autoresearch

Automate machine learning research and iterative experiments using Python.

53|3|Updated Jun 1, 2026

Search Tags:#optimization #benchmarking #machine learning #experimentation #autoresearch

Ready to run

npx skills add https://github.com/AlexWortega/claude-autoresearch-skill --skill autoresearch-alexwortega

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: autoresearch
Source: https://github.com/AlexWortega/claude-autoresearch-skill/tree/main
Command: npx skills add https://github.com/AlexWortega/claude-autoresearch-skill --skill autoresearch-alexwortega

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires pandas, numpy, matplotlib, scikit-learn, torch, and includes scripts (resource) and references (resource) and assets (resource) components.

What problem does it solve?

This Skill automates the process of researching machine learning tasks and running experiments to find the optimal configuration, saving time and effort in the iterative search for the best model setup.

Core Features & Use Cases

Autonomous Research: Automatically surveys existing solutions, benchmarks, and reference code across multiple sources.
Bounded Experiments: Run experiments with constraints on compute budget, time, and resources.
Iterative Search: Utilizes a generational loop to refine experiments and converge on the best configuration.
Use Case: To find the best hyperparameters for a GPT model on a specific dataset, you can use this Skill to automatically run experiments and select the configuration with the highest validation accuracy.

Quick Start

/autoresearch beat the val_bpb baseline on enwik8 with a small GPT, 12 experiments x 5min

Frequently Asked Questions about autoresearch

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

FAQPage Schema

How do I automate machine learning experimentation to find the best hyperparameters?▼

Autonomous ML research automates experimentation by surveying existing solutions on arXiv and GitHub, then executing bounded experiments to iteratively refine and select the best model configuration.

Can I run bounded experiments with strict time and compute budget limits?▼

Yes, bounded experiments support strict constraints on compute budget and time limits, allowing you to execute a defined number of iterations with maximum duration limits per run.

Does this approach use arXiv and GitHub for autonomous ML research?▼

Yes, autonomous ML research surveys existing solutions, benchmarks, and reference code across diverse literature angles using PapersWithCode, arXiv, GitHub, and web search.

What libraries are needed to run automated ML experiments and benchmarking?▼

Automated ML experimentation and benchmarking require Python with pandas, numpy, matplotlib, scikit-learn, and torch installed to execute scripts and validate model configurations.

What is the best way to beat a validation baseline on a specific dataset?▼

To beat a validation baseline, autonomously research reference code and execute a generational loop of bounded experiments to iteratively refine and optimize model configurations.