summaryrefslogtreecommitdiff
path: root/tokenizer/bpe/cmd/tokenize/main.go
diff options
context:
space:
mode:
Diffstat (limited to 'tokenizer/bpe/cmd/tokenize/main.go')
-rw-r--r--tokenizer/bpe/cmd/tokenize/main.go5
1 files changed, 4 insertions, 1 deletions
diff --git a/tokenizer/bpe/cmd/tokenize/main.go b/tokenizer/bpe/cmd/tokenize/main.go
index 13926b7..e6cab4b 100644
--- a/tokenizer/bpe/cmd/tokenize/main.go
+++ b/tokenizer/bpe/cmd/tokenize/main.go
@@ -100,7 +100,10 @@ func data() *dataset.ParquetReader {
func tokenizer() llm.Tokenizer {
var tok *bpe.Tokenizer
- if t, err := bpe.NewTokenizerFromFiles(shelf.Abs("models/gpt2/vocab.json"), shelf.Abs("models/gpt2/merges.txt")); err != nil {
+ v := shelf.Abs("models/gpt2/vocab.json")
+ m := shelf.Abs("models/gpt2/merges.txt")
+
+ if t, err := bpe.NewTokenizerFromFiles(v, m, bpe.DefaultConfig()); err != nil {
log.Fatal(err)
} else {
tok = t