hf-tgi-api-helper

A simple npm package to interact with Hugging Face's Text Generation Inference (TGI) model. It provides easy-to-use functions for generating, summarizing, and rephrasing text using the TGI model API.

Features

• Generate Text: Interact with the TGI model to generate text based on a prompt.
• Summarize Emails: Provide a summary of email content.
• Rephrase Emails: Rephrase email content for professionalism and conciseness.

Installation

Install the package via npm:

npm install hf-tgi-api-helper

Usage

1. Generate Text

You can use the generateText function to send a prompt and system prompt to the TGI model and receive generated text.

const { generateText } = require('hf-tgi-api-helper');

(async () => {
  const result = await generateText('Translate this text to French:', 'Bonjour, comment allez-vous?');
  console.log(result); // "Hello, how are you?"
})();

2. Summarize an Email

You can summarize email content using the summarizeEmail function. The system prompt defaults to "Summarize the following email:" but can be customized.

const { summarizeEmail } = require('hf-tgi-api-helper');

(async () => {
  const summary = await summarizeEmail('Dear John, I hope this email finds you well...');
  console.log(summary); // "Summary of the email content..."
})();

3. Rephrase an Email

Use the rephraseEmail function to rephrase email content for conciseness and professionalism. The system prompt defaults to "Rephrase the following email to be more concise and professional:" but can be customized.

const { rephraseEmail } = require('hf-tgi-api-helper');

(async () => {
  const rephrased = await rephraseEmail('Dear Sir, I am writing to inquire...');
  console.log(rephrased); // "Dear Sir, I would like to inquire..."
})();

API Reference

generateText(prompt, systemPrompt, config, url)

• Description: Interact with the TGI model to generate text.
• Parameters:
  • prompt (string): The main user input.
  • systemPrompt (string): A predefined system prompt (e.g., "Translate this text to French:").
  • config (object, optional): Configuration options for the TGI model, including:
    • max_new_tokens (number, default: 50): Maximum number of tokens to generate.
    • temperature (number, default: 0.7): Controls randomness in generation (higher = more random).
  • url (string, optional): URL of the TGI server. Defaults to the TGI_SERVER_URL environment variable.
• Returns: Promise<string> — The generated text.

summarizeEmail(emailContent, systemPrompt, config, url)

• Description: Summarize email content using the TGI model.
• Parameters:
  • emailContent (string): The content of the email to summarize.
  • systemPrompt (string, optional): Custom system prompt (default: "Summarize the following email:").
  • config (object, optional): Configuration options (same as generateText).
  • url (string, optional): URL of the TGI server. Defaults to the TGI_SERVER_URL environment variable.
• Returns: Promise<string> — The summarized email.

rephraseEmail(emailContent, systemPrompt, config, url)

• Description: Rephrase email content for conciseness and professionalism using the TGI model.
• Parameters:
  • emailContent (string): The content of the email to rephrase.
  • systemPrompt (string, optional): Custom system prompt (default: "Rephrase the following email to be more concise and professional:").
  • config (object, optional): Configuration options (same as generateText).
  • url (string, optional): URL of the TGI server. Defaults to the TGI_SERVER_URL environment variable.
• Returns: Promise<string> — The rephrased email.

Environment Variables

• TGI_SERVER_URL: The URL of the TGI model server. This can be overridden by passing a custom URL to the functions.