Commit 34d4122
committed
vulkan: Update topk_moe fusion to handle gpt's late softmax
Based on ggml-org#16649.1 parent d853036 commit 34d4122
File tree
2 files changed
+251
-113
lines changed- ggml/src/ggml-vulkan
- vulkan-shaders
2 files changed
+251
-113
lines changed
0 commit comments