Hugging Face's logo

Datasets:
greg1506
/
coffee-corpus-v2

Tasks:
Text Generation
Languages:
English
Size:
100M<n<1B
Tags:
coffee
continued-pretraining
corpus
instruction-tuning
domain-adaptation
License:
Dataset card Files
xet
 Community
Dataset Viewer
Duplicate
The dataset viewer is not available for this split.
Cannot load the dataset split (in streaming mode) to extract the first rows.
Error code:   StreamingRowsError
Exception:    CastError
Message:      Couldn't cast
id: string
source: string
domain: string
title: string
text: string
url: string
license: string
pub_year: string
journal: string
corpus: string
split: string
license_tier: string
to
{'id': Value('string'), 'source': Value('string'), 'domain': Value('string'), 'title': Value('string'), 'text': Value('string'), 'url': Value('string'), 'license': Value('string'), 'corpus': Value('string'), 'split': Value('string'), 'license_tier': Value('string')}
because column names don't match
Traceback:    Traceback (most recent call last):
                File "/src/services/worker/src/worker/utils.py", line 149, in get_rows_or_raise
                  return get_rows(
                      dataset=dataset,
                  ...<4 lines>...
                      column_names=column_names,
                  )
                File "/src/libs/libcommon/src/libcommon/utils.py", line 272, in decorator
                  return func(*args, **kwargs)
                File "/src/services/worker/src/worker/utils.py", line 129, in get_rows
                  rows_plus_one = list(itertools.islice(safe_iter(ds, dataset=dataset), rows_max_number + 1))
                File "/src/services/worker/src/worker/utils.py", line 544, in safe_iter
                  yield from ds.decode(False) if ds.features else ds
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 3052, in __iter__
                  for key, example in ex_iterable:
                                      ^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2559, in __iter__
                  for key, pa_table in self._iter_arrow():
                                       ~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2584, in _iter_arrow
                  for key, pa_table in self.ex_iterable._iter_arrow():
                                       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 558, in _iter_arrow
                  for key, pa_table in iterator:
                                       ^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 440, in _iter_arrow
                  for key, pa_table in self.generate_tables_fn(**gen_kwags):
                                       ~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^
                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
              id: string
              source: string
              domain: string
              title: string
              text: string
              url: string
              license: string
              pub_year: string
              journal: string
              corpus: string
              split: string
              license_tier: string
              to
              {'id': Value('string'), 'source': Value('string'), 'domain': Value('string'), 'title': Value('string'), 'text': Value('string'), 'url': Value('string'), 'license': Value('string'), 'corpus': Value('string'), 'split': Value('string'), 'license_tier': Value('string')}
              because column names don't match

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.

coffee-corpus-v2 (open tier)

A cleaned, provenance-tracked coffee + general corpus for continued-pretraining and instruction-tuning experiments. Built from the Phase 1 harvest and the G1 quality gate (13-gram dedup, language/toxicity filters, eval decontamination).

Licence tiers (important for sharing)

LICENSES.csv lists every licence string. license_tier groups them:

tier docs meaning
open 43,421 public domain, CC0, CC BY, ODC-BY — shareable with attribution
share_alike 0 CC BY-SA — derivative redistribution carries share-alike + attribution
restricted 0 CC BY-NC / ND — non-commercial and/or no-derivatives

This bundle contains only open-tier records. Preserve the per-record source, url and license for attribution. The per-record fields remain so further subsets can be traced. A companion full bundle (which additionally contains CC BY-NC/ND and CC BY-SA records and is not cleanly redistributable) is kept outside the Hub.

Files 

corpora/                 one JSONL per bucket (all records)
MANIFEST.csv             per-bucket docs/tokens/hashes
LICENSES.csv             per-bucket licence breakdown
SHA256SUMS               integrity of every file
schema.md                record schema
dataset.json             machine-readable metadata

build_token_stream.sh    rebuild the token stream from corpora/
validate.py              integrity + schema check

Intended use

Continued-pretraining (mid-train) or instruction-tuning experiments for a coffee-focused language model; ablations over coffee domains; eval development.

Not intended as a factual authority.

Known limitations

  • Science-heavy: Europe PMC is ~94% of coffee tokens; modern brewing/practice is only 1,676,377 tokens. Add licensed brewing sources before claiming barista breadth.
  • Coffee total 49,523,453 is just below the project 50M gate; the full corpus reaches 67.7M only by including share-alike/restricted records.
  • General side is 43,834,259 tokens, below the project's 50M aspiration.
  • Language ID and toxicity filtering are keyword heuristics, not classifiers.
  • General benchmark suites were not decontaminated; the frozen coffee battery was (0 matches).
  • Token counts use the Qwen3-4B tokenizer; other tokenizers will differ.

Attribution

Preserve source, url, and license per record. CC BY / ODC-BY require attribution; Project Gutenberg texts are public domain in the US. Europe PMC articles retain their individual licence.

Reproduce / verify 

python3 validate.py
sha256sum -c SHA256SUMS
sh build_token_stream.sh    # from project root
Downloads last month
53
Total file size:
408 MB

Models trained or fine-tuned on greg1506/coffee-corpus-v2