Together AI Blog company blog
Based: Simple linear attention language models
BASED is a linear attention architecture that decodes without a KV-cache, delivering a 24x throughput improvement over Transformers with FlashAttention-2.
- Points
- 165
- HN comments
- 14
- Submissions
- 1
- Cited by
- 0 commenters
AI & Machine LearningProgramming Languages2024
On Hacker News
| Submitted | Title on HN | Points | Comments |
|---|---|---|---|
| Mar 5, 2024 | Based: Simple linear attention language models by swyx | 165 | 14 |
