scopes.

🧮 ML Research Engineering·~30m
Open →