AI研究者キット

Preregistration of hypotheses, design, sampling plan and stopping rule, variables, analysis plan, exclusions and deviations log

Objective and success criteria, data mix and splits, baseline, architecture, 6ND compute budget in KaTeX, ablations, checkpoints, stop rules, gates, risks

Funder-style DMP – storage, sharing, privacy

Hypothesis and decision rule, setup, config and seeds, data versions, results vs baseline with variance, paired test, ablations, failure analysis, decision

Benchmark choice and validity, contamination checks, protocol, prompts and decoding, repeated runs, confidence intervals, McNemar test, comparison table

Task and rubric, rater recruitment and training, ethics and consent, sample size, blinding, Cohen's/Fleiss' kappa, Krippendorff's alpha, QC, analysis

ML conference paper: abstract, method, experiments, results, limitations, broader impact, reproducibility statement, appendix, NeurIPS 2026 checklist, ACL note

REFORMS-aligned ML science checklist: 32 items in 8 modules, leakage audit worksheet, evaluation protocol, code and data release, compute reporting

Model details, intended and out-of-scope uses, factors, metrics, data, disaggregated results, ethics, caveats

Dataset motivation, composition, collection, labelling, uses, distribution, maintenance, privacy and consent

Peer review report: summary, assessment, numbered major and minor comments, methods and ethics checks, recommendation, editor notes