hp-tune

Analyze past experiment results to propose hyperparameter configurations for ML models.

Updated Mar 5, 2026

Search Tags:#optimization #mlops #machine learning #hyperparameter tuning #experimentation #llm-driven

Ready to run

npx skills add https://github.com/ChuaHanChong/ml-optimizer --skill hp-tune

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: hp-tune
Source: https://github.com/ChuaHanChong/ml-optimizer/tree/main/skills/hp-tune
Command: npx skills add https://github.com/ChuaHanChong/ml-optimizer --skill hp-tune

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires json, sys, result_analyzer, error_tracker, experiment_setup, and includes scripts (resource) and references (resource) components.

What problem does it solve?

This Skill automates the complex and time-consuming process of finding optimal hyperparameters for machine learning models, moving beyond traditional search methods to leverage LLM reasoning.

Core Features & Use Cases

LLM-Driven Tuning: Claude reasons directly about past results to propose new configurations, avoiding rigid algorithms like Optuna or grid search.
Iterative Improvement: Analyzes experiment results to intelligently explore and exploit the hyperparameter search space.
Branch-Aware Analysis: Adapts tuning strategies based on different code branches, recognizing that hyperparameter sensitivities can vary.
Use Case: When optimizing a deep learning model's performance, use this Skill to systematically explore learning rates, batch sizes, and other parameters based on previous training runs, leading to better accuracy or lower loss.

Quick Start

Use the hp-tune skill to propose the next set of hyperparameter configurations based on previous experiment results.

Frequently Asked Questions about hp-tune

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

FAQPage Schema

How does LLM-driven hyperparameter tuning work for machine learning models?▼

LLM-driven hyperparameter tuning uses Claude to reason directly about past experiment results, proposing new configurations by intelligently exploring and exploiting the search space rather than relying on rigid algorithms like grid search.

Can I adapt hyperparameter search strategies based on different code branches?▼

Yes, branch-aware analysis adapts hyperparameter tuning strategies based on different code branches, recognizing that model parameter sensitivities can vary across experimental code changes and iteration budgets.

How do I propose the next set of hyperparameters based on previous training runs?▼

You provide past experiment results, and the system analyzes them to propose the next set of hyperparameter configurations, aiming to optimize primary metrics like accuracy or loss through iterative improvement.

What is the best way to optimize deep learning parameters without using Optuna?▼

Using LLM reasoning to analyze previous training runs offers an alternative to traditional search algorithms, systematically exploring learning rates and batch sizes to achieve better model performance.

Does this hyperparameter tuning approach adjust to available iteration budgets?▼

Yes, the tuning strategy adapts based on iteration progress and the available budget, ensuring intelligent exploration and exploitation of the search space to optimize primary metrics within constraints.