highperfocused/ollama
1
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2025-08-27 02:41:21 +02:00
Code Issues Packages Projects Releases Wiki Activity

feat: qwen3 dense and sparse models (#10708)

Browse Source 
* feat: qwen3 dense
* feat: qwen3moe
* fix llama4 moe
This commit is contained in:
Michael Yang
2025-05-21 10:21:07 -07:00
committed by GitHub
parent 139f84cf21
commit e0ed984cde
5 changed files with 258 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

14
ml/backend/ggml/ggml.go
View File

@@ -887,6 +887,13 @@ func (t *Tensor) Mul(ctx ml.Context, t2 ml.Tensor) ml.Tensor {
}
}
func (t *Tensor) Div(ctx ml.Context, t2 ml.Tensor) ml.Tensor {
return &Tensor{
b: t.b,
t: C.ggml_div(ctx.(*Context).ctx, t.t, t2.(*Tensor).t),
}
}
func (t *Tensor) Mulmat(ctx ml.Context, t2 ml.Tensor) ml.Tensor {
return &Tensor{
b: t.b,
@@ -1004,6 +1011,13 @@ func (t *Tensor) Scale(ctx ml.Context, s float64) ml.Tensor {
}
}
func (t *Tensor) SumRows(ctx ml.Context) ml.Tensor {
return &Tensor{
b: t.b,
t: C.ggml_sum_rows(ctx.(*Context).ctx, t.t),
}
}
func (t *Tensor) Softmax(ctx ml.Context) ml.Tensor {
return &Tensor{
b: t.b,