No proof, no honey.

Open datasets with receipts.

Browse, verify, compose, and export fine-tune-ready datasets from a living graph. Community datasets organized by domain, category, license, format, and proof. Curated on sovereign bare-metal RTX 6000 fleet and RTX 3090 systems.

Domain
Category
Dataset
Receipt
Format
Task
11
Indexed datasets
11
Seed domains
11
Verified entries
11
NAS-backed assets
Registry mass
8.8 GB

Receipt-backed local and NAS-indexed assets with file hashes, sizes, and provenance summaries.

Record count
2,958,120

Known records from registry manifests and mounted NAS inventory passes.

Live domain
defendabledatasets.com

Cloudflare Pages static deploy with SSL enabled.

Why this exists

AI builders need dataset assets they can inspect, cite, hash, package, and reuse. DefendableDatasets starts as a local registry and graph, then grows into the open dataset layer for DefendableCloud and DefendableOS.

Discover datasets

Browse by domain, category, format, license, task target, status, quality, and model family.

Verify provenance

Receipts, hashes, validation state, and source summaries sit beside the data instead of in a forgotten spreadsheet.

Compose fine-tune packs

Select multiple datasets and export manifests, cards, README snippets, and SHA256 files client-side.

Free for the community

The public commons keeps metadata, receipts, hashes, and pack manifests open. Large or sensitive files move through access-controlled delivery.

Built for DefendableCloud members, open to builders

Member-ready datasets can be indexed now, with room for signed receipts, object backends, API access, and fine-tune job handoff.

Doctrine
Receipts over vibes. Datasets are the asset.
Read the docs
Tip jar

Thank you. Tips help support the compute to cook, hash, verify, and publish more datasets for the community.

bc1qnfvjpvv08shp8spdfznwmftkmh8895h56kvfqj