summaryrefslogtreecommitdiff
path: root/tokenizer/bpe/utility.go
diff options
context:
space:
mode:
authorJonas Knobloch <jonas.knobloch@t-online.de>2026-04-24 11:49:26 +0200
committerJonas Knobloch <jonas.knobloch@t-online.de>2026-04-24 11:50:02 +0200
commit12f15fdd8c048e5958e294f12a9f163f6f950b3e (patch)
tree16bb23c134f1fe476d977dd846e6dda6c7667c48 /tokenizer/bpe/utility.go
parent408536516bdfedc93d93226dc08280ed224361fc (diff)
Recover tokenization based on config
Diffstat (limited to 'tokenizer/bpe/utility.go')
-rw-r--r--tokenizer/bpe/utility.go4
1 files changed, 2 insertions, 2 deletions
diff --git a/tokenizer/bpe/utility.go b/tokenizer/bpe/utility.go
index fbe99fc..f8fcdbc 100644
--- a/tokenizer/bpe/utility.go
+++ b/tokenizer/bpe/utility.go
@@ -2,7 +2,7 @@ package bpe
import "github.com/jonasknobloch/mbpe"
-func NewTokenizerFromFiles(vocab, merges string) (*Tokenizer, error) {
+func NewTokenizerFromFiles(vocab, merges string, cfg Config) (*Tokenizer, error) {
model := mbpe.NewMBPE()
if err := model.Load(vocab, merges); err != nil {
@@ -16,7 +16,7 @@ func NewTokenizerFromFiles(vocab, merges string) (*Tokenizer, error) {
tokenizer.SetPreTokenizer(pre)
tokenizer.SetDecoder(pre)
- return NewTokenizer(tokenizer), nil
+ return NewTokenizer(tokenizer, cfg), nil
}
func Vocab(t *Tokenizer) []string {