Dataset check¶

plaid-check validates the integrity of a local PLAID dataset.

It checks:

• required on-disk files and directories;
• infos.yaml, metadata and split sample counts;
• sample conversion through the declared storage backend;
• invalid numeric values such as None, empty arrays, NaN and Inf;
• duplicated samples;
• optional problem_definitions/ feature names, splits and indices.

Basic usage¶

plaid-check /path/to/plaid_dataset

A valid dataset prints an [OK] line and returns exit code 0.

Options¶

Check only selected splits:

plaid-check /path/to/plaid_dataset --split train --split test

Check only selected problem definitions:

plaid-check /path/to/plaid_dataset --problem-definition regression_500

Emit a machine-readable report:

plaid-check /path/to/plaid_dataset --json

Make warnings fail the command:

plaid-check /path/to/plaid_dataset --strict

Report format¶

Messages are reported with a severity, a stable code, a location and a short description. Errors return exit code 1; warnings return exit code 2 only in strict mode.

Validation notes¶