Skip to content

GitLab

Unverified Commit aea98512 authored by Qingquan Song's avatar Qingquan Song Committed by GitHub
Browse files

Fix fa3 window size setup (#5316)

parent e4155e96
Show whitespace changes
Inline Side-by-side
Showing with 2 additions and 2 deletions
python/sglang/srt/layers/attention/flashattention_backend.py
View file @ aea98512
...@@ -523,7 +523,7 @@ class FlashAttentionBackend(AttentionBackend): ...@@ -523,7 +523,7 @@ class FlashAttentionBackend(AttentionBackend):
# here is two side inclusive # here is two side inclusive
window_size = ( window_size = (
(layer.sliding_window_size, 0) (layer.sliding_window_size, 0)
if layer.sliding_window_size is not None if layer.sliding_window_size is not None and layer.sliding_window_size > -1
else (-1, -1) else (-1, -1)
) )
k_descale, v_descale = None, None k_descale, v_descale = None, None
...@@ -664,7 +664,7 @@ class FlashAttentionBackend(AttentionBackend): ...@@ -664,7 +664,7 @@ class FlashAttentionBackend(AttentionBackend):
# here is two side inclusive # here is two side inclusive
window_size = ( window_size = (
(layer.sliding_window_size, 0) (layer.sliding_window_size, 0)
if layer.sliding_window_size is not None if layer.sliding_window_size is not None and layer.sliding_window_size > -1
else (-1, -1) else (-1, -1)
) )
causal = not layer.is_cross_attention causal = not layer.is_cross_attention
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment