perf: backup import (#4866)

* i18n

* remove invalid code

* perf: backup import

* backup tip

* fix: indexsize invalid

packages/global/common/error/code/dataset.ts

@@ -27,7 +27,7 @@ const datasetErr = [
   },
   {
     statusText: DatasetErrEnum.unExist,
-    message: 'core.dataset.error.unExistDataset'
+    message: i18nT('common:core.dataset.error.unExistDataset')
   },
   {
     statusText: DatasetErrEnum.unExistCollection,

packages/global/core/dataset/api.d.ts

@@ -147,6 +147,7 @@ export type PushDatasetDataProps = {
   collectionId: string;
   data: PushDatasetDataChunkProps[];
   trainingType?: DatasetCollectionDataProcessModeEnum;
+  indexSize?: number;
   autoIndexes?: boolean;
   imageIndex?: boolean;
   prompt?: string;

packages/global/core/dataset/constants.ts

@@ -120,6 +120,8 @@ export const DatasetCollectionSyncResultMap = {
 export enum DatasetCollectionDataProcessModeEnum {
   chunk = 'chunk',
   qa = 'qa',
+  backup = 'backup',
+
   auto = 'auto' // abandon
 }
 export const DatasetCollectionDataProcessModeMap = {
@@ -131,6 +133,10 @@ export const DatasetCollectionDataProcessModeMap = {
     label: i18nT('common:core.dataset.training.QA mode'),
     tooltip: i18nT('common:core.dataset.import.QA Import Tip')
   },
+  [DatasetCollectionDataProcessModeEnum.backup]: {
+    label: i18nT('dataset:backup_mode'),
+    tooltip: i18nT('dataset:backup_mode')
+  },
   [DatasetCollectionDataProcessModeEnum.auto]: {
     label: i18nT('common:core.dataset.training.Auto mode'),
     tooltip: i18nT('common:core.dataset.training.Auto mode Tip')
@@ -154,7 +160,6 @@ export enum ImportDataSourceEnum {
   fileLocal = 'fileLocal',
   fileLink = 'fileLink',
   fileCustom = 'fileCustom',
-  csvTable = 'csvTable',
   externalFile = 'externalFile',
   apiDataset = 'apiDataset',
   reTraining = 'reTraining'

packages/global/core/dataset/training/utils.ts

@@ -118,7 +118,7 @@ export const computeChunkSize = (params: {
     return getLLMMaxChunkSize(params.llmModel);
   }
 
-  return Math.min(params.chunkSize || chunkAutoChunkSize, getLLMMaxChunkSize(params.llmModel));
+  return Math.min(params.chunkSize ?? chunkAutoChunkSize, getLLMMaxChunkSize(params.llmModel));
 };
 
 export const computeChunkSplitter = (params: {

packages/global/core/dataset/type.d.ts

@@ -175,6 +175,7 @@ export type DatasetTrainingSchemaType = {
   q: string;
   a: string;
   chunkIndex: number;
+  indexSize?: number;
   weight: number;
   indexes: Omit<DatasetDataIndexItemType, 'dataId'>[];
   retryCount: number;