UDA FinQA (orgrctera/uda_fin_qa)

Overview

This dataset is the FinQA slice of the UDA (Unstructured Document Analysis) benchmark: 8,190 question–answer instances derived from real financial reports, packaged for retrieval-oriented evaluation in RAG pipelines.

UDA is a benchmark suite for Retrieval-Augmented Generation (RAG) over messy, real-world documents (PDF/HTML) where evidence mixes narrative text and tables. The finance portion includes large subsets aligned with FinQA-style numerical reasoning over earnings materials.

FinQA (Chen et al., EMNLP 2021) is the foundational dataset of expert-written questions over financial reports, with heterogeneous evidence (tables + text) and multi-step numerical reasoning. UDA adopts FinQA-aligned labeling within its broader document-analysis benchmark (Hui et al., NeurIPS 2024 Datasets & Benchmarks).

In this Hub release, each row is a retrieval task instance: the model must locate and use the right evidence (typically embedded in expected_output as structured context for scoring or teacher forcing) to answer the question in input, consistent with the FinQA / UDA evaluation setting where retrieval quality and parsing of unstructured financial documents are central.

Task

• Task type: Retrieval (within a RAG / document-analysis pipeline) for FinQA-style financial QA.
• Input: A natural-language question (input) about reported figures or relationships in corporate financial disclosures.
• Supervision / reference: expected_output is a JSON string containing gold answers, evidence pointers, and document context (see below). Metadata records UDA benchmark identifiers (sub_benchmark: fin_qa).

Models are typically evaluated by whether retrieved passages support the correct answer and whether generation matches gold reasoning or numeric targets, following the FinQA and UDA protocols.

Background

FinQA

FinQA targets numerical reasoning over financial data: questions are written by finance experts over real reports; annotations include explainable reasoning traces. The original work shows that general-domain pretrained LMs lag humans on finance-specific, multi-step numeric reasoning.

UDA benchmark

UDA revisits RAG and LLM-based document analysis across domains (including finance) using thousands of real documents and tens of thousands of expert-annotated Q&A pairs, with documents kept in original formats to stress parsing, chunking, and retrieval—not only generation.

The FinQA-related finance split in UDA (reported in the UDA paper as part of the finance track) corresponds to the scale of this dataset (8,190 examples in the default split here).

Data fields

Column	Type	Description
input	string	Question text posed over the report.
expected_output	string	JSON string with fields such as answers (e.g. str_answer, exe_answer), evidence (table/text references), and context (supporting pre/post text and table snippets).
metadata	struct	benchmark_name (uda_fin_qa), benchmark_type (uda), split, sub_benchmark (fin_qa), and value (JSON string with identifiers like label_key, label_file, q_uid).

Splits: Single split default with 8,190 examples.

Examples

Example rows are illustrative; long context blocks are abbreviated.

Example 1 — interest expense

• input: what is the the interest expense in 2009?
• expected_output (excerpt):

{
  "answers": {"str_answer": "380", "exe_answer": 3.8},
  "evidence": {
    "text_1": "if libor changes by 100 basis points , our annual interest expense would change by $ 3.8 million ."
  },
  "context": {
    "pre_text": [
      "interest rate to a variable interest rate based on the three-month libor plus 2.05% ...",
      "if libor changes by 100 basis points , our annual interest expense would change by $ 3.8 million .",
      "..."
    ]
  }
}

Example 2 — amortization growth

• input: what is the expected growth rate in amortization expense in 2010?
• expected_output (excerpt):

{
  "answers": {"str_answer": "-27.0%", "exe_answer": -0.26689},
  "evidence": {
    "table_1": "fiscal years the 2010 of amortization expense is $ 5425 ;",
    "text_2": "amortization expense from continuing operations , related to intangibles was $ 7.4 million , $ 9.3 million and $ 9.2 million in fiscal 2009 , 2008 and 2007 , respectively ."
  },
  "context": { "...": "..." }
}

metadata.value (example):

{"label_key": "ADI_2009", "label_file": "fin_qa", "q_uid": "ADI/2009/page_49.pdf-1"}

References

FinQA (source task & data lineage)

Zhiyu Chen, Wenhu Chen, Charese Smiley, Sameena Shah, Iana Borova, Dylan Langdon, Reema Moussa, Matt Beane, Ting-Hao Huang, Bryan Routledge, William Yang Wang. FinQA: A Dataset of Numerical Reasoning over Financial Data. EMNLP 2021, pages 3697–3711.

• Abstract (short): Introduces a large-scale dataset of QA pairs over financial reports with gold reasoning programs for explainability; shows that large pretrained models fall short of experts on finance knowledge and multi-step numerical reasoning.
• ACL Anthology: https://aclanthology.org/2021.emnlp-main.300/
• DOI: 10.18653/v1/2021.emnlp-main.300
• Code & data (original release): https://github.com/czyssrs/FinQA

UDA benchmark (suite containing this FinQA slice)

Yulong Hui, Yao Lu, Huanchen Zhang. UDA: A Benchmark Suite for Retrieval Augmented Generation in Real-world Document Analysis. NeurIPS 2024 (Datasets and Benchmarks Track).

• Abstract (short): Presents UDA with thousands of real-world documents and tens of thousands of expert-annotated Q&A pairs; evaluates LLM- and RAG-based document analysis and highlights parsing and retrieval design choices.
• arXiv: https://arxiv.org/abs/2406.15187
• arXiv DOI: 10.48550/arXiv.2406.15187
• NeurIPS proceedings: https://proceedings.neurips.cc/paper_files/paper/2024/hash/7c06759d1a8567f087b02e8589454917-Abstract-Datasets_and_Benchmarks_Track.html
• Code & resources: https://github.com/qinchuanhui/UDA-Benchmark

Related Hub resources

• UDA QA aggregation (reference): qinchuanhui/UDA-QA

Citation

If you use this dataset, please cite both FinQA and UDA (and this dataset record as appropriate):

@inproceedings{chen-etal-2021-finqa,
  title = {FinQA: A Dataset of Numerical Reasoning over Financial Data},
  author = {Chen, Zhiyu and Chen, Wenhu and Smiley, Charese and Shah, Sameena and Borova, Iana and Langdon, Dylan and Moussa, Reema and Beane, Matt and Huang, Ting-Hao and Routledge, Bryan and Wang, William Yang},
  booktitle = {Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing},
  year = {2021},
  pages = {3697--3711}
}

@article{hui2024uda,
  title   = {UDA: A Benchmark Suite for Retrieval Augmented Generation in Real-world Document Analysis},
  author  = {Hui, Yulong and Lu, Yao and Zhang, Huanchen},
  journal = {arXiv preprint arXiv:2406.15187},
  year    = {2024}
}

License

The original FinQA release is under the MIT License (see the FinQA repository). Use this dataset in compliance with the original data licenses and the UDA benchmark terms.