What this category's data looks like
This page addresses business basis queries for primary and secondary school admissions organizations. Core materials include admission plans, application material checklists, and admission lists. The admission plan contains admission scope, time nodes, and admission rules, with fields including plan name, publication date, and signature/seal; the application material checklist lists items such as household registration, property, and residence permit, with fields including material name, number of copies, and receipt date; the admission list contains student names, admission categories, and publicity dates, with fields including list version, signature, and receipt. The three types of materials are updated by admission season: the plan is issued first, the checklist is adjusted along with applications, and the list is finalized after admission. After the materials enter this function, the plan serves as the rule basis, the checklist as the material verification basis, and the list as the result basis; together they support basis entries that can be traced back to the original text.
Configuration and implementation checks
This part is executed by application settings (response limit, etc.), which output the length of a single answer and the range of context that can be accommodated. "Basis entries that can be traced back to the original text" are then completed by the model. The response limit is controlled by a slider for the length of a single answer; if set too low, it may truncate, and if set too high, it increases consumption.
| Object | Check | Verification action |
|---|---|---|
| Admission plan | The plan name, publication date, and signature/seal must be consistent with the current version, and the entries must match the original text | Check the position of the response limit slider in application settings, and use maxResponse to observe whether the end of a long answer is complete |
| Application material checklist | The material name, number of copies, and receipt date must be consistent between the two materials; missing signatures or receipts must be marked | Check whether the checklist entries and the plan references come from the same source, and record the reference text tokens corresponding to quoteMaxToken |
| Admission list | Student names, admission categories, and publicity dates must correspond to the plan rules; unsynchronized versions must be marked | Check the list version and publicity receipt, record the number of reference entries separately, and do not compare it for equality with quoteMaxToken |
| Context range | The context range that can be accommodated in a single answer must cover cross-material basis questions | Use maxContext to check the context range and confirm that the answer can be traced back to specific material entries |
Three common errors
- Phenomenon: The name, quantity, and date of the same record do not match between the plan and the list. Cause: The two materials are updated separately, and no field-level comparison is performed.
- Phenomenon: The checklist version is not synchronized, and the application material requirements are inconsistent with the current plan. Cause: The update rhythms differ, and there is no version verification step.
- Phenomenon: The list lacks signatures or receipts, and basis entries cannot be traced back to the original text. Cause: The finalization process did not collect all signatures/seals and receipts.
How to verify the setup
This function's acceptance: ask a cross-material basis question that requires a long answer, see whether the end of the answer is complete, and use maxResponse to check whether the response limit matches the page requirements.
Business test: ask a cross-material basis question, see whether the answer gives traceable sources, and confirm that the number of reference entries and quoteMaxToken are recorded separately.
- Check the response limit switch and slider position in application settings, and confirm that they match the answer length required by the page.
- Compare usage from two runs to confirm that the change in single-run usage after adjusting the limit is within expectations.