matt/ollama37
1
0
Fork 0
mirror of https://github.com/dogkeeper886/ollama37.git synced 2025-12-10 15:57:04 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix vocabulary (#10679)

Browse Source
This commit is contained in:
Michael Yang
2025-05-12 17:29:46 -07:00
committed by GitHub
parent a7240c6d63
commit 526b2ed102
7 changed files with 28 additions and 26 deletions
Download Patch File Download Diff File Expand all files Collapse all files

13
model/models/mistral3/model_text.go
View File

@@ -21,7 +21,6 @@ type TextOptions struct {
type TextModel struct {
model.Base
model.BytePairEncoding
TokenEmbedding *nn.Embedding `gguf:"token_embd"`
Layers []Layer `gguf:"blk"`
@@ -148,18 +147,6 @@ func NewTextModel(c fs.Config) (*TextModel, error) {
}
textModel := &TextModel{
BytePairEncoding: model.NewBytePairEncoding(
c.String("tokenizer.ggml.pretokenizer", `[^\r\n\p{L}\p{N}]?[\p{Lu}\p{Lt}\p{Lm}\p{Lo}\p{M}]*[\p{Ll}\p{Lm}\p{Lo}\p{M}]+|[^\r\n\p{L}\p{N}]?[\p{Lu}\p{Lt}\p{Lm}\p{Lo}\p{M}]+[\p{Ll}\p{Lm}\p{Lo}\p{M}]*|\p{N}| ?[^\s\p{L}\p{N}]+[\r\n/]*|\s*[\r\n]+|\s+(?!\S)|\s+`),
&model.Vocabulary{
Values: c.Strings("tokenizer.ggml.tokens"),
Types: c.Ints("tokenizer.ggml.token_type"),
Merges: c.Strings("tokenizer.ggml.merges"),
BOS: int32(c.Uint("tokenizer.ggml.bos_token_id", 1)),
AddBOS: c.Bool("tokenizer.ggml.add_bos_token", true),
EOS: int32(c.Uint("tokenizer.ggml.eos_token_id", 2)),
AddEOS: c.Bool("tokenizer.ggml.add_eos_token", false),
},
),
Layers: make([]Layer, c.Uint("block_count")),
TextOptions: &TextOptions{
hiddenSize: int(c.Uint("embedding_length")),