00_filtered / README.md
jiminHuang's picture
Update dataset card: task, source, license, usage, citation, access form
70e34c7 verified
|
Raw History Blame Contribute Delete
1.7 kB
metadata
pretty_name: arXiv abstracts (filtered)
license: cc0-1.0
language:
  - en
task_categories:
  - text-generation
tags:
  - finance
  - pretraining
  - corpus
  - thefinai
extra_gated_heading: Request access to arXiv abstracts (filtered)
extra_gated_description: >-
  arXiv abstracts (filtered) 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 its license and to cite the The Fin AI
  paper in any resulting publication. Use of the content is also subject to the
  terms of the original source.
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 paper: checkbox

arXiv abstracts (filtered)

🌐 The Fin AI

Pretraining / reference corpus released by The Fin AI. Source: arXiv abstracts (via gfissore/arxiv-abstracts) — https://huggingface.co/datasets/gfissore/arxiv-abstracts-2021.

Note: Despite the shared naming with 01–25_filtered, this shard holds arXiv abstracts, not SEC filings.

Source

arXiv paper metadata/abstracts, which arXiv releases under CC0 1.0.

Structure

  • Rows: 41,633,632
  • Columns: source, date, text, token_count, category

Quick Start

from datasets import load_dataset

ds = load_dataset("TheFinAI/00_filtered", split="train", streaming=True)
print(next(iter(ds)))

License

CC0 1.0, following the original source. Please attribute the original source.