OpenAI Provider

Provider for integrating with OpenAI's GPT models. Supports various models like GPT-3.5, GPT-4, etc.

Features

• Support for GPT models (gpt-3.5-turbo, gpt-4, gpt-4-turbo, etc.)
• Function calling feature support
• Streaming response support
• Control of various model parameters (temperature, top_p, etc.)

Installation

npm install @robota-sdk/openai

Usage

Basic Usage

import { Robota, OpenAIProvider } from 'robota';
import OpenAI from 'openai';

// Create OpenAI client
const openaiClient = new OpenAI({
  apiKey: process.env.OPENAI_API_KEY,
  organization: process.env.OPENAI_ORGANIZATION // Optional
});

// Initialize OpenAI provider
const provider = new OpenAIProvider({
  model: 'gpt-4',
  temperature: 0.7,
  client: openaiClient
});

// Connect provider to Robota instance
const robota = new Robota({ provider });

// Execute
const result = await robota.run('Hello! How is the weather today?');
console.log(result);

Using Function Calling

import { Robota, OpenAIProvider } from 'robota';
import OpenAI from 'openai';

// Create OpenAI client
const openaiClient = new OpenAI({
  apiKey: process.env.OPENAI_API_KEY
});

// Initialize OpenAI provider
const provider = new OpenAIProvider({
  model: 'gpt-4',
  client: openaiClient
});

// Create Robota instance
const robota = new Robota({ provider });

// Register functions
robota.registerFunctions({
  getWeather: async (location: string) => {
    // Weather lookup logic
    return { temperature: 22, condition: 'Clear', location };
  }
});

// Execute
const result = await robota.run('What is the weather in Seoul?');
console.log(result);

Handling Streaming Responses

import { Robota, OpenAIProvider } from 'robota';
import OpenAI from 'openai';

// Create OpenAI client
const openaiClient = new OpenAI({
  apiKey: process.env.OPENAI_API_KEY
});

// Initialize OpenAI provider
const provider = new OpenAIProvider({
  model: 'gpt-4',
  client: openaiClient
});

// Create Robota instance
const robota = new Robota({ provider });

// Handle streaming response
const stream = await robota.stream('Tell me a long story.');

for await (const chunk of stream) {
  process.stdout.write(chunk.content || '');
}

Provider Options

OpenAI provider supports the following options:

interface OpenAIProviderOptions extends ProviderOptions {
  // Required options
  model: string;       // Model name to use (e.g., 'gpt-4')
  client: OpenAI;      // OpenAI client instance

  // Optional options
  temperature?: number;  // Randomness/creativity of the response (0~1)
  maxTokens?: number;    // Maximum number of tokens to generate
  stopSequences?: string[]; // Stop sequences during generation
  topP?: number;         // Top-p sampling (0~1)
  presencePenalty?: number; // Discourage topic repetition (-2.0~2.0)
  frequencyPenalty?: number; // Discourage word repetition (-2.0~2.0)
  responseFormat?: {      // Specify response format
    type: 'text' | 'json_object'
  };
  user?: string;         // User identifier
}

Model Selection Guide

Model	Features	Recommended Use Cases
gpt-3.5-turbo	Fast and economical	Simple conversation, basic information processing
gpt-4	More accurate and better reasoning	Complex problem solving, code writing
gpt-4-turbo	Latest data and enhanced features	Tasks requiring recent information
gpt-4-vision	Image understanding capability	Visual content analysis

Notes

• Always manage your OpenAI API key through environment variables or a secure secret management system.
• Setting the maxTokens option appropriately is recommended for cost management.
• Some features (e.g., token usage tracking) may be limited in streaming mode.