diff options
Diffstat (limited to 'tokenizer/bpe/tokenizer.go')
| -rw-r--r-- | tokenizer/bpe/tokenizer.go | 25 |
1 files changed, 25 insertions, 0 deletions
diff --git a/tokenizer/bpe/tokenizer.go b/tokenizer/bpe/tokenizer.go new file mode 100644 index 0000000..1b65de2 --- /dev/null +++ b/tokenizer/bpe/tokenizer.go @@ -0,0 +1,25 @@ +package bpe + +import "github.com/jonasknobloch/mbpe" + +type Tokenizer struct { + mbpe *mbpe.Tokenizer +} + +func NewTokenizer(mbpe *mbpe.Tokenizer) *Tokenizer { + return &Tokenizer{ + mbpe: mbpe, + } +} + +func (t *Tokenizer) Encode(s string) []int { + return t.mbpe.Tokenize(s) +} + +func (t *Tokenizer) Decode(ids []int) string { + panic("unimplemented") // TODO implement +} + +func (t *Tokenizer) Tokenize(s string) []int { + return t.Encode(s) +} |
