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: ValueError
Message: Expected object or value
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 2840, in __iter__
for key, example in ex_iterable:
^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2373, in __iter__
for key, pa_table in self._iter_arrow():
~~~~~~~~~~~~~~~~^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2398, 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 536, in _iter_arrow
for key, pa_table in iterator:
^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 419, 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 281, in _generate_tables
examples = [ujson_loads(line) for line in 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 valueNeed help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
usta-public-record-tasks-v1-2026-09-06 — sample
20 tasks drawn from the same generator as the full pack, published with their answers so the format can be checked before anything else is exchanged.
What a task is
Each task asks for one fact that appears in a United States public record. The record is read as of a dated snapshot this project already collected, so the answer does not move when the upstream feed changes.
A task row looks like this:
{"task_id":"...","template":"...","source":{"agency":"...","host":"...","as_of":"YYYY-MM-DD"},"question":"...","answer_type":"integer","answer":"..."}
answer_type is one of integer, string, decimal1, id, date.
How to grade
Write one JSON object per line with task_id and answer, then:
python3 grade.py --tasks sample/tasks.jsonl --key sample/tasks.jsonl --answers your_answers.jsonl
Matching rules: integers must match exactly; decimal1 is accepted within
0.05; strings are compared case-insensitively with whitespace collapsed; ids
are compared exactly after trimming; dates accept YYYY-MM-DD or a full ISO
timestamp whose date part matches.
Honest notes
- Answers are dated. They describe the record as of the
as_ofdate in each task, not as of today. An agent that reads the live source may find a different value if the record changed after that date. - This is a sample. It is not a random draw from the full pack; it rotates through every template so each format appears.
- The snapshots are collected from public government feeds. Some of those feeds are rolling windows rather than complete archives. Every counting question therefore names the exact collection window it counts over, and the count is over that dated collection, not over the source's whole history.
- No personal data is used. Company names appear where the record itself names a company. Person names, street addresses, coordinates and business owner fields are excluded by the generator and checked again before release.
Sources in this pack
| database | agency | host | snapshots held | read as of |
|---|---|---|---|---|
business_formation |
municipal open-data business filing register | data.cityofchicago.org |
2026-06-26 to 2026-09-06 | 2026-09-06 |
civic_agenda |
municipal legislative record | webapi.legistar.com |
2026-06-11 to 2026-09-06 | 2026-09-06 |
clinical_trials |
U.S. National Library of Medicine (ClinicalTrials.gov) | clinicaltrials.gov |
2026-06-23 to 2026-08-11 | 2026-08-11 |
fda_enforcement |
U.S. Food and Drug Administration (openFDA) | api.fda.gov |
2026-06-23 to 2026-08-21 | 2026-08-21 |
fed_register |
Office of the Federal Register | www.federalregister.gov |
2026-06-23 to 2026-08-21 | 2026-08-21 |
sec_filings |
U.S. Securities and Exchange Commission (EDGAR) | www.sec.gov |
2026-06-23 to 2026-08-21 | 2026-08-21 |
usgs_quakes |
U.S. Geological Survey Earthquake Hazards Program | earthquake.usgs.gov |
2026-06-23 to 2026-09-06 | 2026-09-06 |
Counts by template
| template | tasks |
|---|---|
bizform_naics_month_count |
1 |
civic_matter_type_count |
1 |
ct_overall_status |
2 |
ct_phase |
2 |
fda_recall_classification |
2 |
fda_recall_month_state_count |
2 |
fedreg_agency |
2 |
fedreg_document_number |
2 |
sec_accession_lookup |
2 |
usgs_quake_daily_count |
2 |
usgs_quake_magnitude |
2 |
Files here
sample/tasks.jsonl— the 20 tasks with answers.grade.py+packlib/normalize.py— the grader, so the command above runs from this folder.seal.json— the published hashes of the full pack, its hidden answer key and this sample, fixed on release day.
- Downloads last month
- 43