Dataset
finsight-sec-filings
FinSight gathers plain-text 10-K and 10-Q SEC filings for twenty major US public firms spanning six sectors, yielding 97 records intended for BERT fine-tuning, retrieval-augmented generation, and multi-agent financial research.
Data preview
Dataset card
finsight-sec-filings is a cleaned plain-text corpus of US SEC EDGAR 10-K and 10-Q filings covering 20 large publicly traded companies across six sectors. The dataset is distributed on Hugging Face by publisher musk1209 and is intended as training and evaluation material for financial NLP research.
What this dataset contains
The corpus holds 10-K annual and 10-Q quarterly filings sourced from the US SEC EDGAR public-domain system. The 97-row train split lists tickers spanning six sectors, with named companies in the README including AAPL, MSFT, AMZN, JNJ, PFE, UNH, WMT, PG, KO, MCD, XOM, CVX, and CAT. Raw HTML has been parsed and stripped of XBRL and iXBRL metadata, then whitespace-normalized, with tables flattened for downstream processing.
Structure and fields
Each record is tabular with a long text field. Schema columns are ticker, cik, company_name, form, filing_date, accession, text, n_chars, n_words, and source_filename. The ticker, cik, company_name, form, accession, and source_filename fields are strings, filing_date is a timestamp, and n_chars and n_words are int64 length counters. The text column holds the cleaned filing body, with example documents reported at roughly 206,939 characters and 71,771 characters for an AAPL 10-K and 10-Q pair.
How analysts use it
Tags list question-answering and text-classification task categories, suggesting use for fine-tuning financial language models or evaluating retrieval-augmented research assistants. Equity and credit analysts at investment funds could treat it as a standardized, machine-readable archive of fundamental disclosures for sentiment extraction, MD&A change tracking, or peer benchmarking across the included sectors. Because coverage is limited to 20 issuers and fewer than 100 filings, it is best suited for methodological prototyping rather than full-universe screening.
Documentation & references 1
Related datasets
Browse moreParsed eligibility criteria from 13,229 ClinicalTrials.gov studies representing the candidate pool gathered by two first-stage retrievers during TREC Clinical Trials 2021–2023, formatted as typed entity-relation graphs for reranking models.
The clinical-trials-patient-graphs dataset is a small tabular collection of patient-level clinical records distributed in Parquet format by the 2001jdev publisher. It organizes entities and relations extracted from trial data into structured rows spanning 2021 through 2023.
This dataset is a parsed snapshot of ClinicalTrials.gov records, distributed in Parquet format on the Hugging Face Hub under an unspecified license. It contains roughly 52,000 trial entries drawn from public registry filings.
clinical-trials-trec-qrels is a tabular relevance-judgment file distributed by the 2001jdev user on the Hugging Face Hub. It maps clinical-trial topic identifiers to NCT registry IDs with graded relevance scores.
The clinical-trials-trec-topics dataset is a small tabular collection of clinical-trial topic records distributed in Parquet format on the Hugging Face Hub. It appears to be derived from TREC clinical-trial retrieval topics, with rows partitioned by year.
SEC EDGAR 10-K & 10-Q financial statements for 2,600+ NASDAQ companies.
Discussion
Questions or observations about this dataset? Join the conversation on X.
No comments yet. Start the conversation on X.