Datasets:

leonmodal
/

arb-finance-web-ff

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
exhibit_labels: int64
page_meta: struct<dated: int64, redated: int64, withheld: int64, tickers: int64, aliases: int64>
  child 0, dated: int64
  child 1, redated: int64
  child 2, withheld: int64
  child 3, tickers: int64
  child 4, aliases: int64
seconds: int64
max_doc_chars: int64
characters: int64
public: bool
chunk_chars: int64
by_source: struct<corpus:ff4-gov-series: int64, raw_corpus:fav2: int64, raw_corpus:bigfinancebench: int64, raw_ (... 351 chars omitted)
  child 0, corpus:ff4-gov-series: int64
  child 1, raw_corpus:fav2: int64
  child 2, raw_corpus:bigfinancebench: int64
  child 3, raw_corpus:diligencebench: int64
  child 4, raw_corpus:frontierfinance: int64
  child 5, corpus:edgar: int64
  child 6, corpus:wiki: int64
  child 7, corpus:ff-sec: int64
  child 8, corpus:ff-sec-history: int64
  child 9, corpus:ff-funds: int64
  child 10, corpus:wiki-events: int64
  child 11, corpus:ff4-sec: int64
  child 12, corpus:ff4-xbrl: int64
  child 13, corpus:ff4-prices: int64
  child 14, corpus:ff4-ftd: int64
  child 15, corpus:defeatbeta: int64
  child 16, corpus:kurry: int64
tantivy: string
normalizer_version: string
built_at: timestamp[s]
inputs: list<item: string>
  child 0, item: string
documents: int64
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': {'corpus:ff4-gov-series': Value('int64'), 'raw_corpus:fav2': Value('int64'), 'raw_corpus:bigfinancebench': Value('int64'), 'raw_corpus:diligencebench': Value('int64'), 'raw_corpus:frontierfinance': Value('int64'), 'corpus:edgar': Value('int64'), 'corpus:wiki': Value('int64'), 'corpus:ff-sec': Value('int64'), 'corpus:ff-sec-history': Value('int64'), 'corpus:ff-funds': Value('int64'), 'corpus:wiki-events': Value('int64'), 'corpus:ff4-sec': Value('int64'), 'corpus:ff4-xbrl': Value('int64'), 'corpus:ff4-prices': Value('int64'), 'corpus:ff4-ftd': Value('int64'), 'corpus:defeatbeta': Value('int64'), 'corpus:kurry': Value('int64')}, 'inputs': List(Value('string')), 'exhibit_labels': Value('int64'), 'page_meta': {'dated': Value('int64'), 'redated': Value('int64'), 'withheld': Value('int64'), 'tickers': Value('int64'), 'aliases': Value('int64')}, '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 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
              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
              exhibit_labels: int64
              page_meta: struct<dated: int64, redated: int64, withheld: int64, tickers: int64, aliases: int64>
                child 0, dated: int64
                child 1, redated: int64
                child 2, withheld: int64
                child 3, tickers: int64
                child 4, aliases: int64
              seconds: int64
              max_doc_chars: int64
              characters: int64
              public: bool
              chunk_chars: int64
              by_source: struct<corpus:ff4-gov-series: int64, raw_corpus:fav2: int64, raw_corpus:bigfinancebench: int64, raw_ (... 351 chars omitted)
                child 0, corpus:ff4-gov-series: int64
                child 1, raw_corpus:fav2: int64
                child 2, raw_corpus:bigfinancebench: int64
                child 3, raw_corpus:diligencebench: int64
                child 4, raw_corpus:frontierfinance: int64
                child 5, corpus:edgar: int64
                child 6, corpus:wiki: int64
                child 7, corpus:ff-sec: int64
                child 8, corpus:ff-sec-history: int64
                child 9, corpus:ff-funds: int64
                child 10, corpus:wiki-events: int64
                child 11, corpus:ff4-sec: int64
                child 12, corpus:ff4-xbrl: int64
                child 13, corpus:ff4-prices: int64
                child 14, corpus:ff4-ftd: int64
                child 15, corpus:defeatbeta: int64
                child 16, corpus:kurry: int64
              tantivy: string
              normalizer_version: string
              built_at: timestamp[s]
              inputs: list<item: string>
                child 0, item: string
              documents: int64
              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': {'corpus:ff4-gov-series': Value('int64'), 'raw_corpus:fav2': Value('int64'), 'raw_corpus:bigfinancebench': Value('int64'), 'raw_corpus:diligencebench': Value('int64'), 'raw_corpus:frontierfinance': Value('int64'), 'corpus:edgar': Value('int64'), 'corpus:wiki': Value('int64'), 'corpus:ff-sec': Value('int64'), 'corpus:ff-sec-history': Value('int64'), 'corpus:ff-funds': Value('int64'), 'corpus:wiki-events': Value('int64'), 'corpus:ff4-sec': Value('int64'), 'corpus:ff4-xbrl': Value('int64'), 'corpus:ff4-prices': Value('int64'), 'corpus:ff4-ftd': Value('int64'), 'corpus:defeatbeta': Value('int64'), 'corpus:kurry': Value('int64')}, 'inputs': List(Value('string')), 'exhibit_labels': Value('int64'), 'page_meta': {'dated': Value('int64'), 'redated': Value('int64'), 'withheld': Value('int64'), 'tickers': Value('int64'), 'aliases': Value('int64')}, 'normalizer_version': Value('string'), 'tantivy': Value('string'), 'seconds': Value('int64')}
              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.

ARB finance simulated internet

A frozen copy of the web pages that the finance benchmarks of Applied RSI Bench (ARB) draw on (fav2, which is Vals AI Finance Agent v2, BigFinanceBench, DiligenceBench and FrontierFinance), with a search index and embedding vectors. Pages keep their real URLs. ARB's finance tasks give the evaluated model two tools over it, web_search and fetch_url, which follow the request and response formats of Tavily's /search and /extract.

This revision (sim-ff4, built 2026-10-27) is the web of ARB's finance-v2 track, which serves all four finance tasks from it. The revision before it (sim-ff3, 293,314 pages) was pinned by FrontierFinance only; ARB's finance track stays on junlinw/arb-finance-web at fd05a101.

Each benchmark sees the web as of its own date (fav2 2026-07-08, BigFinanceBench 2026-05-28, DiligenceBench 2026-05-21, FrontierFinance each query's own date, 2024-08-08 to 2026-04-30): pages dated after it are hidden from search and from reading, and on a time series page the rows dated after it are hidden. Every page carries a date except 828 reference pages whose content does not depend on a date (the SEC's investor education pages on investor.gov and the XBRL taxonomy schemas on xbrl.sec.gov).


pages	420,463 from 70 websites
text	35.0 billion characters
files	`docs.sqlite`, `index/` (tantivy 0.26.2), `vectors/` (Qwen3-Embedding-0.6B, 19,282,198 chunks), `MANIFEST.json`
built	2026-10-27 by ARB's `simweb build --public`

What it holds

source	pages	licence
SEC EDGAR filings of S&P 1500 companies since 2019	203,108	public record
SEC EDGAR filings of the companies FrontierFinance's queries name	32,769	public record
SEC EDGAR filings of the companies the other three tasks' items name, 2017 on (new)	13,371	public record
SEC XBRL financial data, one page per 10-K/10-Q/20-F/40-F of 1,630 companies (new)	77,797	public record
Daily closing prices from SEC's fails-to-deliver data, one page per security and year (new)	12,504	public record
Share prices in SEC filings: insiders' open-market trades (Forms 4) by month, issuer repurchase prices by report (new)	5,391	public record
ARB raw corpora (redistributable builds): FrontierFinance 15,956, DiligenceBench 14,628, fav2 13,828, BigFinanceBench 10,392	54,804	public records, U.S. government works, open licences
U.S. government statistics tables re-dated by row (Treasury yields, Federal Reserve H.15, EIA prices; new form of 154 raw-corpus pages)	154	U.S. government works
Earnings-call transcripts (14,008) and daily share prices (4,051 pages) from `defeatbeta/yahoo-finance-data` (new, open-licence dataset layer)	18,059	ODC-BY as stated by the dataset
Earnings-call transcripts from `kurry/sp500_earnings_transcripts` (new, open-licence dataset layer)	27	MIT as stated by the dataset
Wikipedia articles of S&P 1500 companies and daily "Current events" pages	2,450	CC BY-SA 4.0
SEC N-PORT holdings reports of two ETFs (XBI, XOP)	29	public record

The open-licence dataset layer

Pages taken from public Hugging Face datasets that state an open licence, each pinned to one revision, with a first line naming the dataset, its revision, its licence as stated by the dataset and its original source as stated by its card. Their URLs are the dataset's file URL with the row's key (https://huggingface.co/datasets/<id>/blob/<revision>/<file>?...), so the layer can be told apart and dropped.

dataset	revision	licence as stated	original source as stated by its card	used
`defeatbeta/yahoo-finance-data`	`86776ce50af7`	odc-by	Yahoo! Finance, Nasdaq and the U.S. Department of the Treasury via publicly available APIs; the card adds that the data "is intended for research and educational purposes"	earnings-call transcripts (one page per call, dated by the call), daily prices (one page per ticker and year, prices as traded that day, the dataset's later split adjustment undone)
`kurry/sp500_earnings_transcripts`	`f3ded372da8d`	mit	not stated; the card says "For research and educational use only"	transcripts of 27 calls the first dataset lacks

News items of defeatbeta/yahoo-finance-data and AlphaDojo/dojo_stock_news (Apache-2.0) were prepared too but are not in this revision: in ARB's fact-level coverage check, their 220,000 short, tickered pages pushed SEC filings out of the top search results for the older tasks' questions.

Dates and tickers

SEC pages are dated by their filing date; an XBRL page by its report's filing date; a transcript by its call date; a fails-to-deliver row by its settlement date (its price is the previous trading day's close); a Form 4 price page by its latest Form 4's filing date.
Time series pages (prices, government tables) are dated by their first row, and each row is hidden after the as-of date.
Pages that had no date of their own and were dated by the day they were fetched (October
1. stay hidden from every query, as in sim-ff3.
Renamed or delisted companies carry their earlier tickers too (for example RUBI on Magnite's pages, QRTEA on Qurate's, VSCO on Victoria's Secret's).

Contamination check

ARB's 16-gram check (tests/cloud/finance_web_overlap.py) compared every new page with the four tasks' test questions, claims, rubric lines and reference answers (FrontierFinance's gated file read inside the check's container). No page shares a 16-gram with a test question. Eight pages that share 16-grams with two or more rubric items of one question are left out (4 transcripts, 1 news item and 3 SEC filings whose sentences the rubrics quote; the news item was not going to be in this revision anyway). The benchmarks' repositories, datasets and papers are blocked from every fetch ([blocked] in ARB's raw_corpus/hosts.toml).

Downloads last month: 18

Total file size:

38.6 GB