update flash attention for gemma support: (#1368)
Browse files- requirements.txt +1 -1
- setup.py +1 -1
requirements.txt
CHANGED
|
@@ -12,7 +12,7 @@ fire
|
|
| 12 |
PyYAML>=6.0
|
| 13 |
requests
|
| 14 |
datasets>=2.15.0
|
| 15 |
-
flash-attn==2.
|
| 16 |
sentencepiece
|
| 17 |
wandb
|
| 18 |
einops
|
|
|
|
| 12 |
PyYAML>=6.0
|
| 13 |
requests
|
| 14 |
datasets>=2.15.0
|
| 15 |
+
flash-attn==2.5.5
|
| 16 |
sentencepiece
|
| 17 |
wandb
|
| 18 |
einops
|
setup.py
CHANGED
|
@@ -68,7 +68,7 @@ setup(
|
|
| 68 |
dependency_links=dependency_links,
|
| 69 |
extras_require={
|
| 70 |
"flash-attn": [
|
| 71 |
-
"flash-attn==2.5.
|
| 72 |
],
|
| 73 |
"fused-dense-lib": [
|
| 74 |
"fused-dense-lib @ git+https://github.com/Dao-AILab/flash-attention@v2.3.3#subdirectory=csrc/fused_dense_lib",
|
|
|
|
| 68 |
dependency_links=dependency_links,
|
| 69 |
extras_require={
|
| 70 |
"flash-attn": [
|
| 71 |
+
"flash-attn==2.5.5",
|
| 72 |
],
|
| 73 |
"fused-dense-lib": [
|
| 74 |
"fused-dense-lib @ git+https://github.com/Dao-AILab/flash-attention@v2.3.3#subdirectory=csrc/fused_dense_lib",
|