Explore the data.
Every number, one page.
A live statistical profile generated from the actual exports - coverage rates, distributions, numeric ranges and the full file manifest. Regenerate with python scripts/data_profile.py --out docs/data/profile.json.
—
Anime works
—
QA pairs
—
Instruction examples
—
RAG chunks
—
Characters
—
Episodes
Profile —. Data is read straight from exports/.
OverviewCoverage & quality signals
Works with an episode count—
Works with mean score—
QA pairs grounded in record fields—
Instruction examples validated—
Provenance rows carry license notes—
Year range—–—
Quality tiers & confidence
Split proportions
ShapeWhat kind of works
Media types
Status
TimeWorks by release year
| Year | Works |
|---|
Top 15 years. The dataset spans every age of the medium, weighted toward recent productions.
TagsGenres & themes
Top genres
Top themes
Top studios
RangesNumeric distributions
| Metric | Count | Min | Q1 | Median | Q3 | Max | Mean |
|---|
Quartiles are inclusive. Scores are 0–100 unless a source disclosed otherwise.
TrainablesQA, instructions & RAG anatomy
QA pair types (top 10)
Instruction shapes
Character roles
ManifestExported files
| File | Size |
|---|