Fergus Finn personal blog
Speculative KV coding: losslessly compressing KV cache by up to ~4×
Lossless compression of a target model's KV cache by up to 4×, using a cheaper predictor model to drive an arithmetic coder.
- Points
- 155
- HN comments
- 48
- Submissions
- 1
- Cited by
- 0 commenters
AI & Machine LearningDatabases & Data2026
On Hacker News
| Submitted | Title on HN | Points | Comments |
|---|---|---|---|
| Jun 4, 2026 | Speculative KV coding: losslessly compressing KV cache by up to ~4× by kkm | 155 | 48 |
