Files
ik_llama.cpp/github-data/pull_requests/50-AVX2 Flash Attention 2.md
2025-07-22 18:18:40 +02:00

13 lines
319 B
Markdown

### 🔀 [#50](https://github.com/ikawrakow/ik_llama.cpp/pull/50) - AVX2 Flash Attention 2
| **Author** | `ikawrakow` |
| :--- | :--- |
| **State** | ❌ **Closed** |
| **Created** | 2024-09-11 |
| **Updated** | 2024-09-11 |
---
#### Description
This PR adds the ability to use Q4_0, Q4_1 and Q8_0 for the kv-cache.