teapotai/instruct-teapot

Teapot AI Instruct Documentation

The teapot-instruct model is fine-tuned for optimized web-based Retrieval-Augmented Generation (RAG) question answering. Designed for browser environments, it efficiently retrieves, synthesizes, and answers queries based on web content.

Overview

Teapot AI Instruct is ideal for:

• Question Answering (QA): Extracting answers from web pages.
• RAG: Combining retrieval and generation to answer complex questions.
• Summarization: Condensing web content.
• Translation: Translating between langauges.

Example Usage

import { pipeline, env } from '@xenova/transformers';

const generative_model = await pipeline('text2text-generation', 'teapotai/instruct-teapot');

// Example context from an article (similar to SQuAD)
const context = `
    The Amazon Rainforest, also known as Amazonia, is a large tropical rainforest occupying the drainage basin of the Amazon River 
    and its tributaries in northern South America, and covering an area of 6,000,000 square kilometers (2,300,000 sq mi). 
    Comprising about 40% of Brazil's total area, it is bordered by the Guiana Highlands to the north, the Andes Mountains to the west, 
    the Brazilian Central Plateau to the south, and the Atlantic Ocean to the east.
`;

const query = "What percentage of Brazil's total area does the Amazon Rainforest cover?";
const result = await generative_model(context+query);

console.log(result); // 40%

Features

• Web-Optimized: Tailored for browser-based QA and generation tasks.
• RAG Capability: Integrates retrieval with generative responses.
• Lightweight Deployment: Low latency, high performance in browser environments.
• Zero Setup: Client-side integration with no server requirements.

Use Cases

• Browser-based RAG: Real-time QA within web apps.
• Web Scraping and QA: Generate answers from web content.
• Summarization: Summarize web pages efficiently.

Model Performance

Teapot AI Instruct is fine-tuned with a large web-based corpus, excelling in:

• Accurate, context-aware answers.
• Summarization of long-form text with relevant content focus.

Integration with Browser-Based LLM Agents

Teapot AI Instruct integrates seamlessly with agents like Teapot AI for:

• Web scraping.
• Real-time browser-based QA.
• Summarization and code completion.