Privacy Policy
How we handle your data and privacy at torch.js
Your privacy is a top priority at torch.js. This policy outlines how we collect, use, and protect your information as our platform evolves from a simple library into a collaborative ecosystem.
1. Local-First Computation
The core of torch.js remains a client-side library. When you run models in the Playground or in your own applications, all tensor computations happen locally on your hardware via WebGPU. We do not see or store the internal data of your local computations.
2. Accounts and Authentication
When you create an account using our login system (e.g., via GitHub or Email), we store your email address and basic profile information to manage your projects and playgrounds. This data is used solely for authentication and to provide you with access to your saved work.
3. Public Nature of Uploads
Important Notice:
Any project, playground, code, or dataset you upload to torchjs.org is currently treated as public. Do not upload sensitive information, private keys, or proprietary data that you do not wish to share with the community.
Future versions of the platform may include private project support, but currently, any data pushed to our servers is accessible to other users via their respective URLs.
4. Analytics and Logs
We use anonymized analytics to monitor website performance and identify common errors. We do not track individual users across the web or sell your data to third parties. Our hosting provider, Vercel, maintains standard server logs (IP addresses and request headers) for security and operational purposes.
5. Cookies
We use essential cookies to maintain your session when you are logged in. We do not use advertising or tracking cookies from third-party networks.
Last Updated: January 8, 2026