Workspace / Dataset explorer
Connecting…

HOMEY KNOWLEDGE COLLECTION

Dataset explorer

Source pages, arranged into grounded training examples.

Source pages—Wikipedia articles
Collected examples—Awaiting first update
Approved examples—Reviewed for training
Excerpts remaining—Queue is loading
Checking pipelineConnecting to Cloudflare
WikipediaQwen3 · 4BDataset

Collected examples 0

Generated drafts include source evidence. Approval requires a human review.

Waiting for data
Training examples with questions, relations, sources, and review status
QUESTION & ANSWERSTRUCTURED FACTSOURCE / SPLITREVIEWInspect

Waiting for the first examples

As pages are processed, their questions, answers, and evidence will appear here.

0 examples

Splits are assigned by Wikipedia article, keeping excerpts from the same article together. Related facts across different articles still need a leakage review before evaluation.

Homey Data · Source-grounded collectionWikipedia text and adapted examples: CC BY-SA 4.0
EXAMPLE DETAILS

Operator access

Viewing is public. Adding pages and reviewing examples require your private operator key.

The key is kept only in this tab’s memory.

Add source pages

The collector fetches Wikipedia text and sends each excerpt to the deployed Qwen3 model.

One English Wikipedia article per line. Up to 100 at once.

Export training records

JSON Lines includes the structured fact, question, answer, evidence, source licence, and review status.

Drafts have passed structural checks, not a factual accuracy evaluation.