Skip to content

Commit 0111a34

Browse files
committed
vulkan: Update topk_moe fusion to handle gpt's late softmax
Based on #16649.
1 parent ee09828 commit 0111a34

File tree

2 files changed

+251
-113
lines changed

2 files changed

+251
-113
lines changed

0 commit comments

Comments
 (0)