mirrors/ollama-for-amd
1
0
Fork 0
mirror of https://github.com/likelovewant/ollama-for-amd.git synced 2025-12-21 14:26:30 +00:00
Code Issues Packages Projects Releases Wiki Activity

llm: Enable flash attention by default for qwen3 and qwen3moe

Browse Source
This commit is contained in:
Jesse Gross
2025-10-02 16:51:51 -07:00
committed by Jesse Gross
parent 55ca827267
commit 0bda72892c
1 changed files with 2 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
fs/ggml/ggml.go
View File

@@ -899,6 +899,8 @@ func (f GGML) SupportsFlashAttention() bool {
func (f GGML) FlashAttention() bool {
return slices.Contains([]string{
"gptoss", "gpt-oss",
"qwen3",
"qwen3moe",
}, f.KV().String("general.architecture"))
}