results

Compare completed ML experiment results and identify top configurations.

Updated Aug 27, 2026
One-click install
npx skills add https://github.com/kang-jaehyun/Mac-the-researcher --skill results
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: results
Source: https://github.com/kang-jaehyun/Mac-the-researcher/tree/main/.claude/skills/results
Command: npx skills add https://github.com/kang-jaehyun/Mac-the-researcher --skill results

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Analyze and compare results from completed experiments to identify best configurations and actionable insights.

Core Features & Use Cases

  • Review results from multiple experiments across clusters to understand performance trends.
  • Compare metrics such as final evaluation, training progress, and hyperparameters to highlight differences.
  • Guidance for next steps: surface recommendations for next experiments based on observed results.

Quick Start

Ask Claude Code to compare your last completed experiments on a given cluster.

Frequently Asked Questions about results

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

FAQPage Schema
How do I compare results from completed ML experiments across different clusters?

To compare ML experiment results across clusters, the Skill fetches completed runs, retrieves missing metrics from logs, and generates a comparison table highlighting key hyperparameters and final performance metrics.

What is the best way to analyze metrics from multiple ML experiments to find the top configuration?

Analyzing metrics from multiple ML experiments involves filtering completed runs, extracting final evaluation scores and hyperparameters, and delivering a detailed report that surfaces top configurations and performance patterns.

Can I automatically retrieve missing metrics from experiment logs without manual extraction?

Yes, retrieving missing metrics from experiment logs is handled automatically by leveraging the tail_log function to fetch and parse training progress data for completed runs that lack final evaluation scores.

Does this tool provide recommendations for next steps after comparing ML experiment results?

Yes, the tool provides guidance for next steps by surfacing recommendations for future experiments based on observed performance trends and hyperparameter differences identified during the results comparison.

How do I filter ML experiments to only include completed runs for performance analysis?

To filter ML experiments for performance analysis, the Skill uses the list_experiments function with a status filter set to completed, ensuring only finished runs are included in the cross-cluster comparison report.

What limitations should I expect when comparing ML experiment results across clusters?

Limitations when comparing ML experiment results across clusters include dependency on completed experiment status for filtering and the need to access tail_log data to retrieve any missing training metrics for accurate comparison.