Create Dataset Version from File

POST /api/public/v2/dataset-versions/from-file

cURL

csvFile

curl --request POST \
  --url https://api.promptlayer.com/api/public/v2/dataset-versions/from-file \
  --header 'Content-Type: application/json' \
  --header 'X-API-KEY: <api-key>' \
  --data '
{
  "dataset_group_id": 123,
  "file_name": "support-eval.csv",
  "file_content_base64": "cXVlc3Rpb24sZXhwZWN0ZWRfYW5zd2VyCldoYXQgaXMgUHJvbXB0TGF5ZXI/LEEgcHJvbXB0IG1hbmFnZW1lbnQgcGxhdGZvcm0="
}
'

201

{
  "success": true,
  "message": "Dataset version creation job queued",
  "dataset_id": 456
}

Legacy Dataset, Evaluation, and Report endpoints are deprecated for new workflows. Use the Tables API for new dataset import, evaluation, scoring, recalculation, and reporting workflows.

Create a dataset version by submitting a base64-encoded CSV or JSON file. PromptLayer queues the file for asynchronous processing and creates a draft dataset while the job runs.

Behavior Notes

• The draft dataset starts with version_number = -1 and receives a real version number after processing succeeds.
• The dataset_version_created_by_file webhook event is sent when processing succeeds.
• The dataset_version_created_by_file_failed webhook event is sent when processing fails.
• Decoded file content must be 100MB or smaller.
• Failed drafts are automatically cleaned up.

Related

• Create Dataset Version from Request History
• Create Draft Dataset Version
• Create from History

Authorizations

X-API-KEY · string · header · required

Body

application/json

dataset_group_id · integer · required

ID of the dataset group where the new version will be created.

Required range: x >= 1

file_name · string · required

Name of the CSV or JSON file. Must end with .csv or .json.

Required string length: 1 - 255

file_content_base64 · string · required

Base64-encoded file content. Maximum decoded file size is 100MB.

Minimum string length: 1

Response

201

application/json

Dataset version creation job queued.

success · enum<boolean> · required

Available options: true

message · string · required

dataset_id · integer · required

ID of the draft dataset created for asynchronous processing.