What data formats are supported by these scripts?

The skill supports JSONL, CSV, Parquet, and Arrow formats for comprehensive dataset validation and quality checking.

Does it work with specific ML frameworks?

It is designed to support major ML frameworks including PyTorch, TensorFlow, and JAX, with specific checks for model weights and configs.

Can I use custom validation rules?

Yes, the skill allows you to provide custom JSON schemas to enforce specific data structures, required fields, and value ranges for your unique datasets.

Can it check GPU availability before training?

Yes, it includes specialized scripts to verify CUDA versions, GPU count, driver compatibility, and minimum VRAM requirements.

How does it help prevent training failures?

By performing 'dry runs' of the pipeline and validating token counts and data schemas, it catches common errors before they cause expensive mid-training crashes.

ML Training Validation Scripts

Name: ML Training Validation Scripts
Author: vanman2024

byvanman2024

Data Science & ML

Ensures the integrity of machine learning training workflows by validating datasets, model checkpoints, and system dependencies.

About

This skill provides a comprehensive suite of production-ready validation and testing utilities designed specifically for machine learning projects. It helps developers maintain high data quality, verify model configuration integrity, and ensure end-to-end pipeline correctness before committing to expensive training runs. By automating checks for schema compliance, GPU availability, and model weight integrity, it significantly reduces the risk of training failures and silent data corruption in complex AI development workflows.

Key Features

Model checkpoint and configuration verification for PyTorch and TensorFlow.
0 GitHub stars
End-to-end pipeline testing with sample data and loss range checks.
Detailed JSON reporting for validation passes and failures.
Automated dataset validation with schema enforcement and quality metrics.
System dependency and GPU/VRAM availability auditing.

Use Cases

Verifying large-scale training datasets for format compliance and label distribution before fine-tuning.
Auditing remote server environments to ensure correct CUDA versions and sufficient VRAM exist.
Debugging failed training runs by isolating issues in the data-loading or model-initialization phases.

About

Key Features

Model checkpoint and configuration verification for PyTorch and TensorFlow.
0 GitHub stars
End-to-end pipeline testing with sample data and loss range checks.
Detailed JSON reporting for validation passes and failures.
Automated dataset validation with schema enforcement and quality metrics.
System dependency and GPU/VRAM availability auditing.

Use Cases

Verifying large-scale training datasets for format compliance and label distribution before fine-tuning.
Auditing remote server environments to ensure correct CUDA versions and sufficient VRAM exist.
Debugging failed training runs by isolating issues in the data-loading or model-initialization phases.