Overview:

Chrome AI is a demo project that demonstrates the capabilities of running large language models directly within the browser using modern web technologies. This project highlights the ability to leverage advanced AI functionalities without requiring server-side processing, emphasizing user privacy and data security.

Features:

• Local Execution: Runs large language models directly within the browser environment.
• Real-time Processing: Provides instant responses and interactions without delays.
• Privacy-focused: Ensures user privacy and data security by processing data locally.

Installation:

To set up the Built-in Gemini Nano in Chrome for Chrome AI, follow these steps:

1. Install Chrome Canary version 127 or higher.
2. Enable the Prompt API by visiting chrome://flags/#prompt-api-for-gemini-nano and setting it to “Enabled”.
3. Enable the Optimization Guide by visiting chrome://flags/#optimization-guide-on-device-model and setting it to “Enabled BypassPerfRequirement”. Restart the browser.
4. Download the model by navigating to chrome://components/, locating “Optimization Guide On Device Model”, and ensuring it is fully downloaded. If the version is “0.0.0.0”, click “Check for update”.
5. If the “Optimization Guide On Device Model” is not displayed, disable the settings from steps 2 and 3, restart the browser, and re-enable them.
6. Verify the setup by opening a webpage, pressing F12, and checking window.ai in the console.

Summary:

Chrome AI is a unique project that showcases the power of local execution of large language models in the browser, offering real-time processing capabilities while focusing on user privacy. By following the installation guide, users can experience the features of Chrome AI and contribute to its development for future advancements in AI technologies.