AI Dataset Preparations - TypeScript SDK

Use your deployed MSP management API for management operations and its inference Gateway for model requests. Check the deployed version and required permissions. Example addresses are placeholders.

client.datasets.preparations manages AI-assisted dataset preparation and labeling tasks. Request bodies use the server wire names shown below.

Overview

Available Operations

Method	Description
clusterOptions()	List clusters selectable for preparation workloads.
availableProviderKeys()	List caller-owned provider keys suitable for labeling.
preview(body)	Preflight the selected cluster.
create(body)	Create a preparation task.
list(body)	Search and page preparation tasks.
get(options)	Read task configuration, progress, and artifacts.
update(id, body)	Update an editable preparation task.
generateRules(options)	Start labeling-rule generation.
startLabeling(body)	Start labeling with optional rule text.
action(body)	Cancel, retry, or delete a task.
downloadUrl(body)	Create a temporary URL for a task artifact.

clusterOptions

List clusters selectable for preparation workloads.

Request

This method has no parameters and sends no request body.

Response

WorkloadClusterOptionVO[].

availableProviderKeys

List caller-owned provider keys suitable for labeling.

Request

This method has no parameters and sends no request body.

Response

PrepProviderKeysVO.

preview

Preflight the selected cluster.

Request

Body requires clusterId.

Response

WorkloadAdmissionPreviewVO.

create

Create a preparation task.

Request

PrepTaskCreateRequest; required fields below.

Response

PrepTaskVO.

list

Search and page preparation tasks.

Request

PrepTaskQueryRequest.

Response

PageResult<PrepTaskVO>.

get

Read task configuration, progress, and artifacts.

Request

{id} required.

Response

PrepTaskDetailVO.

update

Update an editable preparation task.

Request

id and PrepTaskUpdateRequest required.

Response

Updated PrepTaskVO.

generateRules

Start labeling-rule generation.

Request

{id} required.

Response

null.

startLabeling

Start labeling with optional rule text.

Request

preparationId required; optional labelRules.

Response

null.

action

Cancel, retry, or delete a task.

Request

preparationId and action (CANCEL, RETRY, DELETE) required.

Response

null.

downloadUrl

Create a temporary URL for a task artifact.

Request

preparationId plus artifactPath or storageRef.

Response

DownloadUrl.

Field Reference and Examples

Use client.waitForDatasetPreparation(id, { until: "rules_ready" }) after rule generation, and omit until after labeling to wait for completed. The rules_ready target also succeeds when the task has already advanced to a later successful processing phase or completed between polls.

Create body fields

Field	Type	Required	Constraints
taskName	string	Yes	Maximum 255 characters; no control characters.
clusterId	number	Yes	Selected cluster ID.
modelType	string	Yes	llm or vlm.
postTrainingMethod	string	Yes	sft or ref_distill.
preparationMode	string	Yes	base or sss-bench.
providerKeyIds	number[]	Yes	Up to 50 IDs; may be empty when the mode does not require one.
scenario	string	No	Scenario text, maximum 4000 characters.
autoSplitPercent	number	No	1..100.
generatedDatasetFileLines	number	No	Requested generated record count.
processingMode	string	No	auto or manual.
unlabeledDataFiles / evaluationDataFiles	DatasetFileItem[]	No	Up to 100 items each.
tBenchConfig	object	No	Benchmark-mode options.

DatasetFileItem accepts fileName, filePath, fileSize, storageRef, sourceType, and preparationGroup. Update accepts the same configuration fields as create, but all are optional.

Key response fields: PrepTaskVO includes id, taskName, clusterId, clusterName, status, progress, progressInfo, savedDatasetId, savedDatasetName, error, resourcePreview, createdAt, and updatedAt. PrepTaskDetailVO additionally includes configuration, selected keys, input files, artifacts, rules, execution summary, and result summary.

const clusters = await client.datasets.preparations.clusterOptions();
const providerKeys = await client.datasets.preparations.availableProviderKeys();
const resource = await client.datasets.preparations.preview({ clusterId: 1 });

const task = await client.datasets.preparations.create<{ id: number }>({
  taskName: "prepare-example-dataset",
  clusterId: 1,
  modelType: "llm",
  postTrainingMethod: "sft",
  preparationMode: "base",
  providerKeyIds: [],
});
const page = await client.datasets.preparations.list({
  taskName: "prepare-example",
  status: ["pending", "completed"],
  pageNum: 1,
  pageSize: 20,
});
const detail = await client.datasets.preparations.get({ id: task.id });
const updated = await client.datasets.preparations.update(task.id, {
  taskName: "new-name",
});
await client.datasets.preparations.generateRules({ id: task.id });
await client.datasets.preparations.startLabeling({
  preparationId: task.id,
  labelRules: "Return one concise label.",
});
const artifact = await client.datasets.preparations.downloadUrl({
  preparationId: task.id,
  artifactPath: "outputs/result.jsonl",
});
await client.datasets.preparations.action({
  preparationId: task.id,
  action: "CANCEL",
});

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