validation-scripts

Validate ML data, models, and pipelines with structured reports.

Updated Jan 15, 2026

Search Tags:#validation #schema #data-quality #data-validation #pipeline-testing #ml-training

Ready to run

npx skills add https://github.com/KaiserWhoLearns/skillsbench --skill validation-scripts-kaiserwholearns

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: validation-scripts
Source: https://github.com/KaiserWhoLearns/skillsbench/tree/main/registry/terminal_bench_1.0/predict-customer-churn/environment/skills/validation-scripts
Command: npx skills add https://github.com/KaiserWhoLearns/skillsbench --skill validation-scripts-kaiserwholearns

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires python3, jq, nvidia-smi, transformers, torch, pandas, pyarrow, PyYAML, and includes scripts (resource) components.

What problem does it solve?

ML training relies on clean data and reliable pipelines; this Skill provides validated data and tested workflows to prevent training-time failures and data quality issues.

Core Features & Use Cases

Data validation: validate datasets against schemas, detect duplicates, and check token counts to ensure compatibility with models.
Pipeline testing: end-to-end checks that pipelines run from data ingestion to inference, catching integration faults early.
Dependency verification: ensure required packages and tools are present before training, with actionable reports and fixes.

Quick Start

Run the validation scripts to audit your ML data and pipelines.

Frequently Asked Questions about validation-scripts

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I validate ML training data for schema and token compatibility before training?▼

Run data validation scripts to check datasets against schemas, detect duplicates, and verify token counts for model compatibility. This prevents training-time failures by catching data quality issues early with structured reports and recommended fixes.

What is pipeline testing in ML workflows and when do I need it?▼

Pipeline testing runs end-to-end checks from data ingestion to inference to catch integration faults early. You need it when validating ML workflows to ensure pipelines execute completely without breaking during training or inference stages.

Can I use these scripts to verify PyTorch and Transformers dependencies?▼

Yes, dependency verification scripts check for required packages like PyTorch, Transformers, pandas, and pyarrow. They ensure required tools are present before training, generating actionable reports with recommended fixes for any missing components.

How do I test an ML pipeline end-to-end from data ingestion to inference?▼

Run the test-pipeline.sh script to validate the complete ML workflow from data ingestion to inference. This catches integration faults early and provides structured reports to fix pipeline reliability issues before training begins.

Does data validation with pandas and pyarrow work for large datasets?▼

Data validation scripts using pandas and pyarrow perform schema checks, duplicate detection, and token validation across ML datasets. They handle data-quality validation tasks, providing structured reports and recommended fixes for any issues found.

What's the best way to prevent data quality issues in ML training pipelines?▼

Implement validation scripts for schema checks, duplicate detection, and dependency verification before training. This ensures clean data and tested workflows, catching integration faults and quality issues early with structured actionable reports.