Commit 716852ab authored by zhuwenwen's avatar zhuwenwen
Browse files

update value_stride_order

parent d8297312
......@@ -128,7 +128,7 @@ class FlashAttentionBackend(AttentionBackend):
cache_layout = get_kv_cache_layout()
if cache_layout == "NHD":
key_stride_order = (0, 1, 2, 3)
value_stride_order = (0, 1, 3, 2)
value_stride_order = (0, 1, 2, 3)
elif cache_layout == "HND":
key_stride_order = (0, 2, 1, 3)
value_stride_order = (0, 3, 1, 2)
......
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