From 12f15fdd8c048e5958e294f12a9f163f6f950b3e Mon Sep 17 00:00:00 2001 From: Jonas Knobloch Date: Fri, 24 Apr 2026 11:49:26 +0200 Subject: Recover tokenization based on config --- tokenizer/bpe/tokenizer.go | 24 ++++++++++++++---------- 1 file changed, 14 insertions(+), 10 deletions(-) (limited to 'tokenizer/bpe/tokenizer.go') diff --git a/tokenizer/bpe/tokenizer.go b/tokenizer/bpe/tokenizer.go index 9bc351c..7e2459a 100644 --- a/tokenizer/bpe/tokenizer.go +++ b/tokenizer/bpe/tokenizer.go @@ -3,12 +3,14 @@ package bpe import "github.com/jonasknobloch/mbpe" type Tokenizer struct { - mbpe *mbpe.Tokenizer + mbpe *mbpe.Tokenizer + config Config } -func NewTokenizer(mbpe *mbpe.Tokenizer) *Tokenizer { +func NewTokenizer(mbpe *mbpe.Tokenizer, cfg Config) *Tokenizer { return &Tokenizer{ - mbpe: mbpe, + mbpe: mbpe, + config: cfg, } } @@ -29,13 +31,15 @@ func (t *Tokenizer) Decode(ids []int) string { } func (t *Tokenizer) Tokenize(s string) (ids []int) { - defer func() { - if r := recover(); r != nil { - // defer UnknownRunes(t, s) - - ids = []int{} - } - }() + if t.config.Recover { + defer func() { + if r := recover(); r != nil { + // defer UnknownRunes(t, s) + + ids = []int{} + } + }() + } return t.Encode(s) } -- cgit v1.3.1