Revision History
PagedAttention · 8 revisions
Sizes are character counts of the article source. The signed number is the change from the previous revision.
Recent edit summaries
Detailed summaries recorded by editors. Generic maintenance summaries are omitted here; every recorded revision remains below. Dates describe the edit, not necessarily the event it covers.
- Correction: Corrected the KV-cache size example (OPT-13B, 1.6 GB, not LLaMA-13B 1.7 GB) and the weights-vs-KV-cache memory split, the throughput gain over Orca (Max) (2.7x to 8x, not 2.3x to 4.3x), removed an unsupported 30% shared-prefix memory saving and an unsupported 29% SGLang-vs-vLLM figure, dated the vLLM line count to the 2023 paper, and noted that TGI is now in maintenance mode (plus 21 minor fixes)
Version 8 · Nov 2, 2026, 12:12 PM
- Update: added a video.
Version 7 · Nov 1, 2026, 08:06 AM
- Improve internal links (2026-10-13)
Version 6 · Oct 13, 2026, 10:20 AM
- Added 6 contextual internal links
Version 5 · Jul 23, 2026, 03:29 PM