JavaScript

import Dataframer from 'dataframer';

const client = new Dataframer({
  apiKey: 'My API Key',
});

const response = await client.dataframer.datasets.createWithFiles({
  dataset_type: 'SINGLE_FILE',
  name: 'name',
});

console.log(response.id);

{
  "id": "550e8400-e29b-41d4-a716-446655440000",
  "name": "Customer Reviews 2025",
  "description": "Product reviews with files",
  "dataset_type": "SINGLE_FILE",
  "dataset_type_display": "Single File",
  "file_count": 1,
  "folder_count": 0,
  "created_at": "2025-01-15T10:30:00Z",
  "message": "Dataset created with 1 file(s)"
}

Datasets

Create dataset with files (combined operation)

Create a new dataset and upload files in a single atomic operation. Files are validated first, and dataset is only created if validation passes. Empty datasets cannot be created with this endpoint.

Dataset Types:

SINGLE_FILE: Upload one file
- Supported types: CSV, JSON, JSONL only
- Max size: 50MB
- Send file as ‘file’ parameter
- Optional: ‘csv_headers’ as JSON array for CSV files
MULTI_FILE: Upload multiple individual files
- Supported types: TXT, MD, JSON, CSV, JSONL
- Max size: 1MB per file, 50MB total
- Max 1000 files per dataset
- Send files as ‘files’ array parameter
MULTI_FOLDER: Upload files in folders
- Supported types: TXT, MD, JSON, CSV, JSONL
- Max size: 1MB per file, 50MB total
- Minimum 2 folders required, each folder must have at least 1 file
- Max 1000 files total, 20 files per folder
- Send files as ‘files’ array parameter
- Send corresponding folder names as ‘folder_names’ array (parallel arrays)

Request Format:

All requests must be multipart/form-data
Dataset metadata: name, description, dataset_type, config_json
Files depend on dataset_type (see above)

Benefits:

Atomic operation - either everything succeeds or nothing is created
Validation before dataset creation
No empty datasets
Single API call instead of two

POST

api

dataframer

datasets

create

JavaScript

import Dataframer from 'dataframer';

const client = new Dataframer({
  apiKey: 'My API Key',
});

const response = await client.dataframer.datasets.createWithFiles({
  dataset_type: 'SINGLE_FILE',
  name: 'name',
});

console.log(response.id);

{
  "id": "550e8400-e29b-41d4-a716-446655440000",
  "name": "Customer Reviews 2025",
  "description": "Product reviews with files",
  "dataset_type": "SINGLE_FILE",
  "dataset_type_display": "Single File",
  "file_count": 1,
  "folder_count": 0,
  "created_at": "2025-01-15T10:30:00Z",
  "message": "Dataset created with 1 file(s)"
}

Authorizations

Authorization

string

header

required

API Key authentication. Format: "Bearer YOUR_API_KEY"

Body

multipart/form-data

name

string

required

Dataset name (unique within company)

dataset_type

enum<string>

required

Dataset type: SINGLE_FILE, MULTI_FILE, or MULTI_FOLDER

Available options:

SINGLE_FILE,

MULTI_FILE,

MULTI_FOLDER

description

string

Optional dataset description

config_json

string

Optional configuration JSON as string

file

Single file for SINGLE_FILE dataset type

files

file[]

Multiple files for MULTI_FILE or MULTI_FOLDER dataset types. Each file will be sent as a separate form field named 'files'. For MULTI_FOLDER: minimum 2 files required (at least 1 file per folder, across at least 2 folders).

Minimum array length: 1

folder_names

string[]

Folder names for MULTI_FOLDER (parallel array with files). Each folder name is sent as a separate 'folder_names' form field. Minimum 2 unique folder names required.

Minimum array length: 2

csv_headers

string

Optional CSV headers as JSON array for SINGLE_FILE CSV files

Response

Dataset created successfully with files

name

string

required

Minimum string length: 1

dataset_type

enum<string>

required

Available options:

SINGLE_FILE,

MULTI_FILE,

MULTI_FOLDER

string<uuid>

description

string | null

dataset_type_display

string

Minimum string length: 1

config_json

Config json · object

created_at

string<date-time>

updated_at

string<date-time>

created_by

integer

created_by_name

string

Minimum string length: 1

company

string<uuid>

company_name

string

Minimum string length: 1

folders

object[]

Show child attributes

folders.name

string

required

Minimum string length: 1

folders.id

string<uuid>

folders.description

string | null

folders.created_at

string<date-time>

folders.files

object[]

Show child attributes

folders.files.file_name

string

required

Minimum string length: 1

folders.files.file_type

enum<string>

required

Available options:

JSON,

JSONL,

CSV,

MD,

TXT,

PDF

folders.files.storage_uri

string

required

Minimum string length: 1

folders.files.id

string<uuid>

folders.files.file_type_display

string

Minimum string length: 1

folders.files.bytes_size

integer | null

Required range: -9223372036854776000 <= x <= 9223372036854776000

folders.files.sha256

string | null

folders.files.folder_id

string<uuid> | null

folders.files.datasets_id

string<uuid>

folders.files.created_at

string<date-time>

folders.file_count

integer

folder_count

integer

file_count

integer

short_sample_compatibility

Short sample compatibility · object

Show child attributes

short_sample_compatibility.is_short_samples_compatible

boolean

Whether the dataset is compatible with short sample generation

short_sample_compatibility.is_long_samples_compatible

boolean

Whether the dataset is compatible with long sample generation

short_sample_compatibility.reason

string | null

Reason for incompatibility if applicable

List datasets Create dataset from ZIP file (auto-detects structure)

⌘I

Specs

Analysis

DataFramer - Evaluation

Datasets

Files

Generation

Health

Runs

Human Labels

Models

Red Teaming

DataFramer - Runs

DataFramer - Specs

Create dataset with files (combined operation)

Authorizations

Body

Response