Dataset Preview

Split (2)

train

The full dataset viewer is not available (click to read why). Only showing a preview of the rows.

The dataset generation failed

Error code:   DatasetGenerationError
Exception:    CastError
Message:      Couldn't cast
text: string
remove_extra_whitespaces: bool
byte_fallback: bool
character_coverage: double
seed: int64
allow_whitespace_only_pieces: bool
threshold: double
model_type: string
special_ids: struct<bos: int64, eos: int64, pad: int64, unk: int64>
  child 0, bos: int64
  child 1, eos: int64
  child 2, pad: int64
  child 3, unk: int64
languages: list<item: string>
  child 0, item: string
vocab_size: int64
corpus_sha256: string
normalization_rule_name: string
created: timestamp[s]
split_digits: bool
special_tokens: list<item: string>
  child 0, item: string
budget_mb: double
train_seconds: double
input_sentence_size: int64
language_count: int64
to
{'language_count': Value('int64'), 'languages': List(Value('string')), 'vocab_size': Value('int64'), 'model_type': Value('string'), 'byte_fallback': Value('bool'), 'split_digits': Value('bool'), 'remove_extra_whitespaces': Value('bool'), 'allow_whitespace_only_pieces': Value('bool'), 'normalization_rule_name': Value('string'), 'special_tokens': List(Value('string')), 'special_ids': {'bos': Value('int64'), 'eos': Value('int64'), 'pad': Value('int64'), 'unk': Value('int64')}, 'character_coverage': Value('float64'), 'input_sentence_size': Value('int64'), 'budget_mb': Value('float64'), 'threshold': Value('float64'), 'seed': Value('int64'), 'corpus_sha256': Value('string'), 'train_seconds': Value('float64'), 'created': Value('timestamp[s]')}
because column names don't match
Traceback:    Traceback (most recent call last):
                File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1835, in _prepare_split_single
                  for key, table in generator:
                                    ^^^^^^^^^
                File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 621, in wrapped
                  for item in generator(*args, **kwargs):
                              ~~~~~~~~~^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 361, in _generate_tables
                  self._cast_table(pa_table, json_field_paths=json_field_paths),
                  ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 132, in _cast_table
                  pa_table = table_cast(pa_table, self.info.features.arrow_schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2381, in table_cast
                  return cast_table_to_schema(table, schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2309, in cast_table_to_schema
                  raise CastError(
                  ...<3 lines>...
                  )
              datasets.table.CastError: Couldn't cast
              text: string
              remove_extra_whitespaces: bool
              byte_fallback: bool
              character_coverage: double
              seed: int64
              allow_whitespace_only_pieces: bool
              threshold: double
              model_type: string
              special_ids: struct<bos: int64, eos: int64, pad: int64, unk: int64>
                child 0, bos: int64
                child 1, eos: int64
                child 2, pad: int64
                child 3, unk: int64
              languages: list<item: string>
                child 0, item: string
              vocab_size: int64
              corpus_sha256: string
              normalization_rule_name: string
              created: timestamp[s]
              split_digits: bool
              special_tokens: list<item: string>
                child 0, item: string
              budget_mb: double
              train_seconds: double
              input_sentence_size: int64
              language_count: int64
              to
              {'language_count': Value('int64'), 'languages': List(Value('string')), 'vocab_size': Value('int64'), 'model_type': Value('string'), 'byte_fallback': Value('bool'), 'split_digits': Value('bool'), 'remove_extra_whitespaces': Value('bool'), 'allow_whitespace_only_pieces': Value('bool'), 'normalization_rule_name': Value('string'), 'special_tokens': List(Value('string')), 'special_ids': {'bos': Value('int64'), 'eos': Value('int64'), 'pad': Value('int64'), 'unk': Value('int64')}, 'character_coverage': Value('float64'), 'input_sentence_size': Value('int64'), 'budget_mb': Value('float64'), 'threshold': Value('float64'), 'seed': Value('int64'), 'corpus_sha256': Value('string'), 'train_seconds': Value('float64'), 'created': Value('timestamp[s]')}
              because column names don't match
              
              The above exception was the direct cause of the following exception:
              
              Traceback (most recent call last):
                File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 1379, in compute_config_parquet_and_info_response
                  parquet_operations, partial, estimated_dataset_info = stream_convert_to_parquet(
                                                                        ~~~~~~~~~~~~~~~~~~~~~~~~~^
                      builder, max_dataset_size_bytes=max_dataset_size_bytes
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                  )
                  ^
                File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 956, in stream_convert_to_parquet
                  builder._prepare_split(split_generator=splits_generators[split], file_format="parquet")
                  ~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1702, in _prepare_split
                  for job_id, done, content in self._prepare_split_single(
                                               ~~~~~~~~~~~~~~~~~~~~~~~~~~^
                      gen_kwargs=gen_kwargs, job_id=job_id, **_prepare_split_args
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                  ):
                  ^
                File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1887, in _prepare_split_single
                  raise DatasetGenerationError("An error occurred while generating the dataset") from e
              datasets.exceptions.DatasetGenerationError: An error occurred while generating the dataset

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

language_count int64	languages list	vocab_size int64	model_type string	byte_fallback bool	split_digits bool	remove_extra_whitespaces bool	allow_whitespace_only_pieces bool	normalization_rule_name string	special_tokens list	special_ids dict	character_coverage float64	input_sentence_size int64	budget_mb float64	threshold float64	seed int64	corpus_sha256 string	train_seconds float64	created timestamp[s]
16	[ "rus_Cyrl", "spa_Latn", "jpn_Jpan", "cmn_Hans", "ind_Latn", "kor_Hang", "tur_Latn", "pes_Arab", "deu_Latn", "arb_Arab", "hun_Latn", "vie_Latn", "fin_Latn", "ell_Grek", "tha_Thai", "hin_Deva" ]	65,536	bpe	true	true	false	true	identity	[ "<\|begin_of_text\|>", "<\|end_of_text\|>", "<\|start_think\|>", "<\|end_think\|>", "<\|finetune_right_pad_id\|>", "<\|tool_call\|>", "<\|start_header_id\|>", "<\|end_header_id\|>", "<\|eom_id\|>", "<\|eot_id\|>", "<\|python_tag\|>", "<\|tool_call_end\|>", "<\|tool_result\|>", "<\|tool_result_end\|>", "<\|reserved_s...	{ "bos": 0, "eos": 1, "pad": 4, "unk": 255 }	0.99995	0	100	0.5	42	63f7c50f198c5d5bd49f1b9f832a329442752453cb2c44caf3890008c6c62b1f	203.855526	2026-10-13T23:50:00

YAML Metadata Warning:empty or missing yaml metadata in repo card

Check out the documentation for more information.

L16_V65536_25B

Pre-training data for delphi-pretrain (train.py --data-dir <this directory> --tokenizer <tokenizer>): 16 languages x 25,000,000,000 tokens = 400,000,000,000 tokens in 368,667,368 documents, uniform random samples of the reject-filtered HPLT 3.0 documents of each language, with their English taken out (HPLT's line labels and sentence-level OpenLID-v3 / GlotLID-v3: short or mostly English lines whole, the English sentences of other lines; see preprocessing/pieces.py), all languages shuffled together (seed 42).

shards/NNNNN.bin: uint16 token ids, one stream of documents, each BOS (id 0), its tokens, EOS (id 1). The last sampled document of every language is cut (keeping its EOS) so that the language total is exact (or 1-2 tokens short when the cut would leave nothing between BOS and EOS). Every file holds whole windows of 4,096 tokens: file j is stream positions [j*S, (j+1)*S + 1) with S = 390,627,328, so the last token of a file is the first of the next one and no window is lost (97,656,249 windows).
shards/NNNNN.docs.npy: offsets of the documents that start in the file; shards/NNNNN.json: its tokens per language.
val.bin, val_langs.json, val_ids.json: 625,000 validation tokens per language from documents that are not in the training stream.
documents/group-NNNNN.parquet: id, language, stream offset and length of every document.
meta.json (written last), manifest.json, selection.json.

Tokenizer: L16_V65536 (vocabulary 65,536).

Downloads last month: 106

Total file size:

803 GB