laksh344

Claude Ml Skill

Name: Claude Ml Skill
Author: laksh344

AI community

SOURCE https://github.com/laksh344/Claude-ml-skill

Description

**An autonomous Grandmaster-level skill for Claude that turns it into a top-1% ML competition partner.**

Installation

This entry records only its repository, not the path inside it, so there is no exact command to give. Open the source below and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

README

🏆Claude Skill for ML/AI Competitions

**An autonomous Grandmaster-level skill for Claude that turns it into a top-1% ML competition partner.**

Built with insights from 2025–2026 research (TRM, SOAR, CompressARC, Gemma 4 MoE, agentic RAG) and battle-tested across 12 competition domains.

[![Version](https://img.shields.io/badge/version-2.0.0-blue.svg)](https://github.com/YOUR_USERNAME/kaggle-ml-skill) [![License: MIT](https://img.shields.io/badge/License-MIT-green.svg)](LICENSE) [![Domains](https://img.shields.io/badge/domains-12-orange.svg)](#supported-competition-types) [![Files](https://img.shields.io/badge/reference_files-17-purple.svg)](#file-overview)

📊 Benchmarks: With Skill vs Without Skill

We evaluated Claude on **6 real competition prompts** spanning tabular, audio, LLM fine-tuning, ARC-AGI, time-series, and math reasoning. Each response was scored 0–10 across 8 quality dimensions by comparing the output against known winning strategies from top Kaggle solutions.

Radar Comparison

Benchmark Radar Chart — With Skill vs Without

Aggregate Scores

Evaluation Criteria	🔴 Without Skill	🟢 With Skill	Δ Improvement
Leakage Detection	2/10	9/10	+350%
CV Strategy Correctness	4/10	9/10	+125%
2026 Research Awareness	1/10	10/10	+900%
Metric-Specific Optimization	3/10	9/10	+200%
Ensemble Strategy	2/10	8/10	+300%
Code Completeness (E2E Runnable)	5/10	9/10	+80%
Domain-Specific Techniques	3/10	10/10	+233%
Submission Validation	1/10	9/10	+800%

Overall Average	2.6/10	9.1/10	+250%