The dataset viewer is not available for this subset.
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/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/webdataset/webdataset.py", line 80, in _split_generators
raise ValueError(
...<2 lines>...
)
ValueError: The TAR archives of the dataset should be in WebDataset format, but the files in the archive don't share the same prefix or the same types.
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.
DiligenceBench data packet
Study material for the diligencebench benchmark of Applied RSI Bench (ARB): the documents an analyst would read to write the 150 DiligenceBench due-diligence memos (SEC filings, bank regulators' guidance and data, federal agency pages and datasets), as
plain text where possible. An ARB post-training agent gets the packet read-only at
/home/agent/workspace/data_packet/ before its model is evaluated.
This is a redistributable build. It keeps only documents from sources whose terms allow redistribution: SEC filings, works of the U.S. federal government, statutes and court opinions, and EU and Wikipedia texts under their licenses. The builder fetched 32,952 documents and left out the 16,610 from other sources (company websites, news, publishers, state and foreign governments, data vendors). Share-price histories are among them, so questions that need market prices cannot be answered from the packet alone.
The packet holds no DiligenceBench questions, answers or rubrics: the builder refuses every
URL in the benchmark's own repository and dataset. A 16-gram check (ARB's
python -m contamination overlap) agrees. No question has 25% or more of its 16-word spans in the packet. 2 of the 150 questions share one span with a filing: one quotes a sentence of the company's own 10-K (5.1% of the question), one shares a stock date phrase (1.1%). None of the rubrics shares a span.
| file | docs | unpacked → archive | tree_sha256 |
|---|---|---|---|
L.tar.gz |
16,342 | 30.19 GB → 5.10 GB | ceec962b3bc682abff897a93c6994a83c242206dcf37b35dd7fcb19c4cc46bdf |
How it was built
ARB's data_packet builder (map → scout → script → list → build), 2026-10-04:
- map: one agent (Claude Fable 5.1) read the benchmark and listed what its work needs: 16 domains, 134 entries; an entry is one source, such as "SEC EDGAR Form 10-K annual reports of the companies the questions name".
- scout: one agent (Claude Sonnet 5.5) per entry found where the entry's documents live online.
- script: one agent (Claude Sonnet 5.5) per website wrote a script that lists the site's documents for an entry, most important first (107 sites).
- list: code kept up to 1,000 documents per entry: 33,087 URLs.
- build: code fetched each URL once (robots.txt honored, per-site rate limits), turned HTML and PDF into text, and hashed every file.
The 33,087 URLs and 611 documents from 1 approved bulk archive gave 32,952 documents; 395 URLs were refused by their source (robots.txt or a 4xx answer) and 348 failed. After the redistribution rule, the packet stores 16,342 files from 46 websites, and 95 of the 134 entries have at least one document.
Most of the bytes are federal datasets the memos draw on: seven Medicare dialysis-facility report files from data.cms.gov hold 17.4 GB, and the SBA's Paycheck Protection Program loan files another 6.2 GB.
| kind | docs | GB |
|---|---|---|
| HTML pages turned into text | 11,321 | 2.17 |
| PDFs turned into text | 2,028 | 1.49 |
| text, CSV, JSON and XML as published | 1,946 | 24.52 |
other formats as fetched (.zip, .xlsx, .pptx, .xls, .bin, …) |
1,047 | 2.00 |
Layout
PACKET.json docs, bytes, tree_sha256, and per file: path, sha256, bytes, url,
entries (the map entries it serves)
README.md the map's 16 domains and 134 entries, each with its document count
<host>/<slug>.txt one normalized document per file (other formats keep their extension)
tree_sha256 is the sha256 of the sorted lines "{path} {sha256}\n", one per
document (PACKET.json and README.md excluded).
Use
hf download junlinw/arb-diligencebench-data-packet L.tar.gz --repo-type dataset --local-dir .
mkdir -p data_packet && tar -xzf L.tar.gz -C data_packet
ARB pins the exact commit it mounts in the benchmark's benchmark.toml; pass
--revision <commit> to get the same bytes.
Sources and licenses
| license (from the builder's table) | docs | GB |
|---|---|---|
| public-record (sec.gov) | 8,618 | 2.26 |
| us-government-work | 7,523 | 27.91 |
| eu-reuse-with-attribution | 162 | 0.01 |
| government-edict | 39 | < 0.01 |
SEC filings are public records. Works of the U.S. federal government are not under
copyright in the United States. Wikipedia texts are CC BY-SA 4.0, converted to plain text: credit Wikipedia's
contributors and share alike; each file's source URL is in PACKET.json. EU texts follow
the Commission's reuse policy (Decision 2011/833/EU), which asks for attribution. These are
not official versions; check the publisher before relying on a text. To have a document
removed, open a discussion on this dataset.
- Downloads last month
- 18