How does the skill help with dataset quality?

It uses scripts to detect duplicates, check image resolutions, and verify that captioning matches the artist's intent while maintaining style-content separation.

What are the Four Gates in the training workflow?

The Four Gates are mandatory validation steps—Dataset Audit, Config Check, Environment Validation, and Dry Run—that ensure your training is configured correctly before any actual GPU time is consumed.

Which training backends does this skill support?

It provides specialized adapters and configuration generation for Kohya (sd-scripts), SimpleTuner, and AI-Toolkit.

Can this skill help prevent Out of Memory (OOM) errors?

Yes, it includes a VRAM calculator and pre-flight environment checks to ensure your hardware can handle the training parameters before you start.

LoRA Training Specialist

Name: LoRA Training Specialist
Author: 0xHoneyJar

by0xHoneyJar

•

Data Science & ML

Guides artists and developers through the end-to-end process of training custom AI art models with a focus on dataset quality and resource optimization.

The LoRA Training Specialist is a comprehensive framework designed to manage the complexities of AI art production within Claude Code. It implements a rigorous 'Four Gates' validation system—covering datasets, configurations, environments, and dry runs—to ensure that training sessions are successful and GPU resources are not wasted. By providing domain-specific guidance on style-aware captioning, VRAM estimation, and hyperparameter selection, it enables creators to effectively replicate specific artistic styles and subjects while maintaining high technical standards.

Key Features

01Automated dataset auditing, duplicate detection, and resolution checking

02Evaluation grid generation to identify the optimal LoRA weight and strength

031 GitHub stars

04Comprehensive style-aware captioning methodology and format selection

05Smart training configuration generation for Kohya, SimpleTuner, and AI-Toolkit

06Four Gates validation system to prevent training failures and wasted GPU time

Use Cases

01Developing consistent character or object LoRAs for high-fidelity image generation

02Training a custom AI model to replicate a specific artist's unique visual aesthetic

03Optimizing machine learning workflows for specific hardware and VRAM constraints

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add 0xhoneyjar/ategnatos train

For use in Claude.ai and ChatGPT

Download Skill