Datasets:

TheFinAI
/

en-convfinqa

pretty_name: FinBen · ConvFinQA
license: mit
language:
  - en
task_categories:
  - question-answering
  - table-question-answering
size_categories:
  - 10K<n<100K
source_datasets:
  - extended
tags:
  - finance
  - benchmark
  - thefinai
  - finben
  - flare
dataset_info:
  features:
    - name: id
      dtype: string
    - name: query
      dtype: string
    - name: answer
      dtype: string
    - name: turn
      dtype: int64
    - name: dialogue_id
      dtype: int64
  splits:
    - name: train
      num_bytes: 44382083
      num_examples: 8891
    - name: valid
      num_bytes: 11171617
      num_examples: 2213
    - name: test
      num_bytes: 7116753
      num_examples: 1490
  download_size: 11803908
  dataset_size: 62670453
extra_gated_heading: Request access to FinBen · ConvFinQA
extra_gated_description: >-
  This FinBen task is released by The Fin AI for research. Access is granted
  automatically after you complete this short form.
extra_gated_button_content: Agree and access
extra_gated_prompt: >-
  By accessing this dataset you agree to the license of the original source (MIT
  License) and to cite the FinBen paper and the original dataset in any
  resulting publication.
extra_gated_fields:
  Full name: text
  Affiliation: text
  Country: country
  Intended use:
    type: select
    options:
      - Research
      - Education
      - Commercial evaluation
      - Other
  I agree to the terms above and will cite the papers: checkbox

FinBen · ConvFinQA

📄 Paper · 💻 Code · 🏆 Leaderboard · 🌐 The Fin AI

Part of FinBen — FinBen: A Holistic Financial Benchmark for Large Language Models (arXiv:2402.12659).


Task	multi-turn question answering
Original dataset	ConvFinQA (Chen et al., 2022)
Evaluation metric	EM Accuracy
Source license	MIT License
Language	en

Quick Start

from datasets import load_dataset

ds = load_dataset("TheFinAI/en-convfinqa", split="test")
print(ds[0])

Example prompt

In the context of this series of interconnected finance-related queries and the additional information provided by the pretext, table data, and posttext from a company's financial filings, please provide a response to the final question. This may require extracting information from the context and performing mathematical calculations. Please take into account the information provided in the preced…

Dataset Structure

Split	Rows
`train`	8,891
`test`	1,490
`valid`	2,213

Field	Description
`id`	Example id
`query`	Full instruction prompt given to the model
`answer`	Gold answer / label text
`turn`	Turn index within the dialogue
`dialogue_id`	Dialogue id

License

The original data is released under MIT License (FinBen paper, Table 2).

Citation

Please cite FinBen and the original dataset (ConvFinQA (Chen et al., 2022)):

@misc{xie2024finbenholisticfinancialbenchmark,
      title={FinBen: A Holistic Financial Benchmark for Large Language Models},
      author={Qianqian Xie and Weiguang Han and Zhengyu Chen and Ruoyu Xiang and Xiao Zhang and Yueru He and Mengxi Xiao and Dong Li and Yongfu Dai and Duanyu Feng and Yijing Xu and Haoqiang Kang and Ziyan Kuang and Chenhan Yuan and Kailai Yang and Zheheng Luo and Tianlin Zhang and Zhiwei Liu and Guojun Xiong and Zhiyang Deng and Yuechen Jiang and Zhiyuan Yao and Haohang Li and Yangyang Yu and Gang Hu and Jiajia Huang and Xiao-Yang Liu and Alejandro Lopez-Lira and Benyou Wang and Yanzhao Lai and Hao Wang and Min Peng and Sophia Ananiadou and Jimin Huang},
      year={2024},
      eprint={2402.12659},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2402.12659},
}

@article{chen2022convfinqa,
  title={Convfinqa: Exploring the chain of numerical reasoning in conversational finance question answering},
  author={Chen, Zhiyu and Li, Shiyang and Smiley, Charese and Ma, Zhiqiang and Shah, Sameena and Wang, William Yang},
  journal={arXiv preprint arXiv:2210.03849},
  year={2022}
}