This repository offers a collection of client-side (in-browser) AI demonstrations, targeting developers and researchers. It showcases practical integration of AI models directly within web applications for tasks like text generation, summarization, and sentiment analysis, utilizing Chrome's experimental APIs or lightweight models such as Google's Gemma. The primary benefit is enabling AI functionalities without server-side inference.

How It Works

The project employs diverse client-side AI strategies. Demos leverage Chrome's experimental Prompt and Summarization APIs for specific tasks. Others integrate generic frameworks like Transformers.js for sentiment analysis and toxicity detection, or utilize Google's Gemma model via MediaPipe for generative AI. A core architectural choice is in-browser inference, often via web workers, to reduce latency and enhance privacy.

Quick Start & Requirements

This is a collection of demos, not a single installable application. Requirements vary but generally include a modern web browser. Specific demos utilize: Chrome's experimental Prompt/Summarization APIs, Transformers.js, Google's Gemma via MediaPipe, OpenWeatherMap API, and Node.js with the Google AI SDK for SSE streaming. Demos are typically run by navigating to their respective pages.

Highlighted Details

• weather-ai: Generates weather descriptions using Chrome's Prompt API and OpenWeatherMap data.
• prompt-api-playground & summarization-api-playground: Showcase Chrome's experimental AI APIs.
• perf-client-side-gemma-worker: Demonstrates client-side Gen AI performance with Gemma 2 via MediaPipe in a web worker.
• right-click-for-superpowers: Adds LLM utilities (summarization, translation) to webpages using Gemma 2B via MediaPipe.
• product-reviews: Features client-side sentiment analysis, toxicity, and rating assessment using Gemma 2B via MediaPipe and Transformers.js.
• gemini-node-sse: Illustrates streaming Gemini content via Server-Sent Events (SSE) with Node.js.

Maintenance & Community

Information on contributors, sponsorships, community channels, or roadmaps is not provided in the README content.

Licensing & Compatibility

The license type and compatibility notes for commercial use are not specified in the README content.

Limitations & Caveats

Functionality relies on experimental Chrome APIs, subject to change. Client-side AI performance is constrained by browser capabilities and user hardware. Demos may require specific API keys or data sources to function fully.