dataset-discovery

Query HuggingFace Hub, OpenML, GitHub, and papers for ranked ML datasets.

Updated Aug 27, 2026

Search Tags:#multi-source #search #ranking #dataset-discovery #data-retrieval #data-curation #ml-datasets

Ready to run

npx skills add https://github.com/unstun/dqn10 --skill dataset-discovery-unstun

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: dataset-discovery
Source: https://github.com/unstun/dqn10/tree/main/.claude/skills/dataset-discovery
Command: npx skills add https://github.com/unstun/dqn10 --skill dataset-discovery-unstun

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires requests, and includes scripts (resource) components.

What problem does it solve?

Multi-source ML dataset discovery accelerates research by surfacing relevant datasets from multiple sources where you work.

Core Features & Use Cases

Aggregate results from HuggingFace Hub, OpenML, GitHub, and papers with ranking and deduplication
Provide metadata like modality, license, size, and source to inform selection
Use cases include identifying datasets for image classification, NLP benchmarks, and tabular data tasks

Quick Start

Search for datasets matching your task and review the ranked results.

Frequently Asked Questions about dataset-discovery

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

FAQPage Schema

How do I find relevant ML datasets across HuggingFace Hub, OpenML, and GitHub?▼

To find relevant ML datasets, query multiple sources like HuggingFace Hub, OpenML, GitHub, and papers to receive a deduplicated, ranked list. Results include metadata such as modality, license, size, and source to inform your selection.

How does multi-source dataset discovery work for machine learning research?▼

Multi-source dataset discovery works by querying public APIs from HuggingFace Hub, OpenML, GitHub, and papers, then returning a deduplicated and ranked list. It gracefully handles API failures with sensible fallbacks to ensure continuous search capability.

Can I search for image classification and NLP benchmark datasets using a single query?▼

You can search for image classification, NLP benchmark, or tabular data datasets by applying your task query to multiple sources. The search aggregates results and provides modality, license, and size metadata for informed selection across these tasks.

Do I need internet access to search for datasets across HuggingFace and OpenML?▼

You need internet access to query public APIs from HuggingFace Hub, OpenML, GitHub, and papers for dataset discovery. The process supports optional local workspace outputs but requires connectivity to retrieve the multi-source results.

What is the best way to deduplicate and rank datasets found from multiple sources?▼

The best way to deduplicate and rank datasets is by querying multiple sources simultaneously, which aggregates results from HuggingFace, OpenML, GitHub, and papers into a single ranked list. This provides modality, license, and size metadata for comparison.

How are API failures handled during multi-source dataset search?▼

API failures during dataset search are handled gracefully with sensible fallbacks to ensure the process continues. This allows the multi-source query to return available results from HuggingFace, OpenML, GitHub, or papers even if one source fails.