Technologies
SR-SWA
Sparse Reference Sliding Window Attention, a technique that caps attention cost so it stays flat as document length grows.
0 stories.
Every story about SR-SWA (0)
No story is about SR-SWA yet: the ones below only mention it.
Also mentioned in (1)
Stories about something else that name SR-SWA in passing, with the sentence that does.
- Baidu releases Unlimited OCR 2, an open model that reads hundreds of pages in one go“The trick that makes it work is a technique called SR-SWA, short for Sparse Reference Sliding Window Attention.”