Lend your machine

What happens when you translate

1. You press Start translating. The page asks the server for a chapter that has no translation.
2. Your browser translates it. It loads the translation model (about 1 GB, downloaded once and kept in the browser) and runs it on your graphics card. The server does not run any model.
3. The result is sent back. The server checks it: complete, no Chinese left over, no repeated sentences. If it passes, it is published for all readers as a machine translation and credited to you.
4. Then it repeats. The page takes the next chapter until you press Stop or the queue is empty. A new translator's chapters wait for a moderator first, unless the chapter had no translation at all.

The pop-up window

If you press Translate in my browser on a chapter in the reader, the chapter is added to your requests and a small separate window opens. That window is this page; it holds the model and translates the chapters you asked for.

• Keep the window open while it works. The reader page can be reloaded or left; it shows the progress again when you return.
• Another chapter requested later uses the same window; no second window or second copy of the model opens.
• With nothing left to translate, the window waits for new requests and frees the graphics card after 10 minutes. It loads the model again when you ask for another chapter.
• If the browser blocks the window, allow pop-ups for this site. Until then the chapter is translated inside the reader page, which has to stay open.

You need Chrome or Edge on a desktop computer with a graphics card, and an account. A chapter takes a few minutes depending on the card, and the computer is busy meanwhile. If you close the window, the finished pieces are kept in the browser and continue next time.