Dataset Viewer
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
index_settings: struct<docstore_compression: string, docstore_blocksize: int64>
child 0, docstore_compression: string
child 1, docstore_blocksize: int64
segments: list<item: struct<segment_id: string, max_doc: int64, deletes: null>>
child 0, item: struct<segment_id: string, max_doc: int64, deletes: null>
child 0, segment_id: string
child 1, max_doc: int64
child 2, deletes: null
schema: list<item: struct<name: string, type: string, options: struct<indexed: bool, fieldnorms: bool, fast: (... 112 chars omitted)
child 0, item: struct<name: string, type: string, options: struct<indexed: bool, fieldnorms: bool, fast: bool, stor (... 100 chars omitted)
child 0, name: string
child 1, type: string
child 2, options: struct<indexed: bool, fieldnorms: bool, fast: bool, stored: bool, indexing: struct<record: string, f (... 55 chars omitted)
child 0, indexed: bool
child 1, fieldnorms: bool
child 2, fast: bool
child 3, stored: bool
child 4, indexing: struct<record: string, fieldnorms: bool, tokenizer: string>
child 0, record: string
child 1, fieldnorms: bool
child 2, tokenizer: string
child 5, precision: string
opstamp: int64
inputs: list<item: string>
child 0, item: string
seconds: int64
max_doc_chars: int64
tantivy: string
documents: int64
built_at: timestamp[s]
characters: int64
by_source: struct<packet:officeqa: int64>
child 0, packet:officeqa: int64
public: bool
chunk_chars: int64
normalizer_version: string
to
{'built_at': Value('timestamp[s]'), 'documents': Value('int64'), 'characters': Value('int64'), 'chunk_chars': Value('int64'), 'max_doc_chars': Value('int64'), 'public': Value('bool'), 'by_source': {'packet:officeqa': Value('int64')}, 'inputs': List(Value('string')), 'normalizer_version': Value('string'), 'tantivy': Value('string'), 'seconds': Value('int64')}
because column names don't match
Traceback: Traceback (most recent call last):
File "/src/services/worker/src/worker/utils.py", line 147, 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 127, 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 483, 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 2951, in __iter__
for key, example in ex_iterable:
^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2461, in __iter__
for key, pa_table in self._iter_arrow():
~~~~~~~~~~~~~~~~^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2486, 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 547, in _iter_arrow
for key, pa_table in iterator:
^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 430, 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 343, 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 2378, in table_cast
return cast_table_to_schema(table, schema)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2306, in cast_table_to_schema
raise CastError(
...<3 lines>...
)
datasets.table.CastError: Couldn't cast
index_settings: struct<docstore_compression: string, docstore_blocksize: int64>
child 0, docstore_compression: string
child 1, docstore_blocksize: int64
segments: list<item: struct<segment_id: string, max_doc: int64, deletes: null>>
child 0, item: struct<segment_id: string, max_doc: int64, deletes: null>
child 0, segment_id: string
child 1, max_doc: int64
child 2, deletes: null
schema: list<item: struct<name: string, type: string, options: struct<indexed: bool, fieldnorms: bool, fast: (... 112 chars omitted)
child 0, item: struct<name: string, type: string, options: struct<indexed: bool, fieldnorms: bool, fast: bool, stor (... 100 chars omitted)
child 0, name: string
child 1, type: string
child 2, options: struct<indexed: bool, fieldnorms: bool, fast: bool, stored: bool, indexing: struct<record: string, f (... 55 chars omitted)
child 0, indexed: bool
child 1, fieldnorms: bool
child 2, fast: bool
child 3, stored: bool
child 4, indexing: struct<record: string, fieldnorms: bool, tokenizer: string>
child 0, record: string
child 1, fieldnorms: bool
child 2, tokenizer: string
child 5, precision: string
opstamp: int64
inputs: list<item: string>
child 0, item: string
seconds: int64
max_doc_chars: int64
tantivy: string
documents: int64
built_at: timestamp[s]
characters: int64
by_source: struct<packet:officeqa: int64>
child 0, packet:officeqa: int64
public: bool
chunk_chars: int64
normalizer_version: string
to
{'built_at': Value('timestamp[s]'), 'documents': Value('int64'), 'characters': Value('int64'), 'chunk_chars': Value('int64'), 'max_doc_chars': Value('int64'), 'public': Value('bool'), 'by_source': {'packet:officeqa': Value('int64')}, 'inputs': List(Value('string')), 'normalizer_version': Value('string'), 'tantivy': Value('string'), 'seconds': Value('int64')}
because column names don't matchNeed help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
ARB OfficeQA simulated internet
A frozen, searchable copy of the web pages Applied RSI Bench's (ARB) OfficeQA benchmark draws
on, with a search index. Pages keep their real URLs. ARB's OfficeQA capsule downloads it and
gives the evaluated model web_search and fetch_url over it, which follow the request and
response formats of Tavily's /search and /extract, beside a Python session.
| pages | 13,934 from 38 websites |
| text | 3.66 billion characters |
| dated pages | 697 (the Treasury Bulletins, 1939-01-01 to 2025-09-01); the other 13,237 carry no date |
| files | docs.sqlite (3.69 GB), index/ (tantivy 0.26.2), MANIFEST.json |
| built | 2026-10-06 by ARB's simweb build from the OfficeQA data packet |
What it holds
Exactly the pages of the OfficeQA data packet (junlinw/arb-officeqa-data-packet) up to 20 MB
each (5 larger files, whole volumes of the Statutes at Large, are left out):
- every issue of the U.S. Treasury Bulletin, January 1939 to September 2025, in Databricks' parsed
text, at the address of its issue page on FRASER
(
https://fraser.stlouisfed.org/title/treasury-bulletin-407/<month>-<year>-<id>); - Bureau of the Fiscal Service, Federal Reserve Board, U.S. Treasury, BLS, IRS, Census and other U.S. federal pages and data a data-packet build found for OfficeQA's questions;
- a few Wikipedia articles.
OfficeQA's capsule hides pages dated after 2026-07-14, the date of the questions' last update. Undated pages are always visible and can hold figures revised after that date.
Licenses and attribution
- The Treasury Bulletin texts are Databricks' parsed text of the bulletins, from the Hugging
Face dataset
databricks/officeqa(revision763a8366,treasury_bulletins_parsed/transformed), licensed CC BY-SA 4.0 by Databricks ("OfficeQA: A Grounded Reasoning Benchmark", 2025). The bulletins themselves are U.S. Treasury publications, archived by FRASER (Federal Reserve Bank of St. Louis). - Wikipedia texts are CC BY-SA 4.0.
- Everything else is a work of the U.S. federal government (17 U.S.C. § 105).
No OfficeQA questions or answers
The web is built from the data packet, which holds no OfficeQA question or answer (OfficeQA's
dataset terms forbid using its answer keys to train models evaluated on OfficeQA). A 16-gram
check (ARB's python -m contamination overlap) finds no 16-word span of any of OfficeQA's 206
test questions in it.
- Downloads last month
- -