Skip to content

Update flash attention to version v2.3.6 #8

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 1 commit into
base: kv-cache
Choose a base branch
from
Open

Update flash attention to version v2.3.6 #8

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 2 additions & 2 deletions server/Makefile-flash-att-v2
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,4 +1,4 @@
flash_att_v2_commit := 4f285b354796fb17df8636485b9a04df3ebbb7dc
flash_att_v2_commit := 92dd5703ecdb99aa4a4aee9817f28557907403a2 # v2.3.6

flash-attention-v2:
# Clone flash attention
Expand All @@ -10,4 +10,4 @@ build-flash-attention-v2: flash-attention-v2
cd flash-attention-v2 && python setup.py build

install-flash-attention-v2: build-flash-attention-v2
cd flash-attention-v2 && python setup.py install
cd flash-attention-v2 && python setup.py install
3 changes: 3 additions & 0 deletions server/text_generation_server/utils/flash_attn.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -66,12 +66,15 @@ def attention(
out,
cu_seqlens,
cu_seqlens,
None, # seqused_k added in ce3e728
max_s,
max_s,
0.0,
softmax_scale,
False,
True,
-1, # window_size[0] added in 083e8f52. -1 means infinite window size
-1, # window_size[1] added in 083e8f52. -1 means infinite window size
False,
None,
)
Expand Down