summaryrefslogtreecommitdiff
path: root/llmc/cmd/peek/main.go
diff options
context:
space:
mode:
Diffstat (limited to 'llmc/cmd/peek/main.go')
-rw-r--r--llmc/cmd/peek/main.go5
1 files changed, 4 insertions, 1 deletions
diff --git a/llmc/cmd/peek/main.go b/llmc/cmd/peek/main.go
index 0a5eeab..1bfcca8 100644
--- a/llmc/cmd/peek/main.go
+++ b/llmc/cmd/peek/main.go
@@ -18,7 +18,10 @@ func main() {
_ = must(llmc.Deserialize(shelf.Abs("llmc/edu_fineweb100B/edu_fineweb_val_000000.bin"), &a))
_ = must(llmc.Deserialize(shelf.Abs("test/llmc/edu_fineweb100B/edu_fineweb_val_000000.bin"), &b))
- t := must(bpe.NewTokenizerFromFiles(shelf.Abs("models/gpt2/vocab.json"), shelf.Abs("models/gpt2/merges.txt")))
+ v := shelf.Abs("models/gpt2/vocab.json")
+ m := shelf.Abs("models/gpt2/merges.txt")
+
+ t := must(bpe.NewTokenizerFromFiles(v, m, bpe.DefaultConfig()))
s := decode(&a, 1024, t)
k := decode(&b, 1024, t)