[mlx-lm] Use sdpa in llama / mistral model (#515)

* use sdpa

* update a few more models

* version

* fix stablelm type
This commit is contained in:
Awni Hannun
2024-03-07 17:41:23 -08:00
committed by GitHub
parent 7cdd1b69ac
commit 8b05bb6d18
7 changed files with 25 additions and 59 deletions

View File

@@ -1,4 +1,4 @@
mlx>=0.4
mlx>=0.6
numpy
transformers>=4.38.0
protobuf