ZovdaSchool coverage
Nine subjects across three stages
Nine subjects across primary, middle, and high school, one record per question. Subject and stage are fields, so a slice can be pulled without re-exporting the whole set.
Datuva by Zovda · Training data
Question banks with the reasoning attached.
Nine school subjects across three stages, plus 16 vocational categories. School items carry stem, options, answer, worked solution, and knowledge point; vocational items carry category, trade, question type, options, and evidence. Scope is confirmed before delivery; sample packs and data cards are shared after review.
- Nine subjects · three stages
- Worked solutions attached
- Reply within 1 business day
Collection contents
What ships in this collection.
Answer and reasoning
Solution and knowledge point attached
Every school item ships with its answer, a worked solution, and the knowledge point it tests — so a model trains on the path, not only the letter.
Vocational coverage
16 categories with structured options
Sixteen vocational categories, including transport, IT, medicine, and law: prompts with structured A–D options, a primary category, a trade, and an evidence field.
Prompt hygiene
Answer leakage removed
Answer leakage and residual numbering are stripped during processing, so the prompt is just the question.
Record structure
What each record holds.
| Field | What it holds |
|---|---|
stem | The question text. |
options | Answer options, structured rather than embedded in the stem. |
answer | The correct option for school items. |
worked_solution | Step-by-step reasoning that reaches the answer. |
knowledge_point | The concept the item tests. |
subject · stage | School item only: subject and school stage. |
category · trade | Vocational item only: primary category and specific trade. |
confidence · evidence | Vocational item only: tagging confidence and the evidence used. |
Training stage
Recommended use by stage.
| Stage | How it is used |
|---|---|
| SFT | Question-and-solution pairs for supervised fine-tuning. |
| Evaluation | Held-out items with reference answers for scoring. |
| Benchmarking | Per-subject and per-stage slices for comparable runs. |
Delivery
One structure per collection.
| Layer | Contents |
|---|---|
| manifest.json | Every file with name, size, date, and collection ID. |
| UTF-8 JSONL | One record per line for text and annotations. |
| Documents | Data card, field schema, and a PDF spec for each collection. |
| UTF-8 JSONL | One record per line; files grouped by category so a subset can ship first. |
Request access
Request the question bank catalog.
Tell us the subject range, school stages, and volume. We reply within one business day. Sample packs, data cards, and field schemas are shared after your request is reviewed.
FAQ
Before you request a collection.
Can we start with one subject?
Yes. Collections can be sliced by subject, stage, or vocational category, and the rest can follow after acceptance.
Are vocational items answer-keyed?
Vocational packs ship prompts, structured options, category and trade tags, and evidence fields. Answer keys are not part of that pack; scope is confirmed per project.
Can the schema be adapted?
Field names, sampling, and delivery format can be adjusted to your pipeline before production.