Hugging Face's logo

Datasets:
SlayerLab
/
slayer139-pretrain-mix-24k

Tasks:
Text Generation
Languages:
English
Size:
10B<n<100B
Tags:
pretraining
tokenized
License:
Dataset card Files
xet
 Community
Dataset Viewer
The dataset viewer is not available for this subset.
Cannot get the split names for the config 'default' of the dataset.
Exception:    SplitsNotFoundError
Message:      The split names could not be parsed from the dataset config.
Traceback:    Traceback (most recent call last):
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 290, in _generate_tables
                  pa_table = paj.read_json(
                      io.BytesIO(batch), read_options=paj.ReadOptions(block_size=block_size)
                  )
                File "pyarrow/_json.pyx", line 364, in pyarrow._json.read_json
                File "pyarrow/error.pxi", line 155, in pyarrow.lib.pyarrow_internal_check_status
                File "pyarrow/error.pxi", line 92, in pyarrow.lib.check_status
                  raise convert_status(status)
              pyarrow.lib.ArrowInvalid: JSON parse error: Column() changed from object to string in row 0
              
              During handling of the above exception, another exception occurred:
              
              Traceback (most recent call last):
                File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 286, in get_dataset_config_info
                  for split_generator in builder._split_generators(
                                         ~~~~~~~~~~~~~~~~~~~~~~~~~^
                      StreamingDownloadManager(base_path=builder.base_path, download_config=download_config)
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                  )
                  ^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 101, in _split_generators
                  pa_table = next(iter(self._generate_tables(**splits[0].gen_kwargs, allow_full_read=False)))[1]
                             ~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 304, in _generate_tables
                  batch = json_encode_fields_in_json_lines(original_batch, json_field_paths)
                File "/usr/local/lib/python3.14/site-packages/datasets/utils/json.py", line 111, in json_encode_fields_in_json_lines
                  examples = [ujson_loads(line) for line in original_batch.splitlines()]
                              ~~~~~~~~~~~^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/utils/json.py", line 20, in ujson_loads
                  return pd.io.json.ujson_loads(*args, **kwargs)
                         ~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^
              ValueError: Expected object or value
              
              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/split_names.py", line 68, in compute_split_names_from_streaming_response
                  for split in get_dataset_split_names(
                               ~~~~~~~~~~~~~~~~~~~~~~~^
                      path=dataset,
                      ^^^^^^^^^^^^^
                      config_name=config,
                      ^^^^^^^^^^^^^^^^^^^
                      token=hf_token,
                      ^^^^^^^^^^^^^^^
                  )
                  ^
                File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 340, in get_dataset_split_names
                  info = get_dataset_config_info(
                      path,
                  ...<6 lines>...
                      **config_kwargs,
                  )
                File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 291, in get_dataset_config_info
                  raise SplitsNotFoundError("The split names could not be parsed from the dataset config.") from err
              datasets.inspect.SplitsNotFoundError: The split names could not be parsed from the dataset config.

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.

Slayer139 pretraining mix (tok24k)

Summary

A pre-tokenized English pretraining mix: 26 882 183 219 tokens in 24 binary shards of uint16 token ids, produced with the tok24k tokenizer (vocab 24 576, included as tokenizer.json, sha256 prefix afeb22b4). It was used to train the Slayer139 model (139 279 760 parameters). No quality or benchmark claims are made here.

The release contains only token ids, not raw text. Token ids can be decoded back to text, so the licences of the upstream sources still apply (see Licensing).

Files: shards/s00.bin–shards/s07.bin, shards/t00.bin–shards/t15.bin, tokenizer.json, README.md, SHA256SUMS. SHA-256 of the concatenation of all shards in order s00…s07, t00…t15: prefix 246e7b83 (full hash in SHA256SUMS).

Format

  • Each shard is a flat little-endian uint16 array of token ids (no header).
  • Documents are concatenated and separated by the EOS token <|endoftext|>, id 0.
  • Global token offsets (lo–hi, half-open) are given in the shard table below.
import numpy as np
from tokenizers import Tokenizer

tok = Tokenizer.from_file("tokenizer.json")
ids = np.memmap("shards/t00.bin", dtype=np.uint16, mode="r")
eos = np.flatnonzero(ids[:1_000_000] == 0)
doc = ids[eos[0] + 1 : eos[1]].tolist()  # first complete document in the shard
print(tok.decode(doc))

Composition

Two parts:

  1. arcmix24k (shards s00–s07, 8 615 887 241 tokens): a base pool re-tokenized with tok24k. The pool is SlayerLab/minimal-en-corpus-5b@bb3aec23 plus new FineWeb-Edu (100BT sample) documents, OpenStax textbooks (×4 copies), and ARC-related upweight copies, minus 5 537 documents (4 362 removed by the decontamination scan, 1 175 removed because they were CC BY-NC-SA).
  2. mix24k (shards t00–t15, 18 266 295 978 tokens): a top-up from 14 Hugging Face / open-data sources, exact-deduplicated against the base pool and against FineWeb-Edu files already used, interleaved with arcmix24k shards s06–s07 (2 153 971 445 tokens) — i.e. those two shards' documents occur in both parts.

Per-source tokens in the top-up part (tok24k, billions). "Declared licence" is the license field of the source's Hugging Face metadata at the pinned revision (as read on 2026-10-12); "—" means no licence is declared there.

Source Repo @ revision Tokens (B) Declared licence Gated
cc-news vblagoje/cc_news@81eb2ce0 0.196 unknown no
DCLM-baseline mlfoundations/dclm-baseline-1.0-parquet@817d6752 1.499 cc-by-4.0 no
FineMath HuggingFaceTB/finemath@e92b25a6 0.456 odc-by no
FineWeb-Edu HuggingFaceFW/fineweb-edu@87f09149 1.917 odc-by no
LoC PD Books storytracer/LoC-PD-Books@5f1f4ee0 0.933 cc0-1.0 no
OpenWebMath open-web-math/open-web-math@fde8ef8d 0.430 — no
Project Gutenberg manu/project_gutenberg@164853d2 0.434 — no
PMC Open Access s3://pmc-oa-opendata (NCBI PMC Article Datasets, AWS Open Data; inventory snapshot 2026-10-05T01-00Z) 0.449 per article (NCBI) —
StackExchange (RedPajama) togethercomputer/RedPajama-Data-1T@398f9257 1.036 — no
StarCoderData bigcode/starcoderdata@9fc30b57 0.974 other yes
tiny-textbooks nampdn-ai/tiny-textbooks@1f88d0fe 0.086 apache-2.0 yes
UltraChat 200k HuggingFaceH4/ultrachat_200k@8049631c 0.092 mit no
Wikipedia (en) wikimedia/wikipedia@b04c8d1c 0.665 cc-by-sa-3.0, gfdl no
WildChat-1M allenai/WildChat-1M@7d6490e4 0.297 odc-by no
FineWeb-Edu expansion (incl. ARC-related copies) HuggingFaceFW/fineweb-edu@87f09149 6.648 odc-by no
Base pool (s06–s07 copy) SlayerLab/minimal-en-corpus-5b@bb3aec23 + additions 2.154 — (see its card) no

The OpenSubtitles source was planned but never fetched; it is not included.

Shards

Shard Part Token range [lo, hi) Tokens sha256 (12 hex)
s00 arcmix24k 0 – 1,076,986,152 1,076,986,152 7d4a8c7f88d3
s01 arcmix24k 1,076,986,152 – 2,153,972,880 1,076,986,728 c4c61d19d28e
s02 arcmix24k 2,153,972,880 – 3,230,957,856 1,076,984,976 f68dda27f157
s03 arcmix24k 3,230,957,856 – 4,307,954,215 1,076,996,359 8800cf707085
s04 arcmix24k 4,307,954,215 – 5,384,940,144 1,076,985,929 06930601eb5f
s05 arcmix24k 5,384,940,144 – 6,461,915,796 1,076,975,652 f7f745f287af
s06 arcmix24k 6,461,915,796 – 7,538,904,772 1,076,988,976 b9852eac6108
s07 arcmix24k 7,538,904,772 – 8,615,887,241 1,076,982,469 3f59f301da90
t00 mix24k 8,615,887,241 – 9,757,595,377 1,141,708,136 14104b2ccf18
t01 mix24k 9,757,595,377 – 10,899,174,951 1,141,579,574 26258dd813ec
t02 mix24k 10,899,174,951 – 12,040,820,507 1,141,645,556 b592697af45b
t03 mix24k 12,040,820,507 – 13,182,468,447 1,141,647,940 02a67e77a05d
t04 mix24k 13,182,468,447 – 14,324,105,205 1,141,636,758 ad1e420681d4
t05 mix24k 14,324,105,205 – 15,465,748,748 1,141,643,543 086af3e1a0b7
t06 mix24k 15,465,748,748 – 16,607,392,260 1,141,643,512 1fc0aafedb97
t07 mix24k 16,607,392,260 – 17,749,035,556 1,141,643,296 558d264e7403
t08 mix24k 17,749,035,556 – 18,890,683,183 1,141,647,627 e2408482098b
t09 mix24k 18,890,683,183 – 20,032,323,075 1,141,639,892 b4f0c9e7f783
t10 mix24k 20,032,323,075 – 21,173,967,608 1,141,644,533 4fc43794aaa3
t11 mix24k 21,173,967,608 – 22,315,609,350 1,141,641,742 60dc948b00f8
t12 mix24k 22,315,609,350 – 23,457,253,667 1,141,644,317 27adb1c43730
t13 mix24k 23,457,253,667 – 24,598,896,900 1,141,643,233 586a079c7c93
t14 mix24k 24,598,896,900 – 25,740,540,806 1,141,643,906 33aa7712a007
t15 mix24k 25,740,540,806 – 26,882,183,219 1,141,642,413 46314cd59cdd

The Slayer139 training rotation used 22 shards (s00–s05 + t00–t15 = 23 670 620 160 tokens) at 32 835 steps per shard; s06–s07 are published for completeness.

Intentional duplication

This mix is not globally deduplicated. Known intentional duplicates:

  • ARC-related documents (web documents related to ARC-style science questions) are duplicated 2–3×. Within arcmix24k about 1.13 B tokens are intentional duplicate copies; the FineWeb-Edu expansion stream contains 1 480 815 intentional ARC-related copies (≈1.89 B tokens).
  • OpenStax textbooks appear 4 times.
  • arcmix24k shards s06–s07 are also mixed into the top-up part.

Decontamination

Scope is narrow; please read precisely.

  • Checked against WikiText-2 and ARC-Easy / ARC-Challenge only, using 13-gram overlap plus exact matching of short questions.
  • After cleaning: 0 hits against the WikiText-2 test and validation splits and the ARC-Easy/ARC-Challenge test and validation splits; positive controls confirmed the scanner detects planted overlap.
  • The top-up stage additionally removed 25 808 documents (union of: 13-gram hits against WikiText-2 and ARC-Easy/Challenge test+validation; documents sharing ≥ 10 distinct 13-grams with a control document; documents containing a BLiMP sentence of ≥ 6 words).
  • The decontamination statement in the base corpus card (SlayerLab/minimal-en-corpus-5b) does not apply to this release; the scans above supersede it.
  • 467 documents overlapping ARC train questions were kept on purpose.
  • Not checked: HellaSwag, PIQA, MMLU, LAMBADA, GSM8K, HumanEval and any other benchmark. BLiMP was scanned only in the top-up part (shards t00–t15), not in arcmix24k. Do not assume this data is clean for those.

Licensing

The dataset is released under license: other because it aggregates sources with different terms. The licence of each upstream source (table above) continues to apply to its part of the data. Users are responsible for complying with the terms of each source, including gated sources (StarCoderData, tiny-textbooks) whose access conditions are set by their maintainers. OpenStax material is CC BY 4.0; ARC is CC BY-SA 4.0.

Dataset curators

Arkadiusz SÅ‚ota (SlayerLab)

Personal information

No PII scrubbing beyond upstream (e.g. FineWeb e-mail/IP anonymisation). A length-weighted random sample of 100,000 documents (575M tokens) contained e-mail-like strings in 2.6% of documents (mostly corresponding-author and organisational addresses in scientific and news text), phone-like strings in 2.7% (many false positives; real ones mostly institutional), IPv4-like strings in 1.1% (mostly non-personal). Do not use this data to identify or contact individuals.

Reproducibility

  • All Hugging Face sources are pinned to the commit revisions in the Composition table.
  • PMC was read from the AWS Open Data bucket s3://pmc-oa-opendata (inventory snapshot 2026-10-05T01-00Z); this part is not bit-reproducible because the bucket changes over time.
  • Shard integrity: verify with sha256sum -c SHA256SUMS.

Citation 

@misc{slayerlab_pretrain_mix_24k_2026,
  title  = {Slayer139 pretraining mix (tok24k)},
  author = {Arkadiusz SÅ‚ota (SlayerLab)},
  year   = {2026},
  howpublished = {\url{https://huggingface.co/datasets/SlayerLab/slayer139-pretrain-mix-24k}}
}

Please also cite the upstream datasets.
Downloads last month
194
Total file size:
53.8 GB