Could an agent reproduce your training run?

Attach evidence to the training agent’s run manifestCode and environment · src/train.py at commit a21c, with locked dependenciesData and features · Immutable orders snapshot and versioned feature definitionsTraining recipe · Versioned config with seed, cutoff date and validation windowEvaluation and artifact · Registered model digest and passing holdout report linked to this run

Registry gate closed. Missing evidence: Code and environment, Data and features, Training recipe, Evaluation and artifact.

Get the ML in Production syllabus →