Skip to main content

​
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);
⌘I