perf: dataset import params code (#4875)

* perf: dataset import params code

* perf: api dataset code

* model
This commit is contained in:
Archer
2025-05-23 10:40:25 +08:00
committed by GitHub
parent 9af92d1eae
commit fae76e887a
23 changed files with 366 additions and 295 deletions

View File

@@ -74,6 +74,15 @@ export const createCollectionAndInsertData = async ({
llmModel: getLLMModel(dataset.agentModel)
});
const chunkSplitter = computeChunkSplitter(createCollectionParams);
if (trainingType === DatasetCollectionDataProcessModeEnum.qa) {
delete createCollectionParams.chunkTriggerType;
delete createCollectionParams.chunkTriggerMinSize;
delete createCollectionParams.dataEnhanceCollectionName;
delete createCollectionParams.imageIndex;
delete createCollectionParams.autoIndexes;
delete createCollectionParams.indexSize;
delete createCollectionParams.qaPrompt;
}
// 1. split chunks
const chunks = rawText2Chunks({