diff options
Diffstat (limited to 'llmc/tokenize.go')
| -rw-r--r-- | llmc/tokenize.go | 8 |
1 files changed, 1 insertions, 7 deletions
diff --git a/llmc/tokenize.go b/llmc/tokenize.go index e90de58..1685168 100644 --- a/llmc/tokenize.go +++ b/llmc/tokenize.go @@ -44,13 +44,7 @@ func WriteShards(path, name string, size int, docs <-chan []uint32) (int, error) n := 0 flush := func() error { - split := "train" - - if n == 0 { - split = "val" - } - - shardName := filepath.Join(path, fmt.Sprintf("%s_%s_%06d.bin", name, split, n)) + shardName := filepath.Join(path, fmt.Sprintf("%s_%06d.bin", name, n)) d := DataFile[uint32]{ Model: GPT2, |