What this category's data looks like
This category's data includes grading and classification tables, hazard source lists, and safety responsibility statements. Sources are department submissions, laboratory self-inspections, and functional department reviews. The update cadence is triggered by changes in experimental projects, personnel adjustments, and annual reviews. The grading and classification table records laboratory name, room number, level, hazard source category and quantity, responsible person, date, and signature; the hazard source list records name, quantity, storage location, risk level, and control measures; the safety responsibility statement records the responsible person, scope of duties, signature, and receipt. When the materials enter the model access and configuration function, their roles differ: the grading and classification table and hazard source list serve as knowledge sources for dialogue and retrieval, while the safety responsibility statement serves as the basis for handover records. Only when fields are complete, the responsible person is clear, and versions and dates are consistent may they proceed to the next step of sign-off.
Configuration and implementation checks
This step is executed by the server-side model channel configuration, outputting model binding relationships and channel responses. "Draft materials and handover records that can be issued" are completed subsequently by AI nodes, code nodes, or manual work. Models are divided into five categories: language, indexing, reranking, speech synthesis, and speech recognition. Different steps bind different model types.
| Object | Check | Verification action |
|---|---|---|
| Grading and classification table | Whether the laboratory name, room number, level, hazard source category and quantity, responsible person, date, and signature are complete, and whether the name, quantity, and date are consistent with those in the hazard source list | Check item by item against the hazard source list for name, quantity, and date; if signature or receipt is missing, return it; in the model configuration, check the type annotation and the model name returned by the channel |
| Hazard source list | Whether the hazard source name, quantity, storage location, risk level, control measures, responsible person, and date are complete, and whether the level and quantity are consistent with those in the grading and classification table | Check the name, quantity, and date of the same record in both materials; if the versions are not synchronized, mark it; confirm that the indexing model binding corresponds to the retrieval step |
| Safety responsibility statement | Whether the responsible person, scope of duties, signature, receipt, and date are complete, and whether the version is synchronized with the grading and classification table | Check whether the signature and receipt are complete; if any item is missing, block it; confirm that the language model binding corresponds to the dialogue step, and that the channel and key are consistent with the accessed service |
| Model binding relationships | Whether the five model categories—language, indexing, reranking, speech synthesis, and speech recognition—correspond to the steps to be used, and whether the model identifiers are consistent with the names listed in the documentation | In the model configuration, check the type annotation against the model name returned by the channel; when references are involved, check quoteMaxToken and maxContext; for tool calls, check toolChoice and maxTokens |
Three common errors
- Symptom: The name, quantity, and date of the same record do not match between the grading and classification table and the hazard source list, and it is returned during handover. Cause: The two materials were drafted separately, without cross-checking and version synchronization.
- Symptom: The safety responsibility statement lacks a signature or receipt, and the next step cannot sign off. Cause: The handover record was not checked item by item according to the template, and the responsible person did not confirm.
- Symptom: Retrieval reports an indexing model-related error, or the answer style changes abruptly after switching models. Cause: The model type and step binding are inconsistent, and the channel and key do not match the accessed service.
How to verify the setup
This function's acceptance: run one dialogue and one retrieval, check that the model identifiers in the responses are consistent with the names listed in the documentation, and check that the configurations of maxContext, maxTokens, quoteMaxToken, and toolChoice correspond to the steps.
Business test: draft one according to the template and complete one handover, and see whether missing items are blocked.
- In the model configuration, check the type annotations of the five model categories—language, indexing, reranking, speech synthesis, and speech recognition—confirm that they correspond to the steps to be used, and compare them with the model names returned by the channel.
- Check that the channel and key are consistent with the accessed service, confirm that the model identifiers are consistent with the names listed in the documentation, and record the tokens and array count of the referenced text.