mirrors/ollama-for-amd
1
0
Fork 0
mirror of https://github.com/likelovewant/ollama-for-amd.git synced 2025-12-26 00:18:02 +00:00
Code Issues Packages Projects Releases Wiki Activity

model: implement bert in ollama engine (#9080)

Browse Source 
* fix truncate

* s/SentencePieceModel/SentencePiece/

* bert

* wordpiece

* refactor pooling

* more tokenizers

* normalize embeddings
This commit is contained in:
Michael Yang
2025-09-15 15:35:59 -07:00
committed by GitHub
parent 6f7117145f
commit 3f6642f6fc
16 changed files with 490 additions and 40 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
model/models/gemma3n/model.go
View File

@@ -10,7 +10,7 @@ import (
type Model struct {
model.Base
model.SentencePieceModel
model.SentencePiece
*TextModel
}
@@ -23,7 +23,7 @@ func (m *Model) Forward(ctx ml.Context, batch input.Batch) (ml.Tensor, error) {
func New(c fs.Config) (model.Model, error) {
m := Model{
TextModel: newTextModel(c),
SentencePieceModel: model.NewSentencePieceModel(
SentencePiece: model.NewSentencePiece(
&model.Vocabulary{
Values: c.Strings("tokenizer.ggml.tokens"),
Scores: c.Floats("tokenizer.ggml.scores"),