The dataset viewer is not available for this split.
Error code: StreamingRowsError
Exception: ValueError
Message: Expected object or value
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 342, 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 "/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 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 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.
MEA Benchmark Query Pools
This dataset contains the fixed extraction-query pools used by the companion anonymous benchmark artifact. It publishes five deterministic budget tiers so that attack implementations use the same query distribution and seed.
Files
| File | Records | Intended use |
|---|---|---|
query_pool_100.json |
100 | Small-budget experiment or smoke run |
query_pool_1000.json |
1,000 | Main benchmark budget |
query_pool_10000.json |
10,000 | Budget-sensitivity experiment |
query_pool_50000.json |
50,000 | Extended pool and held-out selection |
query_pool_100000.json |
100,000 | Master pool and held-out selection |
The tiers are deterministic nested subsets by record ID. The 100- and 1,000-record tiers are independently ordered stratified subsets rather than literal prefixes of the larger JSON files. Consumers should therefore download the exact file corresponding to the requested budget.
Each file is a JSON object with meta and data keys. Records under data
contain:
bank, id, prompt, rank, slice, source, split, tags, templated
The pools contain prompts only. They do not contain victim-model responses, student checkpoints, evaluation results, or API credentials.
Composition
The 100,000-record master pool uses the following fixed composition:
| Source bank | Records | Share |
|---|---|---|
| LMSYS-Chat-1M | 80,000 | 80.0% |
| NuminaMath-CoT | 4,000 | 4.0% |
| TriviaQA | 3,000 | 3.0% |
| MMLU auxiliary train | 2,500 | 2.5% |
| GSM8K | 2,000 | 2.0% |
| WinoGrande | 2,000 | 2.0% |
| HellaSwag | 1,500 | 1.5% |
| PIQA | 1,500 | 1.5% |
| SciQ | 1,000 | 1.0% |
| Social-IQA | 1,000 | 1.0% |
| ARC-Challenge | 750 | 0.75% |
| OpenBookQA | 750 | 0.75% |
The corresponding source identifiers are retained in each record's source
field. The construction seed is 20260701.
Download
from huggingface_hub import hf_hub_download
path = hf_hub_download(
repo_id="anonymous-mea-benchmark/mea-query-pools",
repo_type="dataset",
filename="query_pool_1000.json",
)
The companion code selects a tier automatically when its query-pool option is
set to auto. SHA-256 checksums are provided in SHA256SUMS.
Sources and licenses
This is a mixed-source research artifact assembled from
OpenLeecher/lmsys_chat_1m_clean, AI-MO/NuminaMath-CoT,
mandarjoshi/trivia_qa, cais/mmlu, openai/gsm8k,
allenai/winogrande, Rowan/hellaswag, ybisk/piqa, allenai/sciq,
lighteval/siqa, allenai/ai2_arc, and allenai/openbookqa.
The original datasets retain their respective licenses and terms of use. This repository does not relicense the source records under a single license. Users are responsible for reviewing and complying with the terms of each source dataset. The files are provided for research and reproducibility purposes.
- Downloads last month
- 44