Skip to main content

Datasets - TypeScript SDK

client.datasets manages training and evaluation datasets. The high-level upload(...) workflow signs, transfers, commits, and registers local files in one call.

Overview​

Available Operations​

MethodDescription
upload(files, options)Upload one file, a directory, or a file list and create a Dataset.
list(options?)Search and page datasets.
get(id)Read one Dataset.
preview(id, options?)Preview parsed records without downloading the object.
downloadUrl(body)Create a temporary URL for one Dataset file.
update(id, body)Update Dataset metadata or committed file references.
delete(id)Delete an unreferenced Dataset.

upload​

Upload one file, a directory, or a file list and create a Dataset.

Request​

files; required name, datasetType, trainingCategory; optional onProgress.

Response​

Created DatasetVO.

list​

Search and page datasets.

Request​

Optional datasetName, datasetType, trainingCategory, orderBy, orderDirection, pageNum, pageSize, scope.

Response​

PageResult<DatasetVO>.

get​

Read one Dataset.

Request​

id required.

Response​

DatasetVO.

preview​

Preview parsed records without downloading the object.

Request​

id required; optional {limit, cursor}.

Response​

DatasetPreviewVO.

downloadUrl​

Create a temporary URL for one Dataset file.

Request​

Body requires datasetId and zero-based fileIndex.

Response​

DownloadUrl.

update​

Update Dataset metadata or committed file references.

Request​

id and body required.

Response​

Updated DatasetVO.

delete​

Delete an unreferenced Dataset.

Request​

id required.

Response​

null.

Field Reference and Examples​

upload(...) options

OptionTypeRequiredValues
namestringYesNon-empty Dataset name.
datasetTypestringYestraining or evaluation.
trainingCategorystringYessft-llm, dpo-llm, sft-vlm, dpo-vlm, or cpt-llm.
onProgress(progress: UploadProgress) => voidNoSame progress object as My Model upload.

Update body fields: optional datasetName, type, trainingCategory, and files. Key DatasetVO fields: id, datasetName, type, trainingCategory, files, fileSizeTotal, owner, createdAt, updatedAt. DatasetPreviewVO contains items, totalRecords, datasetRevision, hasNext, and nextCursor.

const dataset = await client.datasets.upload<{ id: number }>("./train.jsonl", {
  name: "example-training-dataset",
  datasetType: "training",
  trainingCategory: "sft-llm",
});

const page = await client.datasets.list({
  datasetName: "example",
  datasetType: ["training"],
  trainingCategory: ["sft-llm"],
  orderBy: "createdAt",
  orderDirection: "DESC",
  pageNum: 1,
  pageSize: 20,
  scope: "self",
});
const detail = await client.datasets.get(dataset.id);
const preview = await client.datasets.preview(dataset.id, { limit: 20 });
const download = await client.datasets.downloadUrl({
  datasetId: dataset.id,
  fileIndex: 0,
});
const updated = await client.datasets.update(dataset.id, {
  datasetName: "renamed-dataset",
});
await client.datasets.delete(dataset.id);

All methods use the shared authentication and typed error behavior described in Response Conventions and Retry and Security.