Audit ML experiment integrity using configurable cross-model reviewer backends.
npx skills add https://github.com/xqinag/ARIS-new --skill experiment-audit-xqinag
Please help me install this Agent Skill. Skill: experiment-audit Source: https://github.com/xqinag/ARIS-new/tree/main/skills/experiment-audit Command: npx skills add https://github.com/xqinag/ARIS-new --skill experiment-audit-xqinag
Audit experiment integrity using cross-model reviewer backends to detect fake ground truth, score normalization fraud, phantom results, and insufficient scope across provided experiment artifacts. It applies to post-experiment evaluation across ML experiments, enabling independent verification of datasets, metrics, and narrative claims. It supports configurable reviewer backends (codex/manual) and artifact collection with traceable reporting.
Provide the path to the experiment directory and artifacts to audit, then run the audit workflow.
High-intent search queries and answers about installing and using this skill.
Verify ML experiment integrity by applying cross-model reviewer backends to independently audit datasets and metrics, detecting fake ground truth, score normalization fraud, and phantom results across provided experiment artifacts.
Cross-model auditing is an independent review process where external reviewer backends evaluate experiment artifacts to verify narrative claims and detect fraud, ensuring the executor does not participate in integrity judgments.
Detect score normalization fraud by providing your experiment directory path to an audit workflow that applies external reviewer backends to trace and verify metrics and datasets across experiment artifacts.
Yes, experiment integrity verification supports configurable reviewer backends including both codex and manual modes to perform independent cross-model audits and generate traceable reports.
Cross-model experiment auditing guards against fake ground truth, score normalization fraud, phantom results, and insufficient scope by applying external reviewer backends to evaluate provided experiment artifacts.
Yes, experiment auditing operates without external dependencies, using configurable internal reviewer backends to collect artifacts and generate traceable reports for post-experiment evaluation.
kaminocorpLog metrics, artifacts, and hyperparameters for ML experiment tracking.
caw111Audit machine learning experiment results for fraud patterns using GPT-5.5.
jkfeeAudit experiment integrity across evaluation scripts, results, and paper claims.
JoaquinCampoExtract final numeric answers from GSM8K model outputs using official ground-truth formats.
