tests/models/test_llama.py · e0fbaa7016e30dff62992706f39cab4a3dade7c4 · gaoqiong / flash-attention · GitLab

Find file Blame History Permalink

Fix Llama GQA/MQA (#546) · 42832575
Kevin Hu authored Sep 19, 2023
```
* Fix llama MQA

* Fix permute shape

* Update llama.py
```
42832575

test_llama.py 27.3 KB