File tree Expand file tree Collapse file tree 3 files changed +2
-5
lines changed
nemo_curator/stages/text/io/writer
tests/stages/text/io/writer
tutorials/text/megatron-tokenizer Expand file tree Collapse file tree 3 files changed +2
-5
lines changed Original file line number Diff line number Diff line change 1313# limitations under the License.
1414
1515from nemo_curator .stages .text .io .writer .jsonl import JsonlWriter
16- from nemo_curator .stages .text .io .writer .megatron_tokenizer import MegatronTokenizerWriter
1716from nemo_curator .stages .text .io .writer .parquet import ParquetWriter
1817
1918__all__ = [
2019 "JsonlWriter" ,
21- "MegatronTokenizerWriter" ,
2220 "ParquetWriter" ,
2321]
Original file line number Diff line number Diff line change 2424import pytest
2525
2626import nemo_curator .stages .text .io .writer .utils as writer_utils
27- from nemo_curator .stages .text .io .writer import MegatronTokenizerWriter
28- from nemo_curator .stages .text .io .writer .megatron_tokenizer import _INDEX_HEADER
27+ from nemo_curator .stages .text .io .writer .megatron_tokenizer import _INDEX_HEADER , MegatronTokenizerWriter
2928from nemo_curator .tasks import DocumentBatch
3029
3130
Original file line number Diff line number Diff line change 2222from nemo_curator .core .client import RayClient
2323from nemo_curator .pipeline import Pipeline
2424from nemo_curator .stages .text .io .reader import ParquetReader
25- from nemo_curator .stages .text .io .writer import MegatronTokenizerWriter
25+ from nemo_curator .stages .text .io .writer . megatron_tokenizer import MegatronTokenizerWriter
2626
2727
2828def main (args : argparse .Namespace ) -> None :
You can’t perform that action at this time.
0 commit comments