summaryrefslogtreecommitdiff
path: root/tokenizer/byte/tokenizer_test.go
diff options
context:
space:
mode:
Diffstat (limited to 'tokenizer/byte/tokenizer_test.go')
-rw-r--r--tokenizer/byte/tokenizer_test.go49
1 files changed, 49 insertions, 0 deletions
diff --git a/tokenizer/byte/tokenizer_test.go b/tokenizer/byte/tokenizer_test.go
new file mode 100644
index 0000000..d2c8c6c
--- /dev/null
+++ b/tokenizer/byte/tokenizer_test.go
@@ -0,0 +1,49 @@
+package byte
+
+import (
+ "fmt"
+ "os"
+ "testing"
+
+ "go.jknobloc.com/x/shelf"
+)
+
+func TestNewTokenizer_Encode(t *testing.T) {
+ var file *os.File
+
+ if f, err := os.Open(shelf.Abs("tokenizers/minipile/tokenizer_gpt2_256_m000_minipile/vocab.json")); err != nil {
+ t.Fatal(err)
+ } else {
+ file = f
+ }
+
+ defer file.Close()
+
+ tok, err := NewTokenizer(file)
+
+ if err != nil {
+ t.Fatal(err)
+ }
+
+ fmt.Println(tok.Encode(" \nabc"))
+}
+
+func TestNewTokenizer_Decode(t *testing.T) {
+ var file *os.File
+
+ if f, err := os.Open(shelf.Abs("tokenizers/minipile/tokenizer_gpt2_256_m000_minipile/vocab.json")); err != nil {
+ t.Fatal(err)
+ } else {
+ file = f
+ }
+
+ defer file.Close()
+
+ tok, err := NewTokenizer(file)
+
+ if err != nil {
+ t.Fatal(err)
+ }
+
+ fmt.Println(tok.Decode([]uint8{220, 198, 64, 65, 66}))
+}