Datasets:
Tasks:
Text Generation
Modalities:
Text
Formats:
parquet
Languages:
English
Size:
10K - 100K
License:
Request access to LLM economic-data memorization (paper version)
LLM economic-data memorization (paper version) is released by The Fin AI for research. Access is granted automatically after you complete this short form.
By accessing this dataset you agree to its license and to cite the The Fin AI paper in any resulting publication.
Log in or Sign Up to review the conditions and access this dataset content.
LLM economic-data memorization (paper version)
Year-by-year (1999–2026) probes of whether LLMs have memorized economic data; the version used in the accompanying paper (under review). The authors have not yet specified a license for this dataset; treat it as research-only until clarified, and respect the terms of the original sources.
| Task | memorization probe over economic data by year |
| Language | en |
| License | other (unspecified) |
Quick Start
from datasets import load_dataset
ds = load_dataset("TheFinAI/llm-econ-memorization-paper", split="1999")
print(ds[0])
| Split | Rows |
|---|---|
1999 |
492 |
2000 |
508 |
2001 |
508 |
2002 |
508 |
2003 |
508 |
2004 |
513 |
2005 |
520 |
2006 |
520 |
2007 |
520 |
2008 |
520 |
2009 |
520 |
2010 |
527 |
2011 |
532 |
2012 |
540 |
2013 |
544 |
2014 |
544 |
2015 |
544 |
2016 |
544 |
2017 |
544 |
2018 |
544 |
2019 |
544 |
2020 |
544 |
2021 |
544 |
2022 |
544 |
2023 |
544 |
2024 |
544 |
2025 |
542 |
2026 |
225 |
Columns: id, task, cutoff_split, year, period, variable, data_name, system, prompt, answer, answer_type, meta_json
- Downloads last month
- 70