extendDataset

---

Function: extendDataset()

function extendDataset(params: object): Promise<object[]>;

Defined in: src/utils/datasets.ts:466

Extends a dataset with synthetically generated data based on the provided parameters.

This function initiates a dataset extension job, waits for it to complete by polling its status, and then returns the content of the extended dataset.

Parameters

params

The parameters for the synthetic dataset extension request.

count?

number

Count

Default

10;

data_types?

| null | ( | "General Query" | "Prompt Injection" | "Off-Topic Query" | "Toxic Content in Query" | "Multiple Questions in Query" | "Sexist Content in Query")[]

Data Types

examples?

string[]

Examples

instructions?

null | string

Instructions

prompt?

null | string

Prompt

prompt_settings?

{ deployment_name?: null | string; echo?: boolean; frequency_penalty?: number; input?: null | string; instructions?: null | string; known_models?: object[]; logprobs?: boolean; max_tokens?: number; model_alias?: string; n?: number; presence_penalty?: number; reasoning_effort?: string; response_format?: | null | { [key: string]: string; }; stop_sequences?: null | string[]; temperature?: number; tool_choice?: | null | string | { function: { name: string; }; type?: string; }; tools?: null | object[]; top_k?: number; top_logprobs?: number; top_p?: number; verbosity?: string; }

Description

Only the model is used.

prompt_settings.deployment_name?

null | string

Deployment Name

prompt_settings.echo?

boolean

Echo

Default

false;

prompt_settings.frequency_penalty?

number

Frequency Penalty

Default

0;

prompt_settings.input?

null | string

Input

prompt_settings.instructions?

null | string

Instructions

prompt_settings.known_models?

object[]

Known Models

prompt_settings.logprobs?

boolean

Logprobs

Default

true;

prompt_settings.max_tokens?

number

Max Tokens

Default

1024;

prompt_settings.model_alias?

string

Model Alias

Default

GPT-4o

prompt_settings.n?

number

N

Default

1;

prompt_settings.presence_penalty?

number

Presence Penalty

Default

0;

prompt_settings.reasoning_effort?

string

Reasoning Effort

Default

medium;

prompt_settings.response_format?

| null | { [key: string]: string; }

Response Format

prompt_settings.stop_sequences?

null | string[]

Stop Sequences

prompt_settings.temperature?

number

Temperature

Default

1;

prompt_settings.tool_choice?

| null | string | { function: { name: string; }; type?: string; }

Tool Choice

prompt_settings.tools?

null | object[]

Tools

prompt_settings.top_k?

number

Top K

Default

40;

prompt_settings.top_logprobs?

number

Top Logprobs

Default

5;

prompt_settings.top_p?

number

Top P

Default

1;

prompt_settings.verbosity?

string

Verbosity

Default

medium;

source_dataset?

| null | { dataset_id: string; dataset_version_index?: null | number; row_ids?: null | string[]; }

Returns

Promise<object[]>

A promise that resolves with the rows of the extended dataset.

Example

const extended_dataset = await extendDataset({
  prompt_settings: {
    model_alias: "GPT-4o mini",
  },
  prompt:
    "Financial planning assistant that helps clients design an investment strategy.",
  instructions:
    "You are a financial planning assistant that helps clients design an investment strategy.",
  examples: ["I want to invest $1000 per month."],
  data_types: ["Prompt Injection"],
  count: 3,
});
console.log("Extended dataset:", extended_dataset);