summaryrefslogtreecommitdiff
path: root/tokenizer/bpe/tokenizer.go
diff options
context:
space:
mode:
authorJonas Knobloch <jonas.knobloch@t-online.de>2026-03-31 16:21:28 +0200
committerJonas Knobloch <jonas.knobloch@t-online.de>2026-03-31 17:29:27 +0200
commitce9c65cb1e13f7ae6eed4b419232a4da9dd63fdb (patch)
tree15aad75e696c50e72d0a0bb6fe9daaf6f87ee1ab /tokenizer/bpe/tokenizer.go
parenta52a3cd0e6d6d999e93877150adf4c4ec19deff0 (diff)
Handle unknown runes
Diffstat (limited to 'tokenizer/bpe/tokenizer.go')
-rw-r--r--tokenizer/bpe/tokenizer.go10
1 files changed, 9 insertions, 1 deletions
diff --git a/tokenizer/bpe/tokenizer.go b/tokenizer/bpe/tokenizer.go
index 1b65de2..39dd7f4 100644
--- a/tokenizer/bpe/tokenizer.go
+++ b/tokenizer/bpe/tokenizer.go
@@ -20,6 +20,14 @@ func (t *Tokenizer) Decode(ids []int) string {
panic("unimplemented") // TODO implement
}
-func (t *Tokenizer) Tokenize(s string) []int {
+func (t *Tokenizer) Tokenize(s string) (ids []int) {
+ defer func() {
+ if r := recover(); r != nil {
+ // defer UnknownRunes(t, s)
+
+ ids = []int{}
+ }
+ }()
+
return t.Encode(s)
}