From 3c5f4c4279e6c9d6479e40eff4fbe1add59c1b3a Mon Sep 17 00:00:00 2001 From: Jonas Knobloch Date: Wed, 9 Sep 2026 02:41:24 +0200 Subject: WIP --- tokenizer/bpe/utility.go | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'tokenizer') diff --git a/tokenizer/bpe/utility.go b/tokenizer/bpe/utility.go index f8fcdbc..79e6faf 100644 --- a/tokenizer/bpe/utility.go +++ b/tokenizer/bpe/utility.go @@ -19,6 +19,10 @@ func NewTokenizerFromFiles(vocab, merges string, cfg Config) (*Tokenizer, error) return NewTokenizer(tokenizer, cfg), nil } +func MBPE(t *Tokenizer) *mbpe.Tokenizer { + return t.mbpe +} + func Vocab(t *Tokenizer) []string { m, ok := t.mbpe.Model().(*mbpe.MBPE) -- cgit v1.3.1