What this category's data looks like
Research project annual progress and acceptance materials mainly include three categories: annual progress reports, mid-term review materials, and acceptance reports. Annual progress reports are updated periodically, with fields covering task name, indicators, completed quantity, start and end dates, and responsible person's signature; mid-term review materials summarize stage results and deviation explanations before the milestone; acceptance reports are formed at project completion, with fields including results list, fund usage, participating personnel, and acceptance opinions. When these three categories of materials enter the model access and configuration stage, they respectively serve as input sources for dialogue and retrieval: progress reports are used to generate drafts, mid-term materials are used to check deviations, and acceptance reports are used to summarize the final draft. Field characteristics emphasize consistency of names, quantities, dates, versions, and signature receipts, and the update rhythm follows project milestones and material version changes.
Configuration and implementation checks
This step is executed by the server-side model channel configuration, outputting model binding relationships and channel responses. Signable material drafts and handover records are then completed by subsequent 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, and type labels need to be checked in the model configuration.
| Object | Check | Verification action |
|---|---|---|
| Annual progress report | Names, quantities, dates, and signatures are complete and consistent with the mid-term review materials | Check the task name, completed quantity, start and end dates, and responsible person's signature in the report; compare whether the model name returned by the channel is consistent with the name listed in the document |
| Mid-term review materials | Dates, versions, and receipts of stage results and deviation explanations are complete | Check the stage result names, quantities, review dates, and receipts; check whether the model identifier corresponds to the accessed service |
| Acceptance report | Fields for the results list, participating personnel, and acceptance opinions are complete, and versions are synchronized | Check the result names, quantities, acceptance dates, and signatures; check whether the channel configuration and key source are consistent with the accessed service |
| Model binding relationships | Language, indexing, reranking, speech synthesis, and speech recognition are respectively bound to the corresponding steps | Check the type labels in the model configuration; confirm that the retrieval step is bound to the indexing model and the dialogue step is bound to the language model |
Three common errors
- Symptom: The names and quantities of the same record do not match between the annual progress report and the mid-term review materials. Cause: The two materials were drafted separately and were not checked against the same field template.
- Symptom: The acceptance report version is not synchronized, and old-version signatures are mixed with new-version content. Cause: After the materials were updated, the version number and date were not unified, and the receipt was not checked during handover.
- Symptom: Retrieval reports errors related to the indexing model, or the answer style changes abruptly after switching models. Cause: The model type is inconsistent with the step binding, or the model identifier does not match the name returned by the channel.
How to verify the setup
Acceptance of this function: run one dialogue and one retrieval, and check the model identifier in the response; confirm that the language model and indexing model are bound correctly respectively, and check whether the values of maxContext, maxTokens, and quoteMaxToken conform to the model configuration.
Business test: draft one copy according to the template and complete one handover, and see whether missing items are blocked.
- In the model configuration, check whether the model type labels correspond to the accessed steps, and confirm that language, indexing, reranking, speech synthesis, and speech recognition each take their proper place.
- Check whether the channel configuration and key source are consistent with the accessed service, and compare the model name returned by the channel with the name listed in the document.