Home › Categories › Technology› Shavit (2023)

Shavit (2023).pdf

2303.11341.pdf

🔗 Source: arxiv.org

📊 Size: 672 KB

📄 Pages: 22 pages

⬇️ Downloads: 87

Summary

This work proposes a monitoring framework to enforce rules on large-scale neural network training using specialized ML chips, aiming to verify compliance with agreed rules and prevent misuse while maintaining privacy and confidentiality of ML models and data. The framework consists of three parts: logging chip activity via firmware, saving information about training runs, and monitoring the chip supply chain. The goal is to provide governments with high confidence that no actor is using large quantities of ML chips to execute training runs in violation of agreed rules, without curtailing the use of consumer computing devices.

Download File

Description

This work proposes a monitoring framework to enforce rules on large-scale neural network training using specialized ML chips, aiming to verify compliance with...

Technical Information

File Format: PDF
File Size: 672 KB
Pages: 22
Language: EN
Total Downloads: 87
Last Updated: 4 weeks ago

Document Overview

This PDF document about Shavit (2023) provides comprehensive information and guidance. Whether you're a beginner or advanced user, this resource offers valuable insights into Shavit (2023).