llm: Enable flash attention for mistral3 by default
This commit is contained in:
parent
1108d8b34e
commit
9191dfaf05
|
|
@ -831,6 +831,7 @@ func (f GGML) FlashAttention() bool {
|
||||||
return slices.Contains([]string{
|
return slices.Contains([]string{
|
||||||
"gemma3",
|
"gemma3",
|
||||||
"gptoss", "gpt-oss",
|
"gptoss", "gpt-oss",
|
||||||
|
"mistral3",
|
||||||
"qwen3", "qwen3moe",
|
"qwen3", "qwen3moe",
|
||||||
"qwen3vl", "qwen3vlmoe",
|
"qwen3vl", "qwen3vlmoe",
|
||||||
}, f.KV().String("general.architecture"))
|
}, f.KV().String("general.architecture"))
|
||||||
|
|
|
||||||
Loading…
Reference in New Issue